generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

vp8: apply clang-format

Browse Source 
Change-Id: I7605b6678014a5426ceb45c27b54885e0c4e06ed
This commit is contained in:
clang-format 2016-07-13 22:26:28 -07:00 committed by James Zern
parent 65daa41378
commit 81a6739533
177 changed files with 45232 additions and 51209 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

245
vp8/common/alloccommon.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "alloccommon.h"
#include "blockd.h"
@ -18,176 +17,166 @@
#include "entropymode.h"
#include "systemdependent.h"
void vp8_de_alloc_frame_buffers(VP8_COMMON *oci)
{
int i;
for (i = 0; i < NUM_YV12_BUFFERS; i++)
vp8_yv12_de_alloc_frame_buffer(&oci->yv12_fb[i]);
void vp8_de_alloc_frame_buffers(VP8_COMMON *oci) {
int i;
for (i = 0; i < NUM_YV12_BUFFERS; i++)
vp8_yv12_de_alloc_frame_buffer(&oci->yv12_fb[i]);
vp8_yv12_de_alloc_frame_buffer(&oci->temp_scale_frame);
vp8_yv12_de_alloc_frame_buffer(&oci->temp_scale_frame);
#if CONFIG_POSTPROC
vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer);
if (oci->post_proc_buffer_int_used)
vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer_int);
vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer);
if (oci->post_proc_buffer_int_used)
vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer_int);
vpx_free(oci->pp_limits_buffer);
oci->pp_limits_buffer = NULL;
vpx_free(oci->pp_limits_buffer);
oci->pp_limits_buffer = NULL;
vpx_free(oci->postproc_state.generated_noise);
oci->postproc_state.generated_noise = NULL;
vpx_free(oci->postproc_state.generated_noise);
oci->postproc_state.generated_noise = NULL;
#endif
vpx_free(oci->above_context);
vpx_free(oci->mip);
vpx_free(oci->above_context);
vpx_free(oci->mip);
#if CONFIG_ERROR_CONCEALMENT
vpx_free(oci->prev_mip);
oci->prev_mip = NULL;
vpx_free(oci->prev_mip);
oci->prev_mip = NULL;
#endif
oci->above_context = NULL;
oci->mip = NULL;
oci->above_context = NULL;
oci->mip = NULL;
}
int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height)
{
int i;
int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height) {
int i;
vp8_de_alloc_frame_buffers(oci);
vp8_de_alloc_frame_buffers(oci);
/* our internal buffers are always multiples of 16 */
if ((width & 0xf) != 0)
width += 16 - (width & 0xf);
/* our internal buffers are always multiples of 16 */
if ((width & 0xf) != 0) width += 16 - (width & 0xf);
if ((height & 0xf) != 0)
height += 16 - (height & 0xf);
if ((height & 0xf) != 0) height += 16 - (height & 0xf);
for (i = 0; i < NUM_YV12_BUFFERS; i++) {
oci->fb_idx_ref_cnt[i] = 0;
oci->yv12_fb[i].flags = 0;
if (vp8_yv12_alloc_frame_buffer(&oci->yv12_fb[i], width, height,
VP8BORDERINPIXELS) < 0)
goto allocation_fail;
}
for (i = 0; i < NUM_YV12_BUFFERS; i++)
{
oci->fb_idx_ref_cnt[i] = 0;
oci->yv12_fb[i].flags = 0;
if (vp8_yv12_alloc_frame_buffer(&oci->yv12_fb[i], width, height, VP8BORDERINPIXELS) < 0)
goto allocation_fail;
}
oci->new_fb_idx = 0;
oci->lst_fb_idx = 1;
oci->gld_fb_idx = 2;
oci->alt_fb_idx = 3;
oci->new_fb_idx = 0;
oci->lst_fb_idx = 1;
oci->gld_fb_idx = 2;
oci->alt_fb_idx = 3;
oci->fb_idx_ref_cnt[0] = 1;
oci->fb_idx_ref_cnt[1] = 1;
oci->fb_idx_ref_cnt[2] = 1;
oci->fb_idx_ref_cnt[3] = 1;
oci->fb_idx_ref_cnt[0] = 1;
oci->fb_idx_ref_cnt[1] = 1;
oci->fb_idx_ref_cnt[2] = 1;
oci->fb_idx_ref_cnt[3] = 1;
if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16,
VP8BORDERINPIXELS) < 0)
goto allocation_fail;
if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16, VP8BORDERINPIXELS) < 0)
goto allocation_fail;
oci->mb_rows = height >> 4;
oci->mb_cols = width >> 4;
oci->MBs = oci->mb_rows * oci->mb_cols;
oci->mode_info_stride = oci->mb_cols + 1;
oci->mip =
vpx_calloc((oci->mb_cols + 1) * (oci->mb_rows + 1), sizeof(MODE_INFO));
oci->mb_rows = height >> 4;
oci->mb_cols = width >> 4;
oci->MBs = oci->mb_rows * oci->mb_cols;
oci->mode_info_stride = oci->mb_cols + 1;
oci->mip = vpx_calloc((oci->mb_cols + 1) * (oci->mb_rows + 1), sizeof(MODE_INFO));
if (!oci->mip) goto allocation_fail;
if (!oci->mip)
goto allocation_fail;
oci->mi = oci->mip + oci->mode_info_stride + 1;
oci->mi = oci->mip + oci->mode_info_stride + 1;
/* Allocation of previous mode info will be done in vp8_decode_frame()
* as it is a decoder only data */
/* Allocation of previous mode info will be done in vp8_decode_frame()
* as it is a decoder only data */
oci->above_context =
vpx_calloc(sizeof(ENTROPY_CONTEXT_PLANES) * oci->mb_cols, 1);
oci->above_context = vpx_calloc(sizeof(ENTROPY_CONTEXT_PLANES) * oci->mb_cols, 1);
if (!oci->above_context)
goto allocation_fail;
if (!oci->above_context) goto allocation_fail;
#if CONFIG_POSTPROC
if (vp8_yv12_alloc_frame_buffer(&oci->post_proc_buffer, width, height, VP8BORDERINPIXELS) < 0)
goto allocation_fail;
if (vp8_yv12_alloc_frame_buffer(&oci->post_proc_buffer, width, height,
VP8BORDERINPIXELS) < 0)
goto allocation_fail;
oci->post_proc_buffer_int_used = 0;
memset(&oci->postproc_state, 0, sizeof(oci->postproc_state));
memset(oci->post_proc_buffer.buffer_alloc, 128,
oci->post_proc_buffer.frame_size);
oci->post_proc_buffer_int_used = 0;
memset(&oci->postproc_state, 0, sizeof(oci->postproc_state));
memset(oci->post_proc_buffer.buffer_alloc, 128,
oci->post_proc_buffer.frame_size);
/* Allocate buffer to store post-processing filter coefficients.
*
* Note: Round up mb_cols to support SIMD reads
*/
oci->pp_limits_buffer = vpx_memalign(16, 24 * ((oci->mb_cols + 1) & ~1));
if (!oci->pp_limits_buffer)
goto allocation_fail;
/* Allocate buffer to store post-processing filter coefficients.
*
* Note: Round up mb_cols to support SIMD reads
*/
oci->pp_limits_buffer = vpx_memalign(16, 24 * ((oci->mb_cols + 1) & ~1));
if (!oci->pp_limits_buffer) goto allocation_fail;
#endif
return 0;
return 0;
allocation_fail:
vp8_de_alloc_frame_buffers(oci);
return 1;
vp8_de_alloc_frame_buffers(oci);
return 1;
}
void vp8_setup_version(VP8_COMMON *cm)
{
switch (cm->version)
{
void vp8_setup_version(VP8_COMMON *cm) {
switch (cm->version) {
case 0:
cm->no_lpf = 0;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
cm->no_lpf = 0;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
case 1:
cm->no_lpf = 0;
cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
cm->no_lpf = 0;
cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
case 2:
cm->no_lpf = 1;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
cm->no_lpf = 1;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 0;
break;
case 3:
cm->no_lpf = 1;
cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 1;
break;
cm->no_lpf = 1;
cm->filter_type = SIMPLE_LOOPFILTER;
cm->use_bilinear_mc_filter = 1;
cm->full_pixel = 1;
break;
default:
/*4,5,6,7 are reserved for future use*/
cm->no_lpf = 0;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
}
/*4,5,6,7 are reserved for future use*/
cm->no_lpf = 0;
cm->filter_type = NORMAL_LOOPFILTER;
cm->use_bilinear_mc_filter = 0;
cm->full_pixel = 0;
break;
}
}
void vp8_create_common(VP8_COMMON *oci)
{
vp8_machine_specific_config(oci);
void vp8_create_common(VP8_COMMON *oci) {
vp8_machine_specific_config(oci);
vp8_init_mbmode_probs(oci);
vp8_default_bmode_probs(oci->fc.bmode_prob);
vp8_init_mbmode_probs(oci);
vp8_default_bmode_probs(oci->fc.bmode_prob);
oci->mb_no_coeff_skip = 1;
oci->no_lpf = 0;
oci->filter_type = NORMAL_LOOPFILTER;
oci->use_bilinear_mc_filter = 0;
oci->full_pixel = 0;
oci->multi_token_partition = ONE_PARTITION;
oci->clamp_type = RECON_CLAMP_REQUIRED;
oci->mb_no_coeff_skip = 1;
oci->no_lpf = 0;
oci->filter_type = NORMAL_LOOPFILTER;
oci->use_bilinear_mc_filter = 0;
oci->full_pixel = 0;
oci->multi_token_partition = ONE_PARTITION;
oci->clamp_type = RECON_CLAMP_REQUIRED;
/* Initialize reference frame sign bias structure to defaults */
memset(oci->ref_frame_sign_bias, 0, sizeof(oci->ref_frame_sign_bias));
/* Initialize reference frame sign bias structure to defaults */
memset(oci->ref_frame_sign_bias, 0, sizeof(oci->ref_frame_sign_bias));
/* Default disable buffer to buffer copying */
oci->copy_buffer_to_gf = 0;
oci->copy_buffer_to_arf = 0;
/* Default disable buffer to buffer copying */
oci->copy_buffer_to_gf = 0;
oci->copy_buffer_to_arf = 0;
}
void vp8_remove_common(VP8_COMMON *oci)
{
vp8_de_alloc_frame_buffers(oci);
}
void vp8_remove_common(VP8_COMMON *oci) { vp8_de_alloc_frame_buffers(oci); }

1
vp8/common/alloccommon.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ALLOCCOMMON_H_
#define VP8_COMMON_ALLOCCOMMON_H_

171
vp8/common/arm/armv6/idct_blk_v6.c
View File

@ -11,105 +11,90 @@
#include "vpx_config.h"
#include "vp8_rtcd.h"
void vp8_dequant_idct_add_y_block_v6(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i;
void vp8_dequant_idct_add_y_block_v6(short *q, short *dq,
unsigned char *dst,
int stride, char *eobs)
{
int i;
for (i = 0; i < 4; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_v6 (q, dq, dst, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_v6 (q[0]*dq[0], dst, stride, dst, stride);
((int *)q)[0] = 0;
}
if (eobs[1] > 1)
vp8_dequant_idct_add_v6 (q+16, dq, dst+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_v6 (q[16]*dq[0], dst+4, stride, dst+4, stride);
((int *)(q+16))[0] = 0;
}
if (eobs[2] > 1)
vp8_dequant_idct_add_v6 (q+32, dq, dst+8, stride);
else if (eobs[2] == 1)
{
vp8_dc_only_idct_add_v6 (q[32]*dq[0], dst+8, stride, dst+8, stride);
((int *)(q+32))[0] = 0;
}
if (eobs[3] > 1)
vp8_dequant_idct_add_v6 (q+48, dq, dst+12, stride);
else if (eobs[3] == 1)
{
vp8_dc_only_idct_add_v6 (q[48]*dq[0], dst+12, stride,dst+12,stride);
((int *)(q+48))[0] = 0;
}
q += 64;
dst += 4*stride;
eobs += 4;
for (i = 0; i < 4; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_v6(q, dq, dst, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_v6(q[0] * dq[0], dst, stride, dst, stride);
((int *)q)[0] = 0;
}
if (eobs[1] > 1)
vp8_dequant_idct_add_v6(q + 16, dq, dst + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_v6(q[16] * dq[0], dst + 4, stride, dst + 4, stride);
((int *)(q + 16))[0] = 0;
}
if (eobs[2] > 1)
vp8_dequant_idct_add_v6(q + 32, dq, dst + 8, stride);
else if (eobs[2] == 1) {
vp8_dc_only_idct_add_v6(q[32] * dq[0], dst + 8, stride, dst + 8, stride);
((int *)(q + 32))[0] = 0;
}
if (eobs[3] > 1)
vp8_dequant_idct_add_v6(q + 48, dq, dst + 12, stride);
else if (eobs[3] == 1) {
vp8_dc_only_idct_add_v6(q[48] * dq[0], dst + 12, stride, dst + 12,
stride);
((int *)(q + 48))[0] = 0;
}
q += 64;
dst += 4 * stride;
eobs += 4;
}
}
void vp8_dequant_idct_add_uv_block_v6(short *q, short *dq,
unsigned char *dstu,
unsigned char *dstv,
int stride, char *eobs)
{
int i;
void vp8_dequant_idct_add_uv_block_v6(short *q, short *dq, unsigned char *dstu,
unsigned char *dstv, int stride,
char *eobs) {
int i;
for (i = 0; i < 2; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_v6 (q, dq, dstu, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_v6 (q[0]*dq[0], dstu, stride, dstu, stride);
((int *)q)[0] = 0;
}
if (eobs[1] > 1)
vp8_dequant_idct_add_v6 (q+16, dq, dstu+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_v6 (q[16]*dq[0], dstu+4, stride,
dstu+4, stride);
((int *)(q+16))[0] = 0;
}
q += 32;
dstu += 4*stride;
eobs += 2;
for (i = 0; i < 2; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_v6(q, dq, dstu, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_v6(q[0] * dq[0], dstu, stride, dstu, stride);
((int *)q)[0] = 0;
}
for (i = 0; i < 2; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_v6 (q, dq, dstv, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_v6 (q[0]*dq[0], dstv, stride, dstv, stride);
((int *)q)[0] = 0;
}
if (eobs[1] > 1)
vp8_dequant_idct_add_v6 (q+16, dq, dstv+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_v6 (q[16]*dq[0], dstv+4, stride,
dstv+4, stride);
((int *)(q+16))[0] = 0;
}
q += 32;
dstv += 4*stride;
eobs += 2;
if (eobs[1] > 1)
vp8_dequant_idct_add_v6(q + 16, dq, dstu + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_v6(q[16] * dq[0], dstu + 4, stride, dstu + 4,
stride);
((int *)(q + 16))[0] = 0;
}
q += 32;
dstu += 4 * stride;
eobs += 2;
}
for (i = 0; i < 2; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_v6(q, dq, dstv, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_v6(q[0] * dq[0], dstv, stride, dstv, stride);
((int *)q)[0] = 0;
}
if (eobs[1] > 1)
vp8_dequant_idct_add_v6(q + 16, dq, dstv + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_v6(q[16] * dq[0], dstv + 4, stride, dstv + 4,
stride);
((int *)(q + 16))[0] = 0;
}
q += 32;
dstv += 4 * stride;
eobs += 2;
}
}

130
vp8/common/arm/bilinearfilter_arm.c
View File

@ -14,100 +14,74 @@
#include "vp8/common/filter.h"
#include "bilinearfilter_arm.h"
void vp8_filter_block2d_bil_armv6
(
unsigned char *src_ptr,
unsigned char *dst_ptr,
unsigned int src_pitch,
unsigned int dst_pitch,
const short *HFilter,
const short *VFilter,
int Width,
int Height
)
{
unsigned short FData[36*16]; /* Temp data buffer used in filtering */
void vp8_filter_block2d_bil_armv6(unsigned char *src_ptr,
unsigned char *dst_ptr,
unsigned int src_pitch,
unsigned int dst_pitch, const short *HFilter,
const short *VFilter, int Width, int Height) {
unsigned short FData[36 * 16]; /* Temp data buffer used in filtering */
/* First filter 1-D horizontally... */
vp8_filter_block2d_bil_first_pass_armv6(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
/* First filter 1-D horizontally... */
vp8_filter_block2d_bil_first_pass_armv6(src_ptr, FData, src_pitch, Height + 1,
Width, HFilter);
/* then 1-D vertically... */
vp8_filter_block2d_bil_second_pass_armv6(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
/* then 1-D vertically... */
vp8_filter_block2d_bil_second_pass_armv6(FData, dst_ptr, dst_pitch, Height,
Width, VFilter);
}
void vp8_bilinear_predict4x4_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict4x4_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch,
HFilter, VFilter, 4, 4);
}
void vp8_bilinear_predict8x8_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict8x8_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch,
HFilter, VFilter, 8, 8);
}
void vp8_bilinear_predict8x4_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict8x4_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch,
HFilter, VFilter, 8, 4);
}
void vp8_bilinear_predict16x16_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict16x16_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
vp8_filter_block2d_bil_armv6(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch,
HFilter, VFilter, 16, 16);
}

26
vp8/common/arm/bilinearfilter_arm.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ARM_BILINEARFILTER_ARM_H_
#define VP8_COMMON_ARM_BILINEARFILTER_ARM_H_
@ -16,25 +15,14 @@
extern "C" {
#endif
extern void vp8_filter_block2d_bil_first_pass_armv6
(
const unsigned char *src_ptr,
unsigned short *dst_ptr,
unsigned int src_pitch,
unsigned int height,
unsigned int width,
const short *vp8_filter
);
extern void vp8_filter_block2d_bil_first_pass_armv6(
const unsigned char *src_ptr, unsigned short *dst_ptr,
unsigned int src_pitch, unsigned int height, unsigned int width,
const short *vp8_filter);
extern void vp8_filter_block2d_bil_second_pass_armv6
(
const unsigned short *src_ptr,
unsigned char *dst_ptr,
int dst_pitch,
unsigned int height,
unsigned int width,
const short *vp8_filter
);
extern void vp8_filter_block2d_bil_second_pass_armv6(
const unsigned short *src_ptr, unsigned char *dst_ptr, int dst_pitch,
unsigned int height, unsigned int width, const short *vp8_filter);
#ifdef __cplusplus
} // extern "C"

10
vp8/common/arm/dequantize_arm.c
View File

@ -8,18 +8,16 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8/common/blockd.h"
#if HAVE_MEDIA
extern void vp8_dequantize_b_loop_v6(short *Q, short *DQC, short *DQ);
void vp8_dequantize_b_v6(BLOCKD *d, short *DQC)
{
short *DQ = d->dqcoeff;
short *Q = d->qcoeff;
void vp8_dequantize_b_v6(BLOCKD *d, short *DQC) {
short *DQ = d->dqcoeff;
short *Q = d->qcoeff;
vp8_dequantize_b_loop_v6(Q, DQC, DQ);
vp8_dequantize_b_loop_v6(Q, DQC, DQ);
}
#endif

311
vp8/common/arm/filter_arm.c
View File

@ -8,214 +8,169 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#include <math.h>
#include "vp8/common/filter.h"
#include "vpx_ports/mem.h"
extern void vp8_filter_block2d_first_pass_armv6
(
unsigned char *src_ptr,
short *output_ptr,
unsigned int src_pixels_per_line,
unsigned int output_width,
unsigned int output_height,
const short *vp8_filter
);
extern void vp8_filter_block2d_first_pass_armv6(
unsigned char *src_ptr, short *output_ptr, unsigned int src_pixels_per_line,
unsigned int output_width, unsigned int output_height,
const short *vp8_filter);
// 8x8
extern void vp8_filter_block2d_first_pass_8x8_armv6
(
unsigned char *src_ptr,
short *output_ptr,
unsigned int src_pixels_per_line,
unsigned int output_width,
unsigned int output_height,
const short *vp8_filter
);
extern void vp8_filter_block2d_first_pass_8x8_armv6(
unsigned char *src_ptr, short *output_ptr, unsigned int src_pixels_per_line,
unsigned int output_width, unsigned int output_height,
const short *vp8_filter);
// 16x16
extern void vp8_filter_block2d_first_pass_16x16_armv6
(
unsigned char *src_ptr,
short *output_ptr,
unsigned int src_pixels_per_line,
unsigned int output_width,
unsigned int output_height,
const short *vp8_filter
);
extern void vp8_filter_block2d_first_pass_16x16_armv6(
unsigned char *src_ptr, short *output_ptr, unsigned int src_pixels_per_line,
unsigned int output_width, unsigned int output_height,
const short *vp8_filter);
extern void vp8_filter_block2d_second_pass_armv6
(
short *src_ptr,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int cnt,
const short *vp8_filter
);
extern void vp8_filter_block2d_second_pass_armv6(short *src_ptr,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int cnt,
const short *vp8_filter);
extern void vp8_filter4_block2d_second_pass_armv6
(
short *src_ptr,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int cnt,
const short *vp8_filter
);
extern void vp8_filter4_block2d_second_pass_armv6(short *src_ptr,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int cnt,
const short *vp8_filter);
extern void vp8_filter_block2d_first_pass_only_armv6
(
unsigned char *src_ptr,
unsigned char *output_ptr,
unsigned int src_pixels_per_line,
unsigned int cnt,
unsigned int output_pitch,
const short *vp8_filter
);
extern void vp8_filter_block2d_first_pass_only_armv6(
unsigned char *src_ptr, unsigned char *output_ptr,
unsigned int src_pixels_per_line, unsigned int cnt,
unsigned int output_pitch, const short *vp8_filter);
extern void vp8_filter_block2d_second_pass_only_armv6
(
unsigned char *src_ptr,
unsigned char *output_ptr,
unsigned int src_pixels_per_line,
unsigned int cnt,
unsigned int output_pitch,
const short *vp8_filter
);
extern void vp8_filter_block2d_second_pass_only_armv6(
unsigned char *src_ptr, unsigned char *output_ptr,
unsigned int src_pixels_per_line, unsigned int cnt,
unsigned int output_pitch, const short *vp8_filter);
#if HAVE_MEDIA
void vp8_sixtap_predict4x4_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short, FData[12*4]); /* Temp data buffer used in filtering */
void vp8_sixtap_predict4x4_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short,
FData[12 * 4]); /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
/* Vfilter is null. First pass only */
if (xoffset && !yoffset) {
/*vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2,
src_pixels_per_line, 4, 4, HFilter );
vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4,
VFilter );*/
/* Vfilter is null. First pass only */
if (xoffset && !yoffset)
{
/*vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2, src_pixels_per_line, 4, 4, HFilter );
vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, VFilter );*/
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, HFilter);
vp8_filter_block2d_first_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, HFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset) {
vp8_filter_block2d_second_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, VFilter);
} else {
/* Vfilter is a 4 tap filter */
if (yoffset & 0x1) {
vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line,
FData + 1, src_pixels_per_line, 4, 7,
HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4,
VFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, VFilter);
}
else
{
/* Vfilter is a 4 tap filter */
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 4, 7, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
}
/* Vfilter is 6 tap filter */
else
{
vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 4, 9, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
}
/* Vfilter is 6 tap filter */
else {
vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line),
FData, src_pixels_per_line, 4, 9,
HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4,
VFilter);
}
}
}
void vp8_sixtap_predict8x8_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short, FData[16*8]); /* Temp data buffer used in filtering */
void vp8_sixtap_predict8x8_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short,
FData[16 * 8]); /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
if (xoffset && !yoffset)
{
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter);
}
else
{
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_8x8_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 8, 11, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
}
else
{
vp8_filter_block2d_first_pass_8x8_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 8, 13, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
}
if (xoffset && !yoffset) {
vp8_filter_block2d_first_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset) {
vp8_filter_block2d_second_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter);
} else {
if (yoffset & 0x1) {
vp8_filter_block2d_first_pass_8x8_armv6(src_ptr - src_pixels_per_line,
FData + 1, src_pixels_per_line, 8,
11, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8,
VFilter);
} else {
vp8_filter_block2d_first_pass_8x8_armv6(
src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 8,
13, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8,
VFilter);
}
}
}
void vp8_sixtap_predict16x16_armv6(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short,
FData[24 * 16]); /* Temp data buffer used in filtering */
void vp8_sixtap_predict16x16_armv6
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED(4, short, FData[24*16]); /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
if (xoffset && !yoffset)
{
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, HFilter);
if (xoffset && !yoffset) {
vp8_filter_block2d_first_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, HFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset) {
vp8_filter_block2d_second_pass_only_armv6(
src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, VFilter);
} else {
if (yoffset & 0x1) {
vp8_filter_block2d_first_pass_16x16_armv6(src_ptr - src_pixels_per_line,
FData + 1, src_pixels_per_line,
16, 19, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16,
VFilter);
} else {
vp8_filter_block2d_first_pass_16x16_armv6(
src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 16,
21, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16,
VFilter);
}
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, VFilter);
}
else
{
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_16x16_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 16, 19, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
}
else
{
vp8_filter_block2d_first_pass_16x16_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 16, 21, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
}
}
}
}
#endif

213
vp8/common/arm/loopfilter_arm.c
View File

@ -8,15 +8,14 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#include "vp8/common/loopfilter.h"
#include "vp8/common/onyxc_int.h"
#define prototype_loopfilter(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
const unsigned char *limit, const unsigned char *thresh, int count)
#define prototype_loopfilter(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit, \
const unsigned char *limit, const unsigned char *thresh, int count)
#if HAVE_MEDIA
extern prototype_loopfilter(vp8_loop_filter_horizontal_edge_armv6);
@ -27,10 +26,11 @@ extern prototype_loopfilter(vp8_mbloop_filter_vertical_edge_armv6);
#if HAVE_NEON
typedef void loopfilter_y_neon(unsigned char *src, int pitch,
unsigned char blimit, unsigned char limit, unsigned char thresh);
unsigned char blimit, unsigned char limit,
unsigned char thresh);
typedef void loopfilter_uv_neon(unsigned char *u, int pitch,
unsigned char blimit, unsigned char limit, unsigned char thresh,
unsigned char *v);
unsigned char blimit, unsigned char limit,
unsigned char thresh, unsigned char *v);
extern loopfilter_y_neon vp8_loop_filter_horizontal_edge_y_neon;
extern loopfilter_y_neon vp8_loop_filter_vertical_edge_y_neon;
@ -46,136 +46,163 @@ extern loopfilter_uv_neon vp8_mbloop_filter_vertical_edge_uv_neon;
#if HAVE_MEDIA
/* ARMV6/MEDIA loopfilter functions*/
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride,
int uv_stride, loop_filter_info *lfi) {
vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride,
int uv_stride, loop_filter_info *lfi) {
vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
}
void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit)
{
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, blimit);
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 4 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 8 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride,
blimit);
}
/* Vertical B Filtering */
void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_loop_filter_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_loop_filter_vertical_edge_armv6(y_ptr + 4, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 8, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 1);
}
void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit)
{
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 12, y_stride, blimit);
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr + 12, y_stride, blimit);
}
#endif
#if HAVE_NEON
/* NEON loopfilter functions */
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
unsigned char mblim = *lfi->mblim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, mblim, lim, hev_thr);
void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
unsigned char mblim = *lfi->mblim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, mblim, lim,
hev_thr);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, mblim, lim, hev_thr, v_ptr);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, mblim, lim,
hev_thr, v_ptr);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
unsigned char mblim = *lfi->mblim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
unsigned char mblim = *lfi->mblim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, mblim, lim, hev_thr);
vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, mblim, lim, hev_thr);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, mblim, lim, hev_thr, v_ptr);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, mblim, lim,
hev_thr, v_ptr);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
unsigned char blim = *lfi->blim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
unsigned char blim = *lfi->blim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 4 * y_stride, y_stride, blim, lim, hev_thr);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 8 * y_stride, y_stride, blim, lim, hev_thr);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, blim, lim, hev_thr);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 4 * y_stride, y_stride, blim,
lim, hev_thr);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 8 * y_stride, y_stride, blim,
lim, hev_thr);
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, blim,
lim, hev_thr);
if (u_ptr)
vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride, blim, lim, hev_thr, v_ptr + 4 * uv_stride);
if (u_ptr)
vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride,
blim, lim, hev_thr,
v_ptr + 4 * uv_stride);
}
/* Vertical B Filtering */
void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
unsigned char blim = *lfi->blim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
unsigned char blim = *lfi->blim;
unsigned char lim = *lfi->lim;
unsigned char hev_thr = *lfi->hev_thr;
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 4, y_stride, blim, lim, hev_thr);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 8, y_stride, blim, lim, hev_thr);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, blim, lim, hev_thr);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 4, y_stride, blim, lim, hev_thr);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 8, y_stride, blim, lim, hev_thr);
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, blim, lim,
hev_thr);
if (u_ptr)
vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, blim, lim, hev_thr, v_ptr + 4);
if (u_ptr)
vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, blim, lim,
hev_thr, v_ptr + 4);
}
#endif

1182
vp8/common/arm/neon/bilinearpredict_neon.c
View File

File diff suppressed because it is too large Load Diff

69
vp8/common/arm/neon/copymem_neon.c
View File

@ -10,50 +10,41 @@
#include <arm_neon.h>
void vp8_copy_mem8x4_neon(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride) {
uint8x8_t vtmp;
int r;
void vp8_copy_mem8x4_neon(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride) {
uint8x8_t vtmp;
int r;
for (r = 0; r < 4; r++) {
vtmp = vld1_u8(src);
vst1_u8(dst, vtmp);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 4; r++) {
vtmp = vld1_u8(src);
vst1_u8(dst, vtmp);
src += src_stride;
dst += dst_stride;
}
}
void vp8_copy_mem8x8_neon(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride) {
uint8x8_t vtmp;
int r;
void vp8_copy_mem8x8_neon(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride) {
uint8x8_t vtmp;
int r;
for (r = 0; r < 8; r++) {
vtmp = vld1_u8(src);
vst1_u8(dst, vtmp);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 8; r++) {
vtmp = vld1_u8(src);
vst1_u8(dst, vtmp);
src += src_stride;
dst += dst_stride;
}
}
void vp8_copy_mem16x16_neon(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride) {
int r;
uint8x16_t qtmp;
void vp8_copy_mem16x16_neon(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride) {
int r;
uint8x16_t qtmp;
for (r = 0; r < 16; r++) {
qtmp = vld1q_u8(src);
vst1q_u8(dst, qtmp);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 16; r++) {
qtmp = vld1q_u8(src);
vst1q_u8(dst, qtmp);
src += src_stride;
dst += dst_stride;
}
}

47
vp8/common/arm/neon/dc_only_idct_add_neon.c
View File

@ -10,33 +10,30 @@
#include <arm_neon.h>
void vp8_dc_only_idct_add_neon(
int16_t input_dc,
unsigned char *pred_ptr,
int pred_stride,
unsigned char *dst_ptr,
int dst_stride) {
int i;
uint16_t a1 = ((input_dc + 4) >> 3);
uint32x2_t d2u32 = vdup_n_u32(0);
uint8x8_t d2u8;
uint16x8_t q1u16;
uint16x8_t qAdd;
void vp8_dc_only_idct_add_neon(int16_t input_dc, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride) {
int i;
uint16_t a1 = ((input_dc + 4) >> 3);
uint32x2_t d2u32 = vdup_n_u32(0);
uint8x8_t d2u8;
uint16x8_t q1u16;
uint16x8_t qAdd;
qAdd = vdupq_n_u16(a1);
qAdd = vdupq_n_u16(a1);
for (i = 0; i < 2; i++) {
d2u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d2u32, 0);
pred_ptr += pred_stride;
d2u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d2u32, 1);
pred_ptr += pred_stride;
for (i = 0; i < 2; i++) {
d2u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d2u32, 0);
pred_ptr += pred_stride;
d2u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d2u32, 1);
pred_ptr += pred_stride;
q1u16 = vaddw_u8(qAdd, vreinterpret_u8_u32(d2u32));
d2u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
q1u16 = vaddw_u8(qAdd, vreinterpret_u8_u32(d2u32));
d2u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d2u8), 0);
dst_ptr += dst_stride;
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d2u8), 1);
dst_ptr += dst_stride;
}
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d2u8), 0);
dst_ptr += dst_stride;
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d2u8), 1);
dst_ptr += dst_stride;
}
}

199
vp8/common/arm/neon/dequant_idct_neon.c
View File

@ -11,132 +11,129 @@
#include <arm_neon.h>
static const int16_t cospi8sqrt2minus1 = 20091;
static const int16_t sinpi8sqrt2 = 35468;
static const int16_t sinpi8sqrt2 = 35468;
void vp8_dequant_idct_add_neon(
int16_t *input,
int16_t *dq,
unsigned char *dst,
int stride) {
unsigned char *dst0;
int32x2_t d14, d15;
int16x4_t d2, d3, d4, d5, d10, d11, d12, d13;
int16x8_t q1, q2, q3, q4, q5, q6;
int16x8_t qEmpty = vdupq_n_s16(0);
int32x2x2_t d2tmp0, d2tmp1;
int16x4x2_t d2tmp2, d2tmp3;
void vp8_dequant_idct_add_neon(int16_t *input, int16_t *dq, unsigned char *dst,
int stride) {
unsigned char *dst0;
int32x2_t d14, d15;
int16x4_t d2, d3, d4, d5, d10, d11, d12, d13;
int16x8_t q1, q2, q3, q4, q5, q6;
int16x8_t qEmpty = vdupq_n_s16(0);
int32x2x2_t d2tmp0, d2tmp1;
int16x4x2_t d2tmp2, d2tmp3;
d14 = d15 = vdup_n_s32(0);
d14 = d15 = vdup_n_s32(0);
// load input
q3 = vld1q_s16(input);
vst1q_s16(input, qEmpty);
input += 8;
q4 = vld1q_s16(input);
vst1q_s16(input, qEmpty);
// load input
q3 = vld1q_s16(input);
vst1q_s16(input, qEmpty);
input += 8;
q4 = vld1q_s16(input);
vst1q_s16(input, qEmpty);
// load dq
q5 = vld1q_s16(dq);
dq += 8;
q6 = vld1q_s16(dq);
// load dq
q5 = vld1q_s16(dq);
dq += 8;
q6 = vld1q_s16(dq);
// load src from dst
dst0 = dst;
d14 = vld1_lane_s32((const int32_t *)dst0, d14, 0);
dst0 += stride;
d14 = vld1_lane_s32((const int32_t *)dst0, d14, 1);
dst0 += stride;
d15 = vld1_lane_s32((const int32_t *)dst0, d15, 0);
dst0 += stride;
d15 = vld1_lane_s32((const int32_t *)dst0, d15, 1);
// load src from dst
dst0 = dst;
d14 = vld1_lane_s32((const int32_t *)dst0, d14, 0);
dst0 += stride;
d14 = vld1_lane_s32((const int32_t *)dst0, d14, 1);
dst0 += stride;
d15 = vld1_lane_s32((const int32_t *)dst0, d15, 0);
dst0 += stride;
d15 = vld1_lane_s32((const int32_t *)dst0, d15, 1);
q1 = vreinterpretq_s16_u16(vmulq_u16(vreinterpretq_u16_s16(q3),
vreinterpretq_u16_s16(q5)));
q2 = vreinterpretq_s16_u16(vmulq_u16(vreinterpretq_u16_s16(q4),
vreinterpretq_u16_s16(q6)));
q1 = vreinterpretq_s16_u16(
vmulq_u16(vreinterpretq_u16_s16(q3), vreinterpretq_u16_s16(q5)));
q2 = vreinterpretq_s16_u16(
vmulq_u16(vreinterpretq_u16_s16(q4), vreinterpretq_u16_s16(q6)));
d12 = vqadd_s16(vget_low_s16(q1), vget_low_s16(q2));
d13 = vqsub_s16(vget_low_s16(q1), vget_low_s16(q2));
d12 = vqadd_s16(vget_low_s16(q1), vget_low_s16(q2));
d13 = vqsub_s16(vget_low_s16(q1), vget_low_s16(q2));
q2 = vcombine_s16(vget_high_s16(q1), vget_high_s16(q2));
q2 = vcombine_s16(vget_high_s16(q1), vget_high_s16(q2));
q3 = vqdmulhq_n_s16(q2, sinpi8sqrt2);
q4 = vqdmulhq_n_s16(q2, cospi8sqrt2minus1);
q3 = vqdmulhq_n_s16(q2, sinpi8sqrt2);
q4 = vqdmulhq_n_s16(q2, cospi8sqrt2minus1);
q3 = vshrq_n_s16(q3, 1);
q4 = vshrq_n_s16(q4, 1);
q3 = vshrq_n_s16(q3, 1);
q4 = vshrq_n_s16(q4, 1);
q3 = vqaddq_s16(q3, q2);
q4 = vqaddq_s16(q4, q2);
q3 = vqaddq_s16(q3, q2);
q4 = vqaddq_s16(q4, q2);
d10 = vqsub_s16(vget_low_s16(q3), vget_high_s16(q4));
d11 = vqadd_s16(vget_high_s16(q3), vget_low_s16(q4));
d10 = vqsub_s16(vget_low_s16(q3), vget_high_s16(q4));
d11 = vqadd_s16(vget_high_s16(q3), vget_low_s16(q4));
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
d2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
d2tmp2 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[0]),
vreinterpret_s16_s32(d2tmp1.val[0]));
d2tmp3 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[1]),
vreinterpret_s16_s32(d2tmp1.val[1]));
d2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
d2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
d2tmp2 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[0]),
vreinterpret_s16_s32(d2tmp1.val[0]));
d2tmp3 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[1]),
vreinterpret_s16_s32(d2tmp1.val[1]));
// loop 2
q2 = vcombine_s16(d2tmp2.val[1], d2tmp3.val[1]);
// loop 2
q2 = vcombine_s16(d2tmp2.val[1], d2tmp3.val[1]);
q3 = vqdmulhq_n_s16(q2, sinpi8sqrt2);
q4 = vqdmulhq_n_s16(q2, cospi8sqrt2minus1);
q3 = vqdmulhq_n_s16(q2, sinpi8sqrt2);
q4 = vqdmulhq_n_s16(q2, cospi8sqrt2minus1);
d12 = vqadd_s16(d2tmp2.val[0], d2tmp3.val[0]);
d13 = vqsub_s16(d2tmp2.val[0], d2tmp3.val[0]);
d12 = vqadd_s16(d2tmp2.val[0], d2tmp3.val[0]);
d13 = vqsub_s16(d2tmp2.val[0], d2tmp3.val[0]);
q3 = vshrq_n_s16(q3, 1);
q4 = vshrq_n_s16(q4, 1);
q3 = vshrq_n_s16(q3, 1);
q4 = vshrq_n_s16(q4, 1);
q3 = vqaddq_s16(q3, q2);
q4 = vqaddq_s16(q4, q2);
q3 = vqaddq_s16(q3, q2);
q4 = vqaddq_s16(q4, q2);
d10 = vqsub_s16(vget_low_s16(q3), vget_high_s16(q4));
d11 = vqadd_s16(vget_high_s16(q3), vget_low_s16(q4));
d10 = vqsub_s16(vget_low_s16(q3), vget_high_s16(q4));
d11 = vqadd_s16(vget_high_s16(q3), vget_low_s16(q4));
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vrshr_n_s16(d2, 3);
d3 = vrshr_n_s16(d3, 3);
d4 = vrshr_n_s16(d4, 3);
d5 = vrshr_n_s16(d5, 3);
d2 = vrshr_n_s16(d2, 3);
d3 = vrshr_n_s16(d3, 3);
d4 = vrshr_n_s16(d4, 3);
d5 = vrshr_n_s16(d5, 3);
d2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
d2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
d2tmp2 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[0]),
vreinterpret_s16_s32(d2tmp1.val[0]));
d2tmp3 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[1]),
vreinterpret_s16_s32(d2tmp1.val[1]));
d2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
d2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
d2tmp2 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[0]),
vreinterpret_s16_s32(d2tmp1.val[0]));
d2tmp3 = vtrn_s16(vreinterpret_s16_s32(d2tmp0.val[1]),
vreinterpret_s16_s32(d2tmp1.val[1]));
q1 = vcombine_s16(d2tmp2.val[0], d2tmp2.val[1]);
q2 = vcombine_s16(d2tmp3.val[0], d2tmp3.val[1]);
q1 = vcombine_s16(d2tmp2.val[0], d2tmp2.val[1]);
q2 = vcombine_s16(d2tmp3.val[0], d2tmp3.val[1]);
q1 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q1),
vreinterpret_u8_s32(d14)));
q2 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q2),
vreinterpret_u8_s32(d15)));
q1 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q1), vreinterpret_u8_s32(d14)));
q2 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q2), vreinterpret_u8_s32(d15)));
d14 = vreinterpret_s32_u8(vqmovun_s16(q1));
d15 = vreinterpret_s32_u8(vqmovun_s16(q2));
d14 = vreinterpret_s32_u8(vqmovun_s16(q1));
d15 = vreinterpret_s32_u8(vqmovun_s16(q2));
dst0 = dst;
vst1_lane_s32((int32_t *)dst0, d14, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d14, 1);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d15, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d15, 1);
return;
dst0 = dst;
vst1_lane_s32((int32_t *)dst0, d14, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d14, 1);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d15, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d15, 1);
return;
}

12
vp8/common/arm/neon/dequantizeb_neon.c
View File

@ -13,13 +13,13 @@
#include "vp8/common/blockd.h"
void vp8_dequantize_b_neon(BLOCKD *d, short *DQC) {
int16x8x2_t qQ, qDQC, qDQ;
int16x8x2_t qQ, qDQC, qDQ;
qQ = vld2q_s16(d->qcoeff);
qDQC = vld2q_s16(DQC);
qQ = vld2q_s16(d->qcoeff);
qDQC = vld2q_s16(DQC);
qDQ.val[0] = vmulq_s16(qQ.val[0], qDQC.val[0]);
qDQ.val[1] = vmulq_s16(qQ.val[1], qDQC.val[1]);
qDQ.val[0] = vmulq_s16(qQ.val[0], qDQC.val[0]);
qDQ.val[1] = vmulq_s16(qQ.val[1], qDQC.val[1]);
vst2q_s16(d->dqcoeff, qDQ);
vst2q_s16(d->dqcoeff, qDQ);
}

120
vp8/common/arm/neon/idct_blk_neon.c
View File

@ -14,83 +14,71 @@
/* place these declarations here because we don't want to maintain them
* outside of this scope
*/
void idct_dequant_full_2x_neon(short *q, short *dq,
unsigned char *dst, int stride);
void idct_dequant_0_2x_neon(short *q, short dq,
unsigned char *dst, int stride);
void idct_dequant_full_2x_neon(short *q, short *dq, unsigned char *dst,
int stride);
void idct_dequant_0_2x_neon(short *q, short dq, unsigned char *dst, int stride);
void vp8_dequant_idct_add_y_block_neon(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i;
void vp8_dequant_idct_add_y_block_neon(short *q, short *dq,
unsigned char *dst,
int stride, char *eobs)
{
int i;
for (i = 0; i < 4; i++)
{
if (((short *)(eobs))[0])
{
if (((short *)eobs)[0] & 0xfefe)
idct_dequant_full_2x_neon (q, dq, dst, stride);
else
idct_dequant_0_2x_neon (q, dq[0], dst, stride);
}
if (((short *)(eobs))[1])
{
if (((short *)eobs)[1] & 0xfefe)
idct_dequant_full_2x_neon (q+32, dq, dst+8, stride);
else
idct_dequant_0_2x_neon (q+32, dq[0], dst+8, stride);
}
q += 64;
dst += 4*stride;
eobs += 4;
for (i = 0; i < 4; i++) {
if (((short *)(eobs))[0]) {
if (((short *)eobs)[0] & 0xfefe)
idct_dequant_full_2x_neon(q, dq, dst, stride);
else
idct_dequant_0_2x_neon(q, dq[0], dst, stride);
}
if (((short *)(eobs))[1]) {
if (((short *)eobs)[1] & 0xfefe)
idct_dequant_full_2x_neon(q + 32, dq, dst + 8, stride);
else
idct_dequant_0_2x_neon(q + 32, dq[0], dst + 8, stride);
}
q += 64;
dst += 4 * stride;
eobs += 4;
}
}
void vp8_dequant_idct_add_uv_block_neon(short *q, short *dq,
unsigned char *dstu,
unsigned char *dstv,
int stride, char *eobs)
{
if (((short *)(eobs))[0])
{
if (((short *)eobs)[0] & 0xfefe)
idct_dequant_full_2x_neon (q, dq, dstu, stride);
else
idct_dequant_0_2x_neon (q, dq[0], dstu, stride);
}
unsigned char *dstv, int stride,
char *eobs) {
if (((short *)(eobs))[0]) {
if (((short *)eobs)[0] & 0xfefe)
idct_dequant_full_2x_neon(q, dq, dstu, stride);
else
idct_dequant_0_2x_neon(q, dq[0], dstu, stride);
}
q += 32;
dstu += 4*stride;
q += 32;
dstu += 4 * stride;
if (((short *)(eobs))[1])
{
if (((short *)eobs)[1] & 0xfefe)
idct_dequant_full_2x_neon (q, dq, dstu, stride);
else
idct_dequant_0_2x_neon (q, dq[0], dstu, stride);
}
if (((short *)(eobs))[1]) {
if (((short *)eobs)[1] & 0xfefe)
idct_dequant_full_2x_neon(q, dq, dstu, stride);
else
idct_dequant_0_2x_neon(q, dq[0], dstu, stride);
}
q += 32;
q += 32;
if (((short *)(eobs))[2])
{
if (((short *)eobs)[2] & 0xfefe)
idct_dequant_full_2x_neon (q, dq, dstv, stride);
else
idct_dequant_0_2x_neon (q, dq[0], dstv, stride);
}
if (((short *)(eobs))[2]) {
if (((short *)eobs)[2] & 0xfefe)
idct_dequant_full_2x_neon(q, dq, dstv, stride);
else
idct_dequant_0_2x_neon(q, dq[0], dstv, stride);
}
q += 32;
dstv += 4*stride;
q += 32;
dstv += 4 * stride;
if (((short *)(eobs))[3])
{
if (((short *)eobs)[3] & 0xfefe)
idct_dequant_full_2x_neon (q, dq, dstv, stride);
else
idct_dequant_0_2x_neon (q, dq[0], dstv, stride);
}
if (((short *)(eobs))[3]) {
if (((short *)eobs)[3] & 0xfefe)
idct_dequant_full_2x_neon(q, dq, dstv, stride);
else
idct_dequant_0_2x_neon(q, dq[0], dstv, stride);
}
}

84
vp8/common/arm/neon/idct_dequant_0_2x_neon.c
View File

@ -10,54 +10,50 @@
#include <arm_neon.h>
void idct_dequant_0_2x_neon(
int16_t *q,
int16_t dq,
unsigned char *dst,
int stride) {
unsigned char *dst0;
int i, a0, a1;
int16x8x2_t q2Add;
int32x2_t d2s32 = vdup_n_s32(0),
d4s32 = vdup_n_s32(0);
uint8x8_t d2u8, d4u8;
uint16x8_t q1u16, q2u16;
void idct_dequant_0_2x_neon(int16_t *q, int16_t dq, unsigned char *dst,
int stride) {
unsigned char *dst0;
int i, a0, a1;
int16x8x2_t q2Add;
int32x2_t d2s32 = vdup_n_s32(0), d4s32 = vdup_n_s32(0);
uint8x8_t d2u8, d4u8;
uint16x8_t q1u16, q2u16;
a0 = ((q[0] * dq) + 4) >> 3;
a1 = ((q[16] * dq) + 4) >> 3;
q[0] = q[16] = 0;
q2Add.val[0] = vdupq_n_s16((int16_t)a0);
q2Add.val[1] = vdupq_n_s16((int16_t)a1);
a0 = ((q[0] * dq) + 4) >> 3;
a1 = ((q[16] * dq) + 4) >> 3;
q[0] = q[16] = 0;
q2Add.val[0] = vdupq_n_s16((int16_t)a0);
q2Add.val[1] = vdupq_n_s16((int16_t)a1);
for (i = 0; i < 2; i++, dst += 4) {
dst0 = dst;
d2s32 = vld1_lane_s32((const int32_t *)dst0, d2s32, 0);
dst0 += stride;
d2s32 = vld1_lane_s32((const int32_t *)dst0, d2s32, 1);
dst0 += stride;
d4s32 = vld1_lane_s32((const int32_t *)dst0, d4s32, 0);
dst0 += stride;
d4s32 = vld1_lane_s32((const int32_t *)dst0, d4s32, 1);
for (i = 0; i < 2; i++, dst += 4) {
dst0 = dst;
d2s32 = vld1_lane_s32((const int32_t *)dst0, d2s32, 0);
dst0 += stride;
d2s32 = vld1_lane_s32((const int32_t *)dst0, d2s32, 1);
dst0 += stride;
d4s32 = vld1_lane_s32((const int32_t *)dst0, d4s32, 0);
dst0 += stride;
d4s32 = vld1_lane_s32((const int32_t *)dst0, d4s32, 1);
q1u16 = vaddw_u8(vreinterpretq_u16_s16(q2Add.val[i]),
vreinterpret_u8_s32(d2s32));
q2u16 = vaddw_u8(vreinterpretq_u16_s16(q2Add.val[i]),
vreinterpret_u8_s32(d4s32));
q1u16 = vaddw_u8(vreinterpretq_u16_s16(q2Add.val[i]),
vreinterpret_u8_s32(d2s32));
q2u16 = vaddw_u8(vreinterpretq_u16_s16(q2Add.val[i]),
vreinterpret_u8_s32(d4s32));
d2u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
d4u8 = vqmovun_s16(vreinterpretq_s16_u16(q2u16));
d2u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
d4u8 = vqmovun_s16(vreinterpretq_s16_u16(q2u16));
d2s32 = vreinterpret_s32_u8(d2u8);
d4s32 = vreinterpret_s32_u8(d4u8);
d2s32 = vreinterpret_s32_u8(d2u8);
d4s32 = vreinterpret_s32_u8(d4u8);
dst0 = dst;
vst1_lane_s32((int32_t *)dst0, d2s32, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d2s32, 1);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d4s32, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d4s32, 1);
}
return;
dst0 = dst;
vst1_lane_s32((int32_t *)dst0, d2s32, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d2s32, 1);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d4s32, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst0, d4s32, 1);
}
return;
}

281
vp8/common/arm/neon/idct_dequant_full_2x_neon.c
View File

@ -11,175 +11,172 @@
#include <arm_neon.h>
static const int16_t cospi8sqrt2minus1 = 20091;
static const int16_t sinpi8sqrt2 = 17734;
static const int16_t sinpi8sqrt2 = 17734;
// because the lowest bit in 0x8a8c is 0, we can pre-shift this
void idct_dequant_full_2x_neon(
int16_t *q,
int16_t *dq,
unsigned char *dst,
int stride) {
unsigned char *dst0, *dst1;
int32x2_t d28, d29, d30, d31;
int16x8_t q0, q1, q2, q3, q4, q5, q6, q7, q8, q9, q10, q11;
int16x8_t qEmpty = vdupq_n_s16(0);
int32x4x2_t q2tmp0, q2tmp1;
int16x8x2_t q2tmp2, q2tmp3;
int16x4_t dLow0, dLow1, dHigh0, dHigh1;
void idct_dequant_full_2x_neon(int16_t *q, int16_t *dq, unsigned char *dst,
int stride) {
unsigned char *dst0, *dst1;
int32x2_t d28, d29, d30, d31;
int16x8_t q0, q1, q2, q3, q4, q5, q6, q7, q8, q9, q10, q11;
int16x8_t qEmpty = vdupq_n_s16(0);
int32x4x2_t q2tmp0, q2tmp1;
int16x8x2_t q2tmp2, q2tmp3;
int16x4_t dLow0, dLow1, dHigh0, dHigh1;
d28 = d29 = d30 = d31 = vdup_n_s32(0);
d28 = d29 = d30 = d31 = vdup_n_s32(0);
// load dq
q0 = vld1q_s16(dq);
dq += 8;
q1 = vld1q_s16(dq);
// load dq
q0 = vld1q_s16(dq);
dq += 8;
q1 = vld1q_s16(dq);
// load q
q2 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q3 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q4 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q5 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
// load q
q2 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q3 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q4 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
q += 8;
q5 = vld1q_s16(q);
vst1q_s16(q, qEmpty);
// load src from dst
dst0 = dst;
dst1 = dst + 4;
d28 = vld1_lane_s32((const int32_t *)dst0, d28, 0);
dst0 += stride;
d28 = vld1_lane_s32((const int32_t *)dst1, d28, 1);
dst1 += stride;
d29 = vld1_lane_s32((const int32_t *)dst0, d29, 0);
dst0 += stride;
d29 = vld1_lane_s32((const int32_t *)dst1, d29, 1);
dst1 += stride;
// load src from dst
dst0 = dst;
dst1 = dst + 4;
d28 = vld1_lane_s32((const int32_t *)dst0, d28, 0);
dst0 += stride;
d28 = vld1_lane_s32((const int32_t *)dst1, d28, 1);
dst1 += stride;
d29 = vld1_lane_s32((const int32_t *)dst0, d29, 0);
dst0 += stride;
d29 = vld1_lane_s32((const int32_t *)dst1, d29, 1);
dst1 += stride;
d30 = vld1_lane_s32((const int32_t *)dst0, d30, 0);
dst0 += stride;
d30 = vld1_lane_s32((const int32_t *)dst1, d30, 1);
dst1 += stride;
d31 = vld1_lane_s32((const int32_t *)dst0, d31, 0);
d31 = vld1_lane_s32((const int32_t *)dst1, d31, 1);
d30 = vld1_lane_s32((const int32_t *)dst0, d30, 0);
dst0 += stride;
d30 = vld1_lane_s32((const int32_t *)dst1, d30, 1);
dst1 += stride;
d31 = vld1_lane_s32((const int32_t *)dst0, d31, 0);
d31 = vld1_lane_s32((const int32_t *)dst1, d31, 1);
q2 = vmulq_s16(q2, q0);
q3 = vmulq_s16(q3, q1);
q4 = vmulq_s16(q4, q0);
q5 = vmulq_s16(q5, q1);
q2 = vmulq_s16(q2, q0);
q3 = vmulq_s16(q3, q1);
q4 = vmulq_s16(q4, q0);
q5 = vmulq_s16(q5, q1);
// vswp
dLow0 = vget_low_s16(q2);
dHigh0 = vget_high_s16(q2);
dLow1 = vget_low_s16(q4);
dHigh1 = vget_high_s16(q4);
q2 = vcombine_s16(dLow0, dLow1);
q4 = vcombine_s16(dHigh0, dHigh1);
// vswp
dLow0 = vget_low_s16(q2);
dHigh0 = vget_high_s16(q2);
dLow1 = vget_low_s16(q4);
dHigh1 = vget_high_s16(q4);
q2 = vcombine_s16(dLow0, dLow1);
q4 = vcombine_s16(dHigh0, dHigh1);
dLow0 = vget_low_s16(q3);
dHigh0 = vget_high_s16(q3);
dLow1 = vget_low_s16(q5);
dHigh1 = vget_high_s16(q5);
q3 = vcombine_s16(dLow0, dLow1);
q5 = vcombine_s16(dHigh0, dHigh1);
dLow0 = vget_low_s16(q3);
dHigh0 = vget_high_s16(q3);
dLow1 = vget_low_s16(q5);
dHigh1 = vget_high_s16(q5);
q3 = vcombine_s16(dLow0, dLow1);
q5 = vcombine_s16(dHigh0, dHigh1);
q6 = vqdmulhq_n_s16(q4, sinpi8sqrt2);
q7 = vqdmulhq_n_s16(q5, sinpi8sqrt2);
q8 = vqdmulhq_n_s16(q4, cospi8sqrt2minus1);
q9 = vqdmulhq_n_s16(q5, cospi8sqrt2minus1);
q6 = vqdmulhq_n_s16(q4, sinpi8sqrt2);
q7 = vqdmulhq_n_s16(q5, sinpi8sqrt2);
q8 = vqdmulhq_n_s16(q4, cospi8sqrt2minus1);
q9 = vqdmulhq_n_s16(q5, cospi8sqrt2minus1);
q10 = vqaddq_s16(q2, q3);
q11 = vqsubq_s16(q2, q3);
q10 = vqaddq_s16(q2, q3);
q11 = vqsubq_s16(q2, q3);
q8 = vshrq_n_s16(q8, 1);
q9 = vshrq_n_s16(q9, 1);
q8 = vshrq_n_s16(q8, 1);
q9 = vshrq_n_s16(q9, 1);
q4 = vqaddq_s16(q4, q8);
q5 = vqaddq_s16(q5, q9);
q4 = vqaddq_s16(q4, q8);
q5 = vqaddq_s16(q5, q9);
q2 = vqsubq_s16(q6, q5);
q3 = vqaddq_s16(q7, q4);
q2 = vqsubq_s16(q6, q5);
q3 = vqaddq_s16(q7, q4);
q4 = vqaddq_s16(q10, q3);
q5 = vqaddq_s16(q11, q2);
q6 = vqsubq_s16(q11, q2);
q7 = vqsubq_s16(q10, q3);
q4 = vqaddq_s16(q10, q3);
q5 = vqaddq_s16(q11, q2);
q6 = vqsubq_s16(q11, q2);
q7 = vqsubq_s16(q10, q3);
q2tmp0 = vtrnq_s32(vreinterpretq_s32_s16(q4), vreinterpretq_s32_s16(q6));
q2tmp1 = vtrnq_s32(vreinterpretq_s32_s16(q5), vreinterpretq_s32_s16(q7));
q2tmp2 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[0]),
vreinterpretq_s16_s32(q2tmp1.val[0]));
q2tmp3 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[1]),
vreinterpretq_s16_s32(q2tmp1.val[1]));
q2tmp0 = vtrnq_s32(vreinterpretq_s32_s16(q4), vreinterpretq_s32_s16(q6));
q2tmp1 = vtrnq_s32(vreinterpretq_s32_s16(q5), vreinterpretq_s32_s16(q7));
q2tmp2 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[0]),
vreinterpretq_s16_s32(q2tmp1.val[0]));
q2tmp3 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[1]),
vreinterpretq_s16_s32(q2tmp1.val[1]));
// loop 2
q8 = vqdmulhq_n_s16(q2tmp2.val[1], sinpi8sqrt2);
q9 = vqdmulhq_n_s16(q2tmp3.val[1], sinpi8sqrt2);
q10 = vqdmulhq_n_s16(q2tmp2.val[1], cospi8sqrt2minus1);
q11 = vqdmulhq_n_s16(q2tmp3.val[1], cospi8sqrt2minus1);
// loop 2
q8 = vqdmulhq_n_s16(q2tmp2.val[1], sinpi8sqrt2);
q9 = vqdmulhq_n_s16(q2tmp3.val[1], sinpi8sqrt2);
q10 = vqdmulhq_n_s16(q2tmp2.val[1], cospi8sqrt2minus1);
q11 = vqdmulhq_n_s16(q2tmp3.val[1], cospi8sqrt2minus1);
q2 = vqaddq_s16(q2tmp2.val[0], q2tmp3.val[0]);
q3 = vqsubq_s16(q2tmp2.val[0], q2tmp3.val[0]);
q2 = vqaddq_s16(q2tmp2.val[0], q2tmp3.val[0]);
q3 = vqsubq_s16(q2tmp2.val[0], q2tmp3.val[0]);
q10 = vshrq_n_s16(q10, 1);
q11 = vshrq_n_s16(q11, 1);
q10 = vshrq_n_s16(q10, 1);
q11 = vshrq_n_s16(q11, 1);
q10 = vqaddq_s16(q2tmp2.val[1], q10);
q11 = vqaddq_s16(q2tmp3.val[1], q11);
q10 = vqaddq_s16(q2tmp2.val[1], q10);
q11 = vqaddq_s16(q2tmp3.val[1], q11);
q8 = vqsubq_s16(q8, q11);
q9 = vqaddq_s16(q9, q10);
q8 = vqsubq_s16(q8, q11);
q9 = vqaddq_s16(q9, q10);
q4 = vqaddq_s16(q2, q9);
q5 = vqaddq_s16(q3, q8);
q6 = vqsubq_s16(q3, q8);
q7 = vqsubq_s16(q2, q9);
q4 = vqaddq_s16(q2, q9);
q5 = vqaddq_s16(q3, q8);
q6 = vqsubq_s16(q3, q8);
q7 = vqsubq_s16(q2, q9);
q4 = vrshrq_n_s16(q4, 3);
q5 = vrshrq_n_s16(q5, 3);
q6 = vrshrq_n_s16(q6, 3);
q7 = vrshrq_n_s16(q7, 3);
q4 = vrshrq_n_s16(q4, 3);
q5 = vrshrq_n_s16(q5, 3);
q6 = vrshrq_n_s16(q6, 3);
q7 = vrshrq_n_s16(q7, 3);
q2tmp0 = vtrnq_s32(vreinterpretq_s32_s16(q4), vreinterpretq_s32_s16(q6));
q2tmp1 = vtrnq_s32(vreinterpretq_s32_s16(q5), vreinterpretq_s32_s16(q7));
q2tmp2 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[0]),
vreinterpretq_s16_s32(q2tmp1.val[0]));
q2tmp3 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[1]),
vreinterpretq_s16_s32(q2tmp1.val[1]));
q2tmp0 = vtrnq_s32(vreinterpretq_s32_s16(q4), vreinterpretq_s32_s16(q6));
q2tmp1 = vtrnq_s32(vreinterpretq_s32_s16(q5), vreinterpretq_s32_s16(q7));
q2tmp2 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[0]),
vreinterpretq_s16_s32(q2tmp1.val[0]));
q2tmp3 = vtrnq_s16(vreinterpretq_s16_s32(q2tmp0.val[1]),
vreinterpretq_s16_s32(q2tmp1.val[1]));
q4 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q2tmp2.val[0]),
vreinterpret_u8_s32(d28)));
q5 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q2tmp2.val[1]),
vreinterpret_u8_s32(d29)));
q6 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q2tmp3.val[0]),
vreinterpret_u8_s32(d30)));
q7 = vreinterpretq_s16_u16(vaddw_u8(vreinterpretq_u16_s16(q2tmp3.val[1]),
vreinterpret_u8_s32(d31)));
q4 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q2tmp2.val[0]), vreinterpret_u8_s32(d28)));
q5 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q2tmp2.val[1]), vreinterpret_u8_s32(d29)));
q6 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q2tmp3.val[0]), vreinterpret_u8_s32(d30)));
q7 = vreinterpretq_s16_u16(
vaddw_u8(vreinterpretq_u16_s16(q2tmp3.val[1]), vreinterpret_u8_s32(d31)));
d28 = vreinterpret_s32_u8(vqmovun_s16(q4));
d29 = vreinterpret_s32_u8(vqmovun_s16(q5));
d30 = vreinterpret_s32_u8(vqmovun_s16(q6));
d31 = vreinterpret_s32_u8(vqmovun_s16(q7));
d28 = vreinterpret_s32_u8(vqmovun_s16(q4));
d29 = vreinterpret_s32_u8(vqmovun_s16(q5));
d30 = vreinterpret_s32_u8(vqmovun_s16(q6));
d31 = vreinterpret_s32_u8(vqmovun_s16(q7));
dst0 = dst;
dst1 = dst + 4;
vst1_lane_s32((int32_t *)dst0, d28, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d28, 1);
dst1 += stride;
vst1_lane_s32((int32_t *)dst0, d29, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d29, 1);
dst1 += stride;
dst0 = dst;
dst1 = dst + 4;
vst1_lane_s32((int32_t *)dst0, d28, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d28, 1);
dst1 += stride;
vst1_lane_s32((int32_t *)dst0, d29, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d29, 1);
dst1 += stride;
vst1_lane_s32((int32_t *)dst0, d30, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d30, 1);
dst1 += stride;
vst1_lane_s32((int32_t *)dst0, d31, 0);
vst1_lane_s32((int32_t *)dst1, d31, 1);
return;
vst1_lane_s32((int32_t *)dst0, d30, 0);
dst0 += stride;
vst1_lane_s32((int32_t *)dst1, d30, 1);
dst1 += stride;
vst1_lane_s32((int32_t *)dst0, d31, 0);
vst1_lane_s32((int32_t *)dst1, d31, 1);
return;
}

148
vp8/common/arm/neon/iwalsh_neon.c
View File

@ -10,93 +10,91 @@
#include <arm_neon.h>
void vp8_short_inv_walsh4x4_neon(
int16_t *input,
int16_t *mb_dqcoeff) {
int16x8_t q0s16, q1s16, q2s16, q3s16;
int16x4_t d4s16, d5s16, d6s16, d7s16;
int16x4x2_t v2tmp0, v2tmp1;
int32x2x2_t v2tmp2, v2tmp3;
int16x8_t qAdd3;
void vp8_short_inv_walsh4x4_neon(int16_t *input, int16_t *mb_dqcoeff) {
int16x8_t q0s16, q1s16, q2s16, q3s16;
int16x4_t d4s16, d5s16, d6s16, d7s16;
int16x4x2_t v2tmp0, v2tmp1;
int32x2x2_t v2tmp2, v2tmp3;
int16x8_t qAdd3;
q0s16 = vld1q_s16(input);
q1s16 = vld1q_s16(input + 8);
q0s16 = vld1q_s16(input);
q1s16 = vld1q_s16(input + 8);
// 1st for loop
d4s16 = vadd_s16(vget_low_s16(q0s16), vget_high_s16(q1s16));
d6s16 = vadd_s16(vget_high_s16(q0s16), vget_low_s16(q1s16));
d5s16 = vsub_s16(vget_low_s16(q0s16), vget_high_s16(q1s16));
d7s16 = vsub_s16(vget_high_s16(q0s16), vget_low_s16(q1s16));
// 1st for loop
d4s16 = vadd_s16(vget_low_s16(q0s16), vget_high_s16(q1s16));
d6s16 = vadd_s16(vget_high_s16(q0s16), vget_low_s16(q1s16));
d5s16 = vsub_s16(vget_low_s16(q0s16), vget_high_s16(q1s16));
d7s16 = vsub_s16(vget_high_s16(q0s16), vget_low_s16(q1s16));
q2s16 = vcombine_s16(d4s16, d5s16);
q3s16 = vcombine_s16(d6s16, d7s16);
q2s16 = vcombine_s16(d4s16, d5s16);
q3s16 = vcombine_s16(d6s16, d7s16);
q0s16 = vaddq_s16(q2s16, q3s16);
q1s16 = vsubq_s16(q2s16, q3s16);
q0s16 = vaddq_s16(q2s16, q3s16);
q1s16 = vsubq_s16(q2s16, q3s16);
v2tmp2 = vtrn_s32(vreinterpret_s32_s16(vget_low_s16(q0s16)),
vreinterpret_s32_s16(vget_low_s16(q1s16)));
v2tmp3 = vtrn_s32(vreinterpret_s32_s16(vget_high_s16(q0s16)),
vreinterpret_s32_s16(vget_high_s16(q1s16)));
v2tmp0 = vtrn_s16(vreinterpret_s16_s32(v2tmp2.val[0]),
vreinterpret_s16_s32(v2tmp3.val[0]));
v2tmp1 = vtrn_s16(vreinterpret_s16_s32(v2tmp2.val[1]),
vreinterpret_s16_s32(v2tmp3.val[1]));
v2tmp2 = vtrn_s32(vreinterpret_s32_s16(vget_low_s16(q0s16)),
vreinterpret_s32_s16(vget_low_s16(q1s16)));
v2tmp3 = vtrn_s32(vreinterpret_s32_s16(vget_high_s16(q0s16)),
vreinterpret_s32_s16(vget_high_s16(q1s16)));
v2tmp0 = vtrn_s16(vreinterpret_s16_s32(v2tmp2.val[0]),
vreinterpret_s16_s32(v2tmp3.val[0]));
v2tmp1 = vtrn_s16(vreinterpret_s16_s32(v2tmp2.val[1]),
vreinterpret_s16_s32(v2tmp3.val[1]));
// 2nd for loop
d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[1]);
d6s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[0]);
d5s16 = vsub_s16(v2tmp0.val[0], v2tmp1.val[1]);
d7s16 = vsub_s16(v2tmp0.val[1], v2tmp1.val[0]);
q2s16 = vcombine_s16(d4s16, d5s16);
q3s16 = vcombine_s16(d6s16, d7s16);
// 2nd for loop
d4s16 = vadd_s16(v2tmp0.val[0], v2tmp1.val[1]);
d6s16 = vadd_s16(v2tmp0.val[1], v2tmp1.val[0]);
d5s16 = vsub_s16(v2tmp0.val[0], v2tmp1.val[1]);
d7s16 = vsub_s16(v2tmp0.val[1], v2tmp1.val[0]);
q2s16 = vcombine_s16(d4s16, d5s16);
q3s16 = vcombine_s16(d6s16, d7s16);
qAdd3 = vdupq_n_s16(3);
qAdd3 = vdupq_n_s16(3);
q0s16 = vaddq_s16(q2s16, q3s16);
q1s16 = vsubq_s16(q2s16, q3s16);
q0s16 = vaddq_s16(q2s16, q3s16);
q1s16 = vsubq_s16(q2s16, q3s16);
q0s16 = vaddq_s16(q0s16, qAdd3);
q1s16 = vaddq_s16(q1s16, qAdd3);
q0s16 = vaddq_s16(q0s16, qAdd3);
q1s16 = vaddq_s16(q1s16, qAdd3);
q0s16 = vshrq_n_s16(q0s16, 3);
q1s16 = vshrq_n_s16(q1s16, 3);
q0s16 = vshrq_n_s16(q0s16, 3);
q1s16 = vshrq_n_s16(q1s16, 3);
// store
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 0);
mb_dqcoeff += 16;
// store
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 0);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 1);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 2);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 3);
mb_dqcoeff += 16;
return;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q0s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q0s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_low_s16(q1s16), 3);
mb_dqcoeff += 16;
vst1_lane_s16(mb_dqcoeff, vget_high_s16(q1s16), 3);
mb_dqcoeff += 16;
return;
}

143
vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.c
View File

@ -12,100 +12,93 @@
#include "./vpx_config.h"
static INLINE void vp8_loop_filter_simple_horizontal_edge_neon(
unsigned char *s,
int p,
const unsigned char *blimit) {
uint8_t *sp;
uint8x16_t qblimit, q0u8;
uint8x16_t q5u8, q6u8, q7u8, q8u8, q9u8, q10u8, q14u8, q15u8;
int16x8_t q2s16, q3s16, q13s16;
int8x8_t d8s8, d9s8;
int8x16_t q2s8, q3s8, q4s8, q10s8, q11s8, q14s8;
unsigned char *s, int p, const unsigned char *blimit) {
uint8_t *sp;
uint8x16_t qblimit, q0u8;
uint8x16_t q5u8, q6u8, q7u8, q8u8, q9u8, q10u8, q14u8, q15u8;
int16x8_t q2s16, q3s16, q13s16;
int8x8_t d8s8, d9s8;
int8x16_t q2s8, q3s8, q4s8, q10s8, q11s8, q14s8;
qblimit = vdupq_n_u8(*blimit);
qblimit = vdupq_n_u8(*blimit);
sp = s - (p << 1);
q5u8 = vld1q_u8(sp);
sp += p;
q6u8 = vld1q_u8(sp);
sp += p;
q7u8 = vld1q_u8(sp);
sp += p;
q8u8 = vld1q_u8(sp);
sp = s - (p << 1);
q5u8 = vld1q_u8(sp);
sp += p;
q6u8 = vld1q_u8(sp);
sp += p;
q7u8 = vld1q_u8(sp);
sp += p;
q8u8 = vld1q_u8(sp);
q15u8 = vabdq_u8(q6u8, q7u8);
q14u8 = vabdq_u8(q5u8, q8u8);
q15u8 = vabdq_u8(q6u8, q7u8);
q14u8 = vabdq_u8(q5u8, q8u8);
q15u8 = vqaddq_u8(q15u8, q15u8);
q14u8 = vshrq_n_u8(q14u8, 1);
q0u8 = vdupq_n_u8(0x80);
q13s16 = vdupq_n_s16(3);
q15u8 = vqaddq_u8(q15u8, q14u8);
q15u8 = vqaddq_u8(q15u8, q15u8);
q14u8 = vshrq_n_u8(q14u8, 1);
q0u8 = vdupq_n_u8(0x80);
q13s16 = vdupq_n_s16(3);
q15u8 = vqaddq_u8(q15u8, q14u8);
q5u8 = veorq_u8(q5u8, q0u8);
q6u8 = veorq_u8(q6u8, q0u8);
q7u8 = veorq_u8(q7u8, q0u8);
q8u8 = veorq_u8(q8u8, q0u8);
q5u8 = veorq_u8(q5u8, q0u8);
q6u8 = veorq_u8(q6u8, q0u8);
q7u8 = veorq_u8(q7u8, q0u8);
q8u8 = veorq_u8(q8u8, q0u8);
q15u8 = vcgeq_u8(qblimit, q15u8);
q15u8 = vcgeq_u8(qblimit, q15u8);
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q7u8)),
vget_low_s8(vreinterpretq_s8_u8(q6u8)));
q3s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q7u8)),
vget_high_s8(vreinterpretq_s8_u8(q6u8)));
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q7u8)),
vget_low_s8(vreinterpretq_s8_u8(q6u8)));
q3s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q7u8)),
vget_high_s8(vreinterpretq_s8_u8(q6u8)));
q4s8 = vqsubq_s8(vreinterpretq_s8_u8(q5u8),
vreinterpretq_s8_u8(q8u8));
q4s8 = vqsubq_s8(vreinterpretq_s8_u8(q5u8), vreinterpretq_s8_u8(q8u8));
q2s16 = vmulq_s16(q2s16, q13s16);
q3s16 = vmulq_s16(q3s16, q13s16);
q2s16 = vmulq_s16(q2s16, q13s16);
q3s16 = vmulq_s16(q3s16, q13s16);
q10u8 = vdupq_n_u8(3);
q9u8 = vdupq_n_u8(4);
q10u8 = vdupq_n_u8(3);
q9u8 = vdupq_n_u8(4);
q2s16 = vaddw_s8(q2s16, vget_low_s8(q4s8));
q3s16 = vaddw_s8(q3s16, vget_high_s8(q4s8));
q2s16 = vaddw_s8(q2s16, vget_low_s8(q4s8));
q3s16 = vaddw_s8(q3s16, vget_high_s8(q4s8));
d8s8 = vqmovn_s16(q2s16);
d9s8 = vqmovn_s16(q3s16);
q4s8 = vcombine_s8(d8s8, d9s8);
d8s8 = vqmovn_s16(q2s16);
d9s8 = vqmovn_s16(q3s16);
q4s8 = vcombine_s8(d8s8, d9s8);
q14s8 = vandq_s8(q4s8, vreinterpretq_s8_u8(q15u8));
q14s8 = vandq_s8(q4s8, vreinterpretq_s8_u8(q15u8));
q2s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q10u8));
q3s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q9u8));
q2s8 = vshrq_n_s8(q2s8, 3);
q3s8 = vshrq_n_s8(q3s8, 3);
q2s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q10u8));
q3s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q9u8));
q2s8 = vshrq_n_s8(q2s8, 3);
q3s8 = vshrq_n_s8(q3s8, 3);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q6u8), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q7u8), q3s8);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q6u8), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q7u8), q3s8);
q6u8 = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
q7u8 = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
q6u8 = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
q7u8 = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
vst1q_u8(s, q7u8);
s -= p;
vst1q_u8(s, q6u8);
return;
vst1q_u8(s, q7u8);
s -= p;
vst1q_u8(s, q6u8);
return;
}
void vp8_loop_filter_bhs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
y_ptr += y_stride * 4;
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
}
void vp8_loop_filter_mbhs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, blimit);
return;
}

359
vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.c
View File

@ -15,34 +15,33 @@
#ifdef VPX_INCOMPATIBLE_GCC
static INLINE void write_2x4(unsigned char *dst, int pitch,
const uint8x8x2_t result) {
/*
* uint8x8x2_t result
00 01 02 03 | 04 05 06 07
10 11 12 13 | 14 15 16 17
---
* after vtrn_u8
00 10 02 12 | 04 14 06 16
01 11 03 13 | 05 15 07 17
*/
const uint8x8x2_t r01_u8 = vtrn_u8(result.val[0],
result.val[1]);
const uint16x4_t x_0_4 = vreinterpret_u16_u8(r01_u8.val[0]);
const uint16x4_t x_1_5 = vreinterpret_u16_u8(r01_u8.val[1]);
vst1_lane_u16((uint16_t *)dst, x_0_4, 0);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 0);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 1);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 1);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 2);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 2);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 3);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 3);
/*
* uint8x8x2_t result
00 01 02 03 | 04 05 06 07
10 11 12 13 | 14 15 16 17
---
* after vtrn_u8
00 10 02 12 | 04 14 06 16
01 11 03 13 | 05 15 07 17
*/
const uint8x8x2_t r01_u8 = vtrn_u8(result.val[0], result.val[1]);
const uint16x4_t x_0_4 = vreinterpret_u16_u8(r01_u8.val[0]);
const uint16x4_t x_1_5 = vreinterpret_u16_u8(r01_u8.val[1]);
vst1_lane_u16((uint16_t *)dst, x_0_4, 0);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 0);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 1);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 1);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 2);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 2);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_0_4, 3);
dst += pitch;
vst1_lane_u16((uint16_t *)dst, x_1_5, 3);
}
static INLINE void write_2x8(unsigned char *dst, int pitch,
@ -91,193 +90,183 @@ static INLINE void write_2x8(unsigned char *dst, int pitch,
}
#endif // VPX_INCOMPATIBLE_GCC
#ifdef VPX_INCOMPATIBLE_GCC
static INLINE
uint8x8x4_t read_4x8(unsigned char *src, int pitch) {
uint8x8x4_t x;
const uint8x8_t a = vld1_u8(src);
const uint8x8_t b = vld1_u8(src + pitch * 1);
const uint8x8_t c = vld1_u8(src + pitch * 2);
const uint8x8_t d = vld1_u8(src + pitch * 3);
const uint8x8_t e = vld1_u8(src + pitch * 4);
const uint8x8_t f = vld1_u8(src + pitch * 5);
const uint8x8_t g = vld1_u8(src + pitch * 6);
const uint8x8_t h = vld1_u8(src + pitch * 7);
const uint32x2x2_t r04_u32 = vtrn_u32(vreinterpret_u32_u8(a),
vreinterpret_u32_u8(e));
const uint32x2x2_t r15_u32 = vtrn_u32(vreinterpret_u32_u8(b),
vreinterpret_u32_u8(f));
const uint32x2x2_t r26_u32 = vtrn_u32(vreinterpret_u32_u8(c),
vreinterpret_u32_u8(g));
const uint32x2x2_t r37_u32 = vtrn_u32(vreinterpret_u32_u8(d),
vreinterpret_u32_u8(h));
const uint16x4x2_t r02_u16 = vtrn_u16(vreinterpret_u16_u32(r04_u32.val[0]),
vreinterpret_u16_u32(r26_u32.val[0]));
const uint16x4x2_t r13_u16 = vtrn_u16(vreinterpret_u16_u32(r15_u32.val[0]),
vreinterpret_u16_u32(r37_u32.val[0]));
const uint8x8x2_t r01_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[0]),
vreinterpret_u8_u16(r13_u16.val[0]));
const uint8x8x2_t r23_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[1]),
vreinterpret_u8_u16(r13_u16.val[1]));
/*
* after vtrn_u32
00 01 02 03 | 40 41 42 43
10 11 12 13 | 50 51 52 53
20 21 22 23 | 60 61 62 63
30 31 32 33 | 70 71 72 73
---
* after vtrn_u16
00 01 20 21 | 40 41 60 61
02 03 22 23 | 42 43 62 63
10 11 30 31 | 50 51 70 71
12 13 32 33 | 52 52 72 73
static INLINE uint8x8x4_t read_4x8(unsigned char *src, int pitch) {
uint8x8x4_t x;
const uint8x8_t a = vld1_u8(src);
const uint8x8_t b = vld1_u8(src + pitch * 1);
const uint8x8_t c = vld1_u8(src + pitch * 2);
const uint8x8_t d = vld1_u8(src + pitch * 3);
const uint8x8_t e = vld1_u8(src + pitch * 4);
const uint8x8_t f = vld1_u8(src + pitch * 5);
const uint8x8_t g = vld1_u8(src + pitch * 6);
const uint8x8_t h = vld1_u8(src + pitch * 7);
const uint32x2x2_t r04_u32 =
vtrn_u32(vreinterpret_u32_u8(a), vreinterpret_u32_u8(e));
const uint32x2x2_t r15_u32 =
vtrn_u32(vreinterpret_u32_u8(b), vreinterpret_u32_u8(f));
const uint32x2x2_t r26_u32 =
vtrn_u32(vreinterpret_u32_u8(c), vreinterpret_u32_u8(g));
const uint32x2x2_t r37_u32 =
vtrn_u32(vreinterpret_u32_u8(d), vreinterpret_u32_u8(h));
const uint16x4x2_t r02_u16 = vtrn_u16(vreinterpret_u16_u32(r04_u32.val[0]),
vreinterpret_u16_u32(r26_u32.val[0]));
const uint16x4x2_t r13_u16 = vtrn_u16(vreinterpret_u16_u32(r15_u32.val[0]),
vreinterpret_u16_u32(r37_u32.val[0]));
const uint8x8x2_t r01_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[0]),
vreinterpret_u8_u16(r13_u16.val[0]));
const uint8x8x2_t r23_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[1]),
vreinterpret_u8_u16(r13_u16.val[1]));
/*
* after vtrn_u32
00 01 02 03 | 40 41 42 43
10 11 12 13 | 50 51 52 53
20 21 22 23 | 60 61 62 63
30 31 32 33 | 70 71 72 73
---
* after vtrn_u16
00 01 20 21 | 40 41 60 61
02 03 22 23 | 42 43 62 63
10 11 30 31 | 50 51 70 71
12 13 32 33 | 52 52 72 73
00 01 20 21 | 40 41 60 61
10 11 30 31 | 50 51 70 71
02 03 22 23 | 42 43 62 63
12 13 32 33 | 52 52 72 73
---
* after vtrn_u8
00 10 20 30 | 40 50 60 70
01 11 21 31 | 41 51 61 71
02 12 22 32 | 42 52 62 72
03 13 23 33 | 43 53 63 73
*/
x.val[0] = r01_u8.val[0];
x.val[1] = r01_u8.val[1];
x.val[2] = r23_u8.val[0];
x.val[3] = r23_u8.val[1];
00 01 20 21 | 40 41 60 61
10 11 30 31 | 50 51 70 71
02 03 22 23 | 42 43 62 63
12 13 32 33 | 52 52 72 73
---
* after vtrn_u8
00 10 20 30 | 40 50 60 70
01 11 21 31 | 41 51 61 71
02 12 22 32 | 42 52 62 72
03 13 23 33 | 43 53 63 73
*/
x.val[0] = r01_u8.val[0];
x.val[1] = r01_u8.val[1];
x.val[2] = r23_u8.val[0];
x.val[3] = r23_u8.val[1];
return x;
return x;
}
#else
static INLINE
uint8x8x4_t read_4x8(unsigned char *src, int pitch) {
uint8x8x4_t x;
x.val[0] = x.val[1] = x.val[2] = x.val[3] = vdup_n_u8(0);
x = vld4_lane_u8(src, x, 0);
src += pitch;
x = vld4_lane_u8(src, x, 1);
src += pitch;
x = vld4_lane_u8(src, x, 2);
src += pitch;
x = vld4_lane_u8(src, x, 3);
src += pitch;
x = vld4_lane_u8(src, x, 4);
src += pitch;
x = vld4_lane_u8(src, x, 5);
src += pitch;
x = vld4_lane_u8(src, x, 6);
src += pitch;
x = vld4_lane_u8(src, x, 7);
return x;
static INLINE uint8x8x4_t read_4x8(unsigned char *src, int pitch) {
uint8x8x4_t x;
x.val[0] = x.val[1] = x.val[2] = x.val[3] = vdup_n_u8(0);
x = vld4_lane_u8(src, x, 0);
src += pitch;
x = vld4_lane_u8(src, x, 1);
src += pitch;
x = vld4_lane_u8(src, x, 2);
src += pitch;
x = vld4_lane_u8(src, x, 3);
src += pitch;
x = vld4_lane_u8(src, x, 4);
src += pitch;
x = vld4_lane_u8(src, x, 5);
src += pitch;
x = vld4_lane_u8(src, x, 6);
src += pitch;
x = vld4_lane_u8(src, x, 7);
return x;
}
#endif // VPX_INCOMPATIBLE_GCC
static INLINE void vp8_loop_filter_simple_vertical_edge_neon(
unsigned char *s,
int p,
const unsigned char *blimit) {
unsigned char *src1;
uint8x16_t qblimit, q0u8;
uint8x16_t q3u8, q4u8, q5u8, q6u8, q7u8, q11u8, q12u8, q14u8, q15u8;
int16x8_t q2s16, q13s16, q11s16;
int8x8_t d28s8, d29s8;
int8x16_t q2s8, q3s8, q10s8, q11s8, q14s8;
uint8x8x4_t d0u8x4; // d6, d7, d8, d9
uint8x8x4_t d1u8x4; // d10, d11, d12, d13
uint8x8x2_t d2u8x2; // d12, d13
uint8x8x2_t d3u8x2; // d14, d15
unsigned char *s, int p, const unsigned char *blimit) {
unsigned char *src1;
uint8x16_t qblimit, q0u8;
uint8x16_t q3u8, q4u8, q5u8, q6u8, q7u8, q11u8, q12u8, q14u8, q15u8;
int16x8_t q2s16, q13s16, q11s16;
int8x8_t d28s8, d29s8;
int8x16_t q2s8, q3s8, q10s8, q11s8, q14s8;
uint8x8x4_t d0u8x4; // d6, d7, d8, d9
uint8x8x4_t d1u8x4; // d10, d11, d12, d13
uint8x8x2_t d2u8x2; // d12, d13
uint8x8x2_t d3u8x2; // d14, d15
qblimit = vdupq_n_u8(*blimit);
qblimit = vdupq_n_u8(*blimit);
src1 = s - 2;
d0u8x4 = read_4x8(src1, p);
src1 += p * 8;
d1u8x4 = read_4x8(src1, p);
src1 = s - 2;
d0u8x4 = read_4x8(src1, p);
src1 += p * 8;
d1u8x4 = read_4x8(src1, p);
q3u8 = vcombine_u8(d0u8x4.val[0], d1u8x4.val[0]); // d6 d10
q4u8 = vcombine_u8(d0u8x4.val[2], d1u8x4.val[2]); // d8 d12
q5u8 = vcombine_u8(d0u8x4.val[1], d1u8x4.val[1]); // d7 d11
q6u8 = vcombine_u8(d0u8x4.val[3], d1u8x4.val[3]); // d9 d13
q3u8 = vcombine_u8(d0u8x4.val[0], d1u8x4.val[0]); // d6 d10
q4u8 = vcombine_u8(d0u8x4.val[2], d1u8x4.val[2]); // d8 d12
q5u8 = vcombine_u8(d0u8x4.val[1], d1u8x4.val[1]); // d7 d11
q6u8 = vcombine_u8(d0u8x4.val[3], d1u8x4.val[3]); // d9 d13
q15u8 = vabdq_u8(q5u8, q4u8);
q14u8 = vabdq_u8(q3u8, q6u8);
q15u8 = vabdq_u8(q5u8, q4u8);
q14u8 = vabdq_u8(q3u8, q6u8);
q15u8 = vqaddq_u8(q15u8, q15u8);
q14u8 = vshrq_n_u8(q14u8, 1);
q0u8 = vdupq_n_u8(0x80);
q11s16 = vdupq_n_s16(3);
q15u8 = vqaddq_u8(q15u8, q14u8);
q15u8 = vqaddq_u8(q15u8, q15u8);
q14u8 = vshrq_n_u8(q14u8, 1);
q0u8 = vdupq_n_u8(0x80);
q11s16 = vdupq_n_s16(3);
q15u8 = vqaddq_u8(q15u8, q14u8);
q3u8 = veorq_u8(q3u8, q0u8);
q4u8 = veorq_u8(q4u8, q0u8);
q5u8 = veorq_u8(q5u8, q0u8);
q6u8 = veorq_u8(q6u8, q0u8);
q3u8 = veorq_u8(q3u8, q0u8);
q4u8 = veorq_u8(q4u8, q0u8);
q5u8 = veorq_u8(q5u8, q0u8);
q6u8 = veorq_u8(q6u8, q0u8);
q15u8 = vcgeq_u8(qblimit, q15u8);
q15u8 = vcgeq_u8(qblimit, q15u8);
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q4u8)),
vget_low_s8(vreinterpretq_s8_u8(q5u8)));
q13s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q4u8)),
vget_high_s8(vreinterpretq_s8_u8(q5u8)));
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q4u8)),
vget_low_s8(vreinterpretq_s8_u8(q5u8)));
q13s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q4u8)),
vget_high_s8(vreinterpretq_s8_u8(q5u8)));
q14s8 = vqsubq_s8(vreinterpretq_s8_u8(q3u8),
vreinterpretq_s8_u8(q6u8));
q14s8 = vqsubq_s8(vreinterpretq_s8_u8(q3u8), vreinterpretq_s8_u8(q6u8));
q2s16 = vmulq_s16(q2s16, q11s16);
q13s16 = vmulq_s16(q13s16, q11s16);
q2s16 = vmulq_s16(q2s16, q11s16);
q13s16 = vmulq_s16(q13s16, q11s16);
q11u8 = vdupq_n_u8(3);
q12u8 = vdupq_n_u8(4);
q11u8 = vdupq_n_u8(3);
q12u8 = vdupq_n_u8(4);
q2s16 = vaddw_s8(q2s16, vget_low_s8(q14s8));
q13s16 = vaddw_s8(q13s16, vget_high_s8(q14s8));
q2s16 = vaddw_s8(q2s16, vget_low_s8(q14s8));
q13s16 = vaddw_s8(q13s16, vget_high_s8(q14s8));
d28s8 = vqmovn_s16(q2s16);
d29s8 = vqmovn_s16(q13s16);
q14s8 = vcombine_s8(d28s8, d29s8);
d28s8 = vqmovn_s16(q2s16);
d29s8 = vqmovn_s16(q13s16);
q14s8 = vcombine_s8(d28s8, d29s8);
q14s8 = vandq_s8(q14s8, vreinterpretq_s8_u8(q15u8));
q14s8 = vandq_s8(q14s8, vreinterpretq_s8_u8(q15u8));
q2s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q11u8));
q3s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q12u8));
q2s8 = vshrq_n_s8(q2s8, 3);
q14s8 = vshrq_n_s8(q3s8, 3);
q2s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q11u8));
q3s8 = vqaddq_s8(q14s8, vreinterpretq_s8_u8(q12u8));
q2s8 = vshrq_n_s8(q2s8, 3);
q14s8 = vshrq_n_s8(q3s8, 3);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q5u8), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q4u8), q14s8);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q5u8), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q4u8), q14s8);
q6u8 = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
q7u8 = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
q6u8 = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
q7u8 = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
d2u8x2.val[0] = vget_low_u8(q6u8); // d12
d2u8x2.val[1] = vget_low_u8(q7u8); // d14
d3u8x2.val[0] = vget_high_u8(q6u8); // d13
d3u8x2.val[1] = vget_high_u8(q7u8); // d15
d2u8x2.val[0] = vget_low_u8(q6u8); // d12
d2u8x2.val[1] = vget_low_u8(q7u8); // d14
d3u8x2.val[0] = vget_high_u8(q6u8); // d13
d3u8x2.val[1] = vget_high_u8(q7u8); // d15
src1 = s - 1;
write_2x8(src1, p, d2u8x2, d3u8x2);
src1 = s - 1;
write_2x8(src1, p, d2u8x2, d3u8x2);
}
void vp8_loop_filter_bvs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
return;
void vp8_loop_filter_bvs_neon(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
y_ptr += 4;
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
return;
}
void vp8_loop_filter_mbvs_neon(
unsigned char *y_ptr,
int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
return;
void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, blimit);
return;
}

1066
vp8/common/arm/neon/mbloopfilter_neon.c
View File

File diff suppressed because it is too large Load Diff

168
vp8/common/arm/neon/shortidct4x4llm_neon.c
View File

@ -11,113 +11,109 @@
#include <arm_neon.h>
static const int16_t cospi8sqrt2minus1 = 20091;
static const int16_t sinpi8sqrt2 = 35468;
static const int16_t sinpi8sqrt2 = 35468;
void vp8_short_idct4x4llm_neon(
int16_t *input,
unsigned char *pred_ptr,
int pred_stride,
unsigned char *dst_ptr,
int dst_stride) {
int i;
uint32x2_t d6u32 = vdup_n_u32(0);
uint8x8_t d1u8;
int16x4_t d2, d3, d4, d5, d10, d11, d12, d13;
uint16x8_t q1u16;
int16x8_t q1s16, q2s16, q3s16, q4s16;
int32x2x2_t v2tmp0, v2tmp1;
int16x4x2_t v2tmp2, v2tmp3;
void vp8_short_idct4x4llm_neon(int16_t *input, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride) {
int i;
uint32x2_t d6u32 = vdup_n_u32(0);
uint8x8_t d1u8;
int16x4_t d2, d3, d4, d5, d10, d11, d12, d13;
uint16x8_t q1u16;
int16x8_t q1s16, q2s16, q3s16, q4s16;
int32x2x2_t v2tmp0, v2tmp1;
int16x4x2_t v2tmp2, v2tmp3;
d2 = vld1_s16(input);
d3 = vld1_s16(input + 4);
d4 = vld1_s16(input + 8);
d5 = vld1_s16(input + 12);
d2 = vld1_s16(input);
d3 = vld1_s16(input + 4);
d4 = vld1_s16(input + 8);
d5 = vld1_s16(input + 12);
// 1st for loop
q1s16 = vcombine_s16(d2, d4); // Swap d3 d4 here
q2s16 = vcombine_s16(d3, d5);
// 1st for loop
q1s16 = vcombine_s16(d2, d4); // Swap d3 d4 here
q2s16 = vcombine_s16(d3, d5);
q3s16 = vqdmulhq_n_s16(q2s16, sinpi8sqrt2);
q4s16 = vqdmulhq_n_s16(q2s16, cospi8sqrt2minus1);
q3s16 = vqdmulhq_n_s16(q2s16, sinpi8sqrt2);
q4s16 = vqdmulhq_n_s16(q2s16, cospi8sqrt2minus1);
d12 = vqadd_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // a1
d13 = vqsub_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // b1
d12 = vqadd_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // a1
d13 = vqsub_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // b1
q3s16 = vshrq_n_s16(q3s16, 1);
q4s16 = vshrq_n_s16(q4s16, 1);
q3s16 = vshrq_n_s16(q3s16, 1);
q4s16 = vshrq_n_s16(q4s16, 1);
q3s16 = vqaddq_s16(q3s16, q2s16);
q4s16 = vqaddq_s16(q4s16, q2s16);
q3s16 = vqaddq_s16(q3s16, q2s16);
q4s16 = vqaddq_s16(q4s16, q2s16);
d10 = vqsub_s16(vget_low_s16(q3s16), vget_high_s16(q4s16)); // c1
d11 = vqadd_s16(vget_high_s16(q3s16), vget_low_s16(q4s16)); // d1
d10 = vqsub_s16(vget_low_s16(q3s16), vget_high_s16(q4s16)); // c1
d11 = vqadd_s16(vget_high_s16(q3s16), vget_low_s16(q4s16)); // d1
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
v2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
v2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
v2tmp2 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[0]),
vreinterpret_s16_s32(v2tmp1.val[0]));
v2tmp3 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[1]),
vreinterpret_s16_s32(v2tmp1.val[1]));
v2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
v2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
v2tmp2 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[0]),
vreinterpret_s16_s32(v2tmp1.val[0]));
v2tmp3 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[1]),
vreinterpret_s16_s32(v2tmp1.val[1]));
// 2nd for loop
q1s16 = vcombine_s16(v2tmp2.val[0], v2tmp3.val[0]);
q2s16 = vcombine_s16(v2tmp2.val[1], v2tmp3.val[1]);
// 2nd for loop
q1s16 = vcombine_s16(v2tmp2.val[0], v2tmp3.val[0]);
q2s16 = vcombine_s16(v2tmp2.val[1], v2tmp3.val[1]);
q3s16 = vqdmulhq_n_s16(q2s16, sinpi8sqrt2);
q4s16 = vqdmulhq_n_s16(q2s16, cospi8sqrt2minus1);
q3s16 = vqdmulhq_n_s16(q2s16, sinpi8sqrt2);
q4s16 = vqdmulhq_n_s16(q2s16, cospi8sqrt2minus1);
d12 = vqadd_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // a1
d13 = vqsub_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // b1
d12 = vqadd_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // a1
d13 = vqsub_s16(vget_low_s16(q1s16), vget_high_s16(q1s16)); // b1
q3s16 = vshrq_n_s16(q3s16, 1);
q4s16 = vshrq_n_s16(q4s16, 1);
q3s16 = vshrq_n_s16(q3s16, 1);
q4s16 = vshrq_n_s16(q4s16, 1);
q3s16 = vqaddq_s16(q3s16, q2s16);
q4s16 = vqaddq_s16(q4s16, q2s16);
q3s16 = vqaddq_s16(q3s16, q2s16);
q4s16 = vqaddq_s16(q4s16, q2s16);
d10 = vqsub_s16(vget_low_s16(q3s16), vget_high_s16(q4s16)); // c1
d11 = vqadd_s16(vget_high_s16(q3s16), vget_low_s16(q4s16)); // d1
d10 = vqsub_s16(vget_low_s16(q3s16), vget_high_s16(q4s16)); // c1
d11 = vqadd_s16(vget_high_s16(q3s16), vget_low_s16(q4s16)); // d1
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vqadd_s16(d12, d11);
d3 = vqadd_s16(d13, d10);
d4 = vqsub_s16(d13, d10);
d5 = vqsub_s16(d12, d11);
d2 = vrshr_n_s16(d2, 3);
d3 = vrshr_n_s16(d3, 3);
d4 = vrshr_n_s16(d4, 3);
d5 = vrshr_n_s16(d5, 3);
d2 = vrshr_n_s16(d2, 3);
d3 = vrshr_n_s16(d3, 3);
d4 = vrshr_n_s16(d4, 3);
d5 = vrshr_n_s16(d5, 3);
v2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
v2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
v2tmp2 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[0]),
vreinterpret_s16_s32(v2tmp1.val[0]));
v2tmp3 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[1]),
vreinterpret_s16_s32(v2tmp1.val[1]));
v2tmp0 = vtrn_s32(vreinterpret_s32_s16(d2), vreinterpret_s32_s16(d4));
v2tmp1 = vtrn_s32(vreinterpret_s32_s16(d3), vreinterpret_s32_s16(d5));
v2tmp2 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[0]),
vreinterpret_s16_s32(v2tmp1.val[0]));
v2tmp3 = vtrn_s16(vreinterpret_s16_s32(v2tmp0.val[1]),
vreinterpret_s16_s32(v2tmp1.val[1]));
q1s16 = vcombine_s16(v2tmp2.val[0], v2tmp2.val[1]);
q2s16 = vcombine_s16(v2tmp3.val[0], v2tmp3.val[1]);
q1s16 = vcombine_s16(v2tmp2.val[0], v2tmp2.val[1]);
q2s16 = vcombine_s16(v2tmp3.val[0], v2tmp3.val[1]);
// dc_only_idct_add
for (i = 0; i < 2; i++, q1s16 = q2s16) {
d6u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d6u32, 0);
pred_ptr += pred_stride;
d6u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d6u32, 1);
pred_ptr += pred_stride;
// dc_only_idct_add
for (i = 0; i < 2; i++, q1s16 = q2s16) {
d6u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d6u32, 0);
pred_ptr += pred_stride;
d6u32 = vld1_lane_u32((const uint32_t *)pred_ptr, d6u32, 1);
pred_ptr += pred_stride;
q1u16 = vaddw_u8(vreinterpretq_u16_s16(q1s16),
vreinterpret_u8_u32(d6u32));
d1u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
q1u16 = vaddw_u8(vreinterpretq_u16_s16(q1s16), vreinterpret_u8_u32(d6u32));
d1u8 = vqmovun_s16(vreinterpretq_s16_u16(q1u16));
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d1u8), 0);
dst_ptr += dst_stride;
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d1u8), 1);
dst_ptr += dst_stride;
}
return;
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d1u8), 0);
dst_ptr += dst_stride;
vst1_lane_u32((uint32_t *)dst_ptr, vreinterpret_u32_u8(d1u8), 1);
dst_ptr += dst_stride;
}
return;
}

2382
vp8/common/arm/neon/sixtappredict_neon.c
View File

File diff suppressed because it is too large Load Diff

918
vp8/common/arm/neon/vp8_loopfilter_neon.c
View File

@ -12,539 +12,525 @@
#include "./vpx_config.h"
#include "vpx_ports/arm.h"
static INLINE void vp8_loop_filter_neon(
uint8x16_t qblimit, // flimit
uint8x16_t qlimit, // limit
uint8x16_t qthresh, // thresh
uint8x16_t q3, // p3
uint8x16_t q4, // p2
uint8x16_t q5, // p1
uint8x16_t q6, // p0
uint8x16_t q7, // q0
uint8x16_t q8, // q1
uint8x16_t q9, // q2
uint8x16_t q10, // q3
uint8x16_t *q5r, // p1
uint8x16_t *q6r, // p0
uint8x16_t *q7r, // q0
uint8x16_t *q8r) { // q1
uint8x16_t q0u8, q1u8, q2u8, q11u8, q12u8, q13u8, q14u8, q15u8;
int16x8_t q2s16, q11s16;
uint16x8_t q4u16;
int8x16_t q1s8, q2s8, q10s8, q11s8, q12s8, q13s8;
int8x8_t d2s8, d3s8;
static INLINE void vp8_loop_filter_neon(uint8x16_t qblimit, // flimit
uint8x16_t qlimit, // limit
uint8x16_t qthresh, // thresh
uint8x16_t q3, // p3
uint8x16_t q4, // p2
uint8x16_t q5, // p1
uint8x16_t q6, // p0
uint8x16_t q7, // q0
uint8x16_t q8, // q1
uint8x16_t q9, // q2
uint8x16_t q10, // q3
uint8x16_t *q5r, // p1
uint8x16_t *q6r, // p0
uint8x16_t *q7r, // q0
uint8x16_t *q8r) { // q1
uint8x16_t q0u8, q1u8, q2u8, q11u8, q12u8, q13u8, q14u8, q15u8;
int16x8_t q2s16, q11s16;
uint16x8_t q4u16;
int8x16_t q1s8, q2s8, q10s8, q11s8, q12s8, q13s8;
int8x8_t d2s8, d3s8;
q11u8 = vabdq_u8(q3, q4);
q12u8 = vabdq_u8(q4, q5);
q13u8 = vabdq_u8(q5, q6);
q14u8 = vabdq_u8(q8, q7);
q3 = vabdq_u8(q9, q8);
q4 = vabdq_u8(q10, q9);
q11u8 = vabdq_u8(q3, q4);
q12u8 = vabdq_u8(q4, q5);
q13u8 = vabdq_u8(q5, q6);
q14u8 = vabdq_u8(q8, q7);
q3 = vabdq_u8(q9, q8);
q4 = vabdq_u8(q10, q9);
q11u8 = vmaxq_u8(q11u8, q12u8);
q12u8 = vmaxq_u8(q13u8, q14u8);
q3 = vmaxq_u8(q3, q4);
q15u8 = vmaxq_u8(q11u8, q12u8);
q11u8 = vmaxq_u8(q11u8, q12u8);
q12u8 = vmaxq_u8(q13u8, q14u8);
q3 = vmaxq_u8(q3, q4);
q15u8 = vmaxq_u8(q11u8, q12u8);
q9 = vabdq_u8(q6, q7);
q9 = vabdq_u8(q6, q7);
// vp8_hevmask
q13u8 = vcgtq_u8(q13u8, qthresh);
q14u8 = vcgtq_u8(q14u8, qthresh);
q15u8 = vmaxq_u8(q15u8, q3);
// vp8_hevmask
q13u8 = vcgtq_u8(q13u8, qthresh);
q14u8 = vcgtq_u8(q14u8, qthresh);
q15u8 = vmaxq_u8(q15u8, q3);
q2u8 = vabdq_u8(q5, q8);
q9 = vqaddq_u8(q9, q9);
q2u8 = vabdq_u8(q5, q8);
q9 = vqaddq_u8(q9, q9);
q15u8 = vcgeq_u8(qlimit, q15u8);
q15u8 = vcgeq_u8(qlimit, q15u8);
// vp8_filter() function
// convert to signed
q10 = vdupq_n_u8(0x80);
q8 = veorq_u8(q8, q10);
q7 = veorq_u8(q7, q10);
q6 = veorq_u8(q6, q10);
q5 = veorq_u8(q5, q10);
// vp8_filter() function
// convert to signed
q10 = vdupq_n_u8(0x80);
q8 = veorq_u8(q8, q10);
q7 = veorq_u8(q7, q10);
q6 = veorq_u8(q6, q10);
q5 = veorq_u8(q5, q10);
q2u8 = vshrq_n_u8(q2u8, 1);
q9 = vqaddq_u8(q9, q2u8);
q2u8 = vshrq_n_u8(q2u8, 1);
q9 = vqaddq_u8(q9, q2u8);
q10 = vdupq_n_u8(3);
q10 = vdupq_n_u8(3);
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q7)),
vget_low_s8(vreinterpretq_s8_u8(q6)));
q11s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q7)),
vget_high_s8(vreinterpretq_s8_u8(q6)));
q2s16 = vsubl_s8(vget_low_s8(vreinterpretq_s8_u8(q7)),
vget_low_s8(vreinterpretq_s8_u8(q6)));
q11s16 = vsubl_s8(vget_high_s8(vreinterpretq_s8_u8(q7)),
vget_high_s8(vreinterpretq_s8_u8(q6)));
q9 = vcgeq_u8(qblimit, q9);
q9 = vcgeq_u8(qblimit, q9);
q1s8 = vqsubq_s8(vreinterpretq_s8_u8(q5),
vreinterpretq_s8_u8(q8));
q1s8 = vqsubq_s8(vreinterpretq_s8_u8(q5), vreinterpretq_s8_u8(q8));
q14u8 = vorrq_u8(q13u8, q14u8);
q14u8 = vorrq_u8(q13u8, q14u8);
q4u16 = vmovl_u8(vget_low_u8(q10));
q2s16 = vmulq_s16(q2s16, vreinterpretq_s16_u16(q4u16));
q11s16 = vmulq_s16(q11s16, vreinterpretq_s16_u16(q4u16));
q4u16 = vmovl_u8(vget_low_u8(q10));
q2s16 = vmulq_s16(q2s16, vreinterpretq_s16_u16(q4u16));
q11s16 = vmulq_s16(q11s16, vreinterpretq_s16_u16(q4u16));
q1u8 = vandq_u8(vreinterpretq_u8_s8(q1s8), q14u8);
q15u8 = vandq_u8(q15u8, q9);
q1u8 = vandq_u8(vreinterpretq_u8_s8(q1s8), q14u8);
q15u8 = vandq_u8(q15u8, q9);
q1s8 = vreinterpretq_s8_u8(q1u8);
q2s16 = vaddw_s8(q2s16, vget_low_s8(q1s8));
q11s16 = vaddw_s8(q11s16, vget_high_s8(q1s8));
q1s8 = vreinterpretq_s8_u8(q1u8);
q2s16 = vaddw_s8(q2s16, vget_low_s8(q1s8));
q11s16 = vaddw_s8(q11s16, vget_high_s8(q1s8));
q9 = vdupq_n_u8(4);
// vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
d2s8 = vqmovn_s16(q2s16);
d3s8 = vqmovn_s16(q11s16);
q1s8 = vcombine_s8(d2s8, d3s8);
q1u8 = vandq_u8(vreinterpretq_u8_s8(q1s8), q15u8);
q1s8 = vreinterpretq_s8_u8(q1u8);
q9 = vdupq_n_u8(4);
// vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
d2s8 = vqmovn_s16(q2s16);
d3s8 = vqmovn_s16(q11s16);
q1s8 = vcombine_s8(d2s8, d3s8);
q1u8 = vandq_u8(vreinterpretq_u8_s8(q1s8), q15u8);
q1s8 = vreinterpretq_s8_u8(q1u8);
q2s8 = vqaddq_s8(q1s8, vreinterpretq_s8_u8(q10));
q1s8 = vqaddq_s8(q1s8, vreinterpretq_s8_u8(q9));
q2s8 = vshrq_n_s8(q2s8, 3);
q1s8 = vshrq_n_s8(q1s8, 3);
q2s8 = vqaddq_s8(q1s8, vreinterpretq_s8_u8(q10));
q1s8 = vqaddq_s8(q1s8, vreinterpretq_s8_u8(q9));
q2s8 = vshrq_n_s8(q2s8, 3);
q1s8 = vshrq_n_s8(q1s8, 3);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q6), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q7), q1s8);
q11s8 = vqaddq_s8(vreinterpretq_s8_u8(q6), q2s8);
q10s8 = vqsubq_s8(vreinterpretq_s8_u8(q7), q1s8);
q1s8 = vrshrq_n_s8(q1s8, 1);
q1s8 = vbicq_s8(q1s8, vreinterpretq_s8_u8(q14u8));
q1s8 = vrshrq_n_s8(q1s8, 1);
q1s8 = vbicq_s8(q1s8, vreinterpretq_s8_u8(q14u8));
q13s8 = vqaddq_s8(vreinterpretq_s8_u8(q5), q1s8);
q12s8 = vqsubq_s8(vreinterpretq_s8_u8(q8), q1s8);
q13s8 = vqaddq_s8(vreinterpretq_s8_u8(q5), q1s8);
q12s8 = vqsubq_s8(vreinterpretq_s8_u8(q8), q1s8);
q0u8 = vdupq_n_u8(0x80);
*q8r = veorq_u8(vreinterpretq_u8_s8(q12s8), q0u8);
*q7r = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
*q6r = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
*q5r = veorq_u8(vreinterpretq_u8_s8(q13s8), q0u8);
return;
q0u8 = vdupq_n_u8(0x80);
*q8r = veorq_u8(vreinterpretq_u8_s8(q12s8), q0u8);
*q7r = veorq_u8(vreinterpretq_u8_s8(q10s8), q0u8);
*q6r = veorq_u8(vreinterpretq_u8_s8(q11s8), q0u8);
*q5r = veorq_u8(vreinterpretq_u8_s8(q13s8), q0u8);
return;
}
void vp8_loop_filter_horizontal_edge_y_neon(
unsigned char *src,
int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh) {
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
void vp8_loop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh) {
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
src -= (pitch << 2);
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
src -= (pitch << 2);
q3 = vld1q_u8(src);
src += pitch;
q4 = vld1q_u8(src);
src += pitch;
q5 = vld1q_u8(src);
src += pitch;
q6 = vld1q_u8(src);
src += pitch;
q7 = vld1q_u8(src);
src += pitch;
q8 = vld1q_u8(src);
src += pitch;
q9 = vld1q_u8(src);
src += pitch;
q10 = vld1q_u8(src);
q3 = vld1q_u8(src);
src += pitch;
q4 = vld1q_u8(src);
src += pitch;
q5 = vld1q_u8(src);
src += pitch;
q6 = vld1q_u8(src);
src += pitch;
q7 = vld1q_u8(src);
src += pitch;
q8 = vld1q_u8(src);
src += pitch;
q9 = vld1q_u8(src);
src += pitch;
q10 = vld1q_u8(src);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4,
q5, q6, q7, q8, q9, q10,
&q5, &q6, &q7, &q8);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4, q5, q6, q7, q8, q9,
q10, &q5, &q6, &q7, &q8);
src -= (pitch * 5);
vst1q_u8(src, q5);
src += pitch;
vst1q_u8(src, q6);
src += pitch;
vst1q_u8(src, q7);
src += pitch;
vst1q_u8(src, q8);
return;
src -= (pitch * 5);
vst1q_u8(src, q5);
src += pitch;
vst1q_u8(src, q6);
src += pitch;
vst1q_u8(src, q7);
src += pitch;
vst1q_u8(src, q8);
return;
}
void vp8_loop_filter_horizontal_edge_uv_neon(
unsigned char *u,
int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh,
unsigned char *v) {
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
void vp8_loop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh,
unsigned char *v) {
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
u -= (pitch << 2);
v -= (pitch << 2);
u -= (pitch << 2);
v -= (pitch << 2);
d6 = vld1_u8(u);
u += pitch;
d7 = vld1_u8(v);
v += pitch;
d8 = vld1_u8(u);
u += pitch;
d9 = vld1_u8(v);
v += pitch;
d10 = vld1_u8(u);
u += pitch;
d11 = vld1_u8(v);
v += pitch;
d12 = vld1_u8(u);
u += pitch;
d13 = vld1_u8(v);
v += pitch;
d14 = vld1_u8(u);
u += pitch;
d15 = vld1_u8(v);
v += pitch;
d16 = vld1_u8(u);
u += pitch;
d17 = vld1_u8(v);
v += pitch;
d18 = vld1_u8(u);
u += pitch;
d19 = vld1_u8(v);
v += pitch;
d20 = vld1_u8(u);
d21 = vld1_u8(v);
d6 = vld1_u8(u);
u += pitch;
d7 = vld1_u8(v);
v += pitch;
d8 = vld1_u8(u);
u += pitch;
d9 = vld1_u8(v);
v += pitch;
d10 = vld1_u8(u);
u += pitch;
d11 = vld1_u8(v);
v += pitch;
d12 = vld1_u8(u);
u += pitch;
d13 = vld1_u8(v);
v += pitch;
d14 = vld1_u8(u);
u += pitch;
d15 = vld1_u8(v);
v += pitch;
d16 = vld1_u8(u);
u += pitch;
d17 = vld1_u8(v);
v += pitch;
d18 = vld1_u8(u);
u += pitch;
d19 = vld1_u8(v);
v += pitch;
d20 = vld1_u8(u);
d21 = vld1_u8(v);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4,
q5, q6, q7, q8, q9, q10,
&q5, &q6, &q7, &q8);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4, q5, q6, q7, q8, q9,
q10, &q5, &q6, &q7, &q8);
u -= (pitch * 5);
vst1_u8(u, vget_low_u8(q5));
u += pitch;
vst1_u8(u, vget_low_u8(q6));
u += pitch;
vst1_u8(u, vget_low_u8(q7));
u += pitch;
vst1_u8(u, vget_low_u8(q8));
u -= (pitch * 5);
vst1_u8(u, vget_low_u8(q5));
u += pitch;
vst1_u8(u, vget_low_u8(q6));
u += pitch;
vst1_u8(u, vget_low_u8(q7));
u += pitch;
vst1_u8(u, vget_low_u8(q8));
v -= (pitch * 5);
vst1_u8(v, vget_high_u8(q5));
v += pitch;
vst1_u8(v, vget_high_u8(q6));
v += pitch;
vst1_u8(v, vget_high_u8(q7));
v += pitch;
vst1_u8(v, vget_high_u8(q8));
return;
v -= (pitch * 5);
vst1_u8(v, vget_high_u8(q5));
v += pitch;
vst1_u8(v, vget_high_u8(q6));
v += pitch;
vst1_u8(v, vget_high_u8(q7));
v += pitch;
vst1_u8(v, vget_high_u8(q8));
return;
}
static INLINE void write_4x8(unsigned char *dst, int pitch,
const uint8x8x4_t result) {
#ifdef VPX_INCOMPATIBLE_GCC
/*
* uint8x8x4_t result
00 01 02 03 | 04 05 06 07
10 11 12 13 | 14 15 16 17
20 21 22 23 | 24 25 26 27
30 31 32 33 | 34 35 36 37
---
* after vtrn_u16
00 01 20 21 | 04 05 24 25
02 03 22 23 | 06 07 26 27
10 11 30 31 | 14 15 34 35
12 13 32 33 | 16 17 36 37
---
* after vtrn_u8
00 10 20 30 | 04 14 24 34
01 11 21 31 | 05 15 25 35
02 12 22 32 | 06 16 26 36
03 13 23 33 | 07 17 27 37
*/
const uint16x4x2_t r02_u16 = vtrn_u16(vreinterpret_u16_u8(result.val[0]),
vreinterpret_u16_u8(result.val[2]));
const uint16x4x2_t r13_u16 = vtrn_u16(vreinterpret_u16_u8(result.val[1]),
vreinterpret_u16_u8(result.val[3]));
const uint8x8x2_t r01_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[0]),
vreinterpret_u8_u16(r13_u16.val[0]));
const uint8x8x2_t r23_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[1]),
vreinterpret_u8_u16(r13_u16.val[1]));
const uint32x2_t x_0_4 = vreinterpret_u32_u8(r01_u8.val[0]);
const uint32x2_t x_1_5 = vreinterpret_u32_u8(r01_u8.val[1]);
const uint32x2_t x_2_6 = vreinterpret_u32_u8(r23_u8.val[0]);
const uint32x2_t x_3_7 = vreinterpret_u32_u8(r23_u8.val[1]);
vst1_lane_u32((uint32_t *)dst, x_0_4, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_1_5, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_2_6, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_3_7, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_0_4, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_1_5, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_2_6, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_3_7, 1);
/*
* uint8x8x4_t result
00 01 02 03 | 04 05 06 07
10 11 12 13 | 14 15 16 17
20 21 22 23 | 24 25 26 27
30 31 32 33 | 34 35 36 37
---
* after vtrn_u16
00 01 20 21 | 04 05 24 25
02 03 22 23 | 06 07 26 27
10 11 30 31 | 14 15 34 35
12 13 32 33 | 16 17 36 37
---
* after vtrn_u8
00 10 20 30 | 04 14 24 34
01 11 21 31 | 05 15 25 35
02 12 22 32 | 06 16 26 36
03 13 23 33 | 07 17 27 37
*/
const uint16x4x2_t r02_u16 = vtrn_u16(vreinterpret_u16_u8(result.val[0]),
vreinterpret_u16_u8(result.val[2]));
const uint16x4x2_t r13_u16 = vtrn_u16(vreinterpret_u16_u8(result.val[1]),
vreinterpret_u16_u8(result.val[3]));
const uint8x8x2_t r01_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[0]),
vreinterpret_u8_u16(r13_u16.val[0]));
const uint8x8x2_t r23_u8 = vtrn_u8(vreinterpret_u8_u16(r02_u16.val[1]),
vreinterpret_u8_u16(r13_u16.val[1]));
const uint32x2_t x_0_4 = vreinterpret_u32_u8(r01_u8.val[0]);
const uint32x2_t x_1_5 = vreinterpret_u32_u8(r01_u8.val[1]);
const uint32x2_t x_2_6 = vreinterpret_u32_u8(r23_u8.val[0]);
const uint32x2_t x_3_7 = vreinterpret_u32_u8(r23_u8.val[1]);
vst1_lane_u32((uint32_t *)dst, x_0_4, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_1_5, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_2_6, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_3_7, 0);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_0_4, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_1_5, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_2_6, 1);
dst += pitch;
vst1_lane_u32((uint32_t *)dst, x_3_7, 1);
#else
vst4_lane_u8(dst, result, 0);
dst += pitch;
vst4_lane_u8(dst, result, 1);
dst += pitch;
vst4_lane_u8(dst, result, 2);
dst += pitch;
vst4_lane_u8(dst, result, 3);
dst += pitch;
vst4_lane_u8(dst, result, 4);
dst += pitch;
vst4_lane_u8(dst, result, 5);
dst += pitch;
vst4_lane_u8(dst, result, 6);
dst += pitch;
vst4_lane_u8(dst, result, 7);
vst4_lane_u8(dst, result, 0);
dst += pitch;
vst4_lane_u8(dst, result, 1);
dst += pitch;
vst4_lane_u8(dst, result, 2);
dst += pitch;
vst4_lane_u8(dst, result, 3);
dst += pitch;
vst4_lane_u8(dst, result, 4);
dst += pitch;
vst4_lane_u8(dst, result, 5);
dst += pitch;
vst4_lane_u8(dst, result, 6);
dst += pitch;
vst4_lane_u8(dst, result, 7);
#endif // VPX_INCOMPATIBLE_GCC
}
void vp8_loop_filter_vertical_edge_y_neon(
unsigned char *src,
int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh) {
unsigned char *s, *d;
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
uint32x4x2_t q2tmp0, q2tmp1, q2tmp2, q2tmp3;
uint16x8x2_t q2tmp4, q2tmp5, q2tmp6, q2tmp7;
uint8x16x2_t q2tmp8, q2tmp9, q2tmp10, q2tmp11;
uint8x8x4_t q4ResultH, q4ResultL;
void vp8_loop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh) {
unsigned char *s, *d;
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
uint32x4x2_t q2tmp0, q2tmp1, q2tmp2, q2tmp3;
uint16x8x2_t q2tmp4, q2tmp5, q2tmp6, q2tmp7;
uint8x16x2_t q2tmp8, q2tmp9, q2tmp10, q2tmp11;
uint8x8x4_t q4ResultH, q4ResultL;
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
s = src - 4;
d6 = vld1_u8(s);
s += pitch;
d8 = vld1_u8(s);
s += pitch;
d10 = vld1_u8(s);
s += pitch;
d12 = vld1_u8(s);
s += pitch;
d14 = vld1_u8(s);
s += pitch;
d16 = vld1_u8(s);
s += pitch;
d18 = vld1_u8(s);
s += pitch;
d20 = vld1_u8(s);
s += pitch;
d7 = vld1_u8(s);
s += pitch;
d9 = vld1_u8(s);
s += pitch;
d11 = vld1_u8(s);
s += pitch;
d13 = vld1_u8(s);
s += pitch;
d15 = vld1_u8(s);
s += pitch;
d17 = vld1_u8(s);
s += pitch;
d19 = vld1_u8(s);
s += pitch;
d21 = vld1_u8(s);
s = src - 4;
d6 = vld1_u8(s);
s += pitch;
d8 = vld1_u8(s);
s += pitch;
d10 = vld1_u8(s);
s += pitch;
d12 = vld1_u8(s);
s += pitch;
d14 = vld1_u8(s);
s += pitch;
d16 = vld1_u8(s);
s += pitch;
d18 = vld1_u8(s);
s += pitch;
d20 = vld1_u8(s);
s += pitch;
d7 = vld1_u8(s);
s += pitch;
d9 = vld1_u8(s);
s += pitch;
d11 = vld1_u8(s);
s += pitch;
d13 = vld1_u8(s);
s += pitch;
d15 = vld1_u8(s);
s += pitch;
d17 = vld1_u8(s);
s += pitch;
d19 = vld1_u8(s);
s += pitch;
d21 = vld1_u8(s);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
q2tmp0 = vtrnq_u32(vreinterpretq_u32_u8(q3), vreinterpretq_u32_u8(q7));
q2tmp1 = vtrnq_u32(vreinterpretq_u32_u8(q4), vreinterpretq_u32_u8(q8));
q2tmp2 = vtrnq_u32(vreinterpretq_u32_u8(q5), vreinterpretq_u32_u8(q9));
q2tmp3 = vtrnq_u32(vreinterpretq_u32_u8(q6), vreinterpretq_u32_u8(q10));
q2tmp0 = vtrnq_u32(vreinterpretq_u32_u8(q3), vreinterpretq_u32_u8(q7));
q2tmp1 = vtrnq_u32(vreinterpretq_u32_u8(q4), vreinterpretq_u32_u8(q8));
q2tmp2 = vtrnq_u32(vreinterpretq_u32_u8(q5), vreinterpretq_u32_u8(q9));
q2tmp3 = vtrnq_u32(vreinterpretq_u32_u8(q6), vreinterpretq_u32_u8(q10));
q2tmp4 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[0]),
vreinterpretq_u16_u32(q2tmp2.val[0]));
q2tmp5 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[0]),
vreinterpretq_u16_u32(q2tmp3.val[0]));
q2tmp6 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[1]),
vreinterpretq_u16_u32(q2tmp2.val[1]));
q2tmp7 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[1]),
vreinterpretq_u16_u32(q2tmp3.val[1]));
q2tmp4 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[0]),
vreinterpretq_u16_u32(q2tmp2.val[0]));
q2tmp5 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[0]),
vreinterpretq_u16_u32(q2tmp3.val[0]));
q2tmp6 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[1]),
vreinterpretq_u16_u32(q2tmp2.val[1]));
q2tmp7 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[1]),
vreinterpretq_u16_u32(q2tmp3.val[1]));
q2tmp8 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[0]),
vreinterpretq_u8_u16(q2tmp5.val[0]));
q2tmp9 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[1]),
vreinterpretq_u8_u16(q2tmp5.val[1]));
q2tmp10 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[0]),
vreinterpretq_u8_u16(q2tmp7.val[0]));
q2tmp11 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[1]),
vreinterpretq_u8_u16(q2tmp7.val[1]));
q2tmp8 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[0]),
vreinterpretq_u8_u16(q2tmp5.val[0]));
q2tmp9 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[1]),
vreinterpretq_u8_u16(q2tmp5.val[1]));
q2tmp10 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[0]),
vreinterpretq_u8_u16(q2tmp7.val[0]));
q2tmp11 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[1]),
vreinterpretq_u8_u16(q2tmp7.val[1]));
q3 = q2tmp8.val[0];
q4 = q2tmp8.val[1];
q5 = q2tmp9.val[0];
q6 = q2tmp9.val[1];
q7 = q2tmp10.val[0];
q8 = q2tmp10.val[1];
q9 = q2tmp11.val[0];
q10 = q2tmp11.val[1];
q3 = q2tmp8.val[0];
q4 = q2tmp8.val[1];
q5 = q2tmp9.val[0];
q6 = q2tmp9.val[1];
q7 = q2tmp10.val[0];
q8 = q2tmp10.val[1];
q9 = q2tmp11.val[0];
q10 = q2tmp11.val[1];
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4,
q5, q6, q7, q8, q9, q10,
&q5, &q6, &q7, &q8);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4, q5, q6, q7, q8, q9,
q10, &q5, &q6, &q7, &q8);
q4ResultL.val[0] = vget_low_u8(q5); // d10
q4ResultL.val[1] = vget_low_u8(q6); // d12
q4ResultL.val[2] = vget_low_u8(q7); // d14
q4ResultL.val[3] = vget_low_u8(q8); // d16
q4ResultH.val[0] = vget_high_u8(q5); // d11
q4ResultH.val[1] = vget_high_u8(q6); // d13
q4ResultH.val[2] = vget_high_u8(q7); // d15
q4ResultH.val[3] = vget_high_u8(q8); // d17
q4ResultL.val[0] = vget_low_u8(q5); // d10
q4ResultL.val[1] = vget_low_u8(q6); // d12
q4ResultL.val[2] = vget_low_u8(q7); // d14
q4ResultL.val[3] = vget_low_u8(q8); // d16
q4ResultH.val[0] = vget_high_u8(q5); // d11
q4ResultH.val[1] = vget_high_u8(q6); // d13
q4ResultH.val[2] = vget_high_u8(q7); // d15
q4ResultH.val[3] = vget_high_u8(q8); // d17
d = src - 2;
write_4x8(d, pitch, q4ResultL);
d += pitch * 8;
write_4x8(d, pitch, q4ResultH);
d = src - 2;
write_4x8(d, pitch, q4ResultL);
d += pitch * 8;
write_4x8(d, pitch, q4ResultH);
}
void vp8_loop_filter_vertical_edge_uv_neon(
unsigned char *u,
int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh,
unsigned char *v) {
unsigned char *us, *ud;
unsigned char *vs, *vd;
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
uint32x4x2_t q2tmp0, q2tmp1, q2tmp2, q2tmp3;
uint16x8x2_t q2tmp4, q2tmp5, q2tmp6, q2tmp7;
uint8x16x2_t q2tmp8, q2tmp9, q2tmp10, q2tmp11;
uint8x8x4_t q4ResultH, q4ResultL;
void vp8_loop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch,
unsigned char blimit,
unsigned char limit,
unsigned char thresh,
unsigned char *v) {
unsigned char *us, *ud;
unsigned char *vs, *vd;
uint8x16_t qblimit, qlimit, qthresh, q3, q4;
uint8x16_t q5, q6, q7, q8, q9, q10;
uint8x8_t d6, d7, d8, d9, d10, d11, d12, d13, d14;
uint8x8_t d15, d16, d17, d18, d19, d20, d21;
uint32x4x2_t q2tmp0, q2tmp1, q2tmp2, q2tmp3;
uint16x8x2_t q2tmp4, q2tmp5, q2tmp6, q2tmp7;
uint8x16x2_t q2tmp8, q2tmp9, q2tmp10, q2tmp11;
uint8x8x4_t q4ResultH, q4ResultL;
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
qblimit = vdupq_n_u8(blimit);
qlimit = vdupq_n_u8(limit);
qthresh = vdupq_n_u8(thresh);
us = u - 4;
d6 = vld1_u8(us);
us += pitch;
d8 = vld1_u8(us);
us += pitch;
d10 = vld1_u8(us);
us += pitch;
d12 = vld1_u8(us);
us += pitch;
d14 = vld1_u8(us);
us += pitch;
d16 = vld1_u8(us);
us += pitch;
d18 = vld1_u8(us);
us += pitch;
d20 = vld1_u8(us);
us = u - 4;
d6 = vld1_u8(us);
us += pitch;
d8 = vld1_u8(us);
us += pitch;
d10 = vld1_u8(us);
us += pitch;
d12 = vld1_u8(us);
us += pitch;
d14 = vld1_u8(us);
us += pitch;
d16 = vld1_u8(us);
us += pitch;
d18 = vld1_u8(us);
us += pitch;
d20 = vld1_u8(us);
vs = v - 4;
d7 = vld1_u8(vs);
vs += pitch;
d9 = vld1_u8(vs);
vs += pitch;
d11 = vld1_u8(vs);
vs += pitch;
d13 = vld1_u8(vs);
vs += pitch;
d15 = vld1_u8(vs);
vs += pitch;
d17 = vld1_u8(vs);
vs += pitch;
d19 = vld1_u8(vs);
vs += pitch;
d21 = vld1_u8(vs);
vs = v - 4;
d7 = vld1_u8(vs);
vs += pitch;
d9 = vld1_u8(vs);
vs += pitch;
d11 = vld1_u8(vs);
vs += pitch;
d13 = vld1_u8(vs);
vs += pitch;
d15 = vld1_u8(vs);
vs += pitch;
d17 = vld1_u8(vs);
vs += pitch;
d19 = vld1_u8(vs);
vs += pitch;
d21 = vld1_u8(vs);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
q3 = vcombine_u8(d6, d7);
q4 = vcombine_u8(d8, d9);
q5 = vcombine_u8(d10, d11);
q6 = vcombine_u8(d12, d13);
q7 = vcombine_u8(d14, d15);
q8 = vcombine_u8(d16, d17);
q9 = vcombine_u8(d18, d19);
q10 = vcombine_u8(d20, d21);
q2tmp0 = vtrnq_u32(vreinterpretq_u32_u8(q3), vreinterpretq_u32_u8(q7));
q2tmp1 = vtrnq_u32(vreinterpretq_u32_u8(q4), vreinterpretq_u32_u8(q8));
q2tmp2 = vtrnq_u32(vreinterpretq_u32_u8(q5), vreinterpretq_u32_u8(q9));
q2tmp3 = vtrnq_u32(vreinterpretq_u32_u8(q6), vreinterpretq_u32_u8(q10));
q2tmp0 = vtrnq_u32(vreinterpretq_u32_u8(q3), vreinterpretq_u32_u8(q7));
q2tmp1 = vtrnq_u32(vreinterpretq_u32_u8(q4), vreinterpretq_u32_u8(q8));
q2tmp2 = vtrnq_u32(vreinterpretq_u32_u8(q5), vreinterpretq_u32_u8(q9));
q2tmp3 = vtrnq_u32(vreinterpretq_u32_u8(q6), vreinterpretq_u32_u8(q10));
q2tmp4 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[0]),
vreinterpretq_u16_u32(q2tmp2.val[0]));
q2tmp5 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[0]),
vreinterpretq_u16_u32(q2tmp3.val[0]));
q2tmp6 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[1]),
vreinterpretq_u16_u32(q2tmp2.val[1]));
q2tmp7 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[1]),
vreinterpretq_u16_u32(q2tmp3.val[1]));
q2tmp4 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[0]),
vreinterpretq_u16_u32(q2tmp2.val[0]));
q2tmp5 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[0]),
vreinterpretq_u16_u32(q2tmp3.val[0]));
q2tmp6 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp0.val[1]),
vreinterpretq_u16_u32(q2tmp2.val[1]));
q2tmp7 = vtrnq_u16(vreinterpretq_u16_u32(q2tmp1.val[1]),
vreinterpretq_u16_u32(q2tmp3.val[1]));
q2tmp8 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[0]),
vreinterpretq_u8_u16(q2tmp5.val[0]));
q2tmp9 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[1]),
vreinterpretq_u8_u16(q2tmp5.val[1]));
q2tmp10 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[0]),
vreinterpretq_u8_u16(q2tmp7.val[0]));
q2tmp11 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[1]),
vreinterpretq_u8_u16(q2tmp7.val[1]));
q2tmp8 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[0]),
vreinterpretq_u8_u16(q2tmp5.val[0]));
q2tmp9 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp4.val[1]),
vreinterpretq_u8_u16(q2tmp5.val[1]));
q2tmp10 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[0]),
vreinterpretq_u8_u16(q2tmp7.val[0]));
q2tmp11 = vtrnq_u8(vreinterpretq_u8_u16(q2tmp6.val[1]),
vreinterpretq_u8_u16(q2tmp7.val[1]));
q3 = q2tmp8.val[0];
q4 = q2tmp8.val[1];
q5 = q2tmp9.val[0];
q6 = q2tmp9.val[1];
q7 = q2tmp10.val[0];
q8 = q2tmp10.val[1];
q9 = q2tmp11.val[0];
q10 = q2tmp11.val[1];
q3 = q2tmp8.val[0];
q4 = q2tmp8.val[1];
q5 = q2tmp9.val[0];
q6 = q2tmp9.val[1];
q7 = q2tmp10.val[0];
q8 = q2tmp10.val[1];
q9 = q2tmp11.val[0];
q10 = q2tmp11.val[1];
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4,
q5, q6, q7, q8, q9, q10,
&q5, &q6, &q7, &q8);
vp8_loop_filter_neon(qblimit, qlimit, qthresh, q3, q4, q5, q6, q7, q8, q9,
q10, &q5, &q6, &q7, &q8);
q4ResultL.val[0] = vget_low_u8(q5); // d10
q4ResultL.val[1] = vget_low_u8(q6); // d12
q4ResultL.val[2] = vget_low_u8(q7); // d14
q4ResultL.val[3] = vget_low_u8(q8); // d16
ud = u - 2;
write_4x8(ud, pitch, q4ResultL);
q4ResultL.val[0] = vget_low_u8(q5); // d10
q4ResultL.val[1] = vget_low_u8(q6); // d12
q4ResultL.val[2] = vget_low_u8(q7); // d14
q4ResultL.val[3] = vget_low_u8(q8); // d16
ud = u - 2;
write_4x8(ud, pitch, q4ResultL);
q4ResultH.val[0] = vget_high_u8(q5); // d11
q4ResultH.val[1] = vget_high_u8(q6); // d13
q4ResultH.val[2] = vget_high_u8(q7); // d15
q4ResultH.val[3] = vget_high_u8(q8); // d17
vd = v - 2;
write_4x8(vd, pitch, q4ResultH);
q4ResultH.val[0] = vget_high_u8(q5); // d11
q4ResultH.val[1] = vget_high_u8(q6); // d13
q4ResultH.val[2] = vget_high_u8(q7); // d15
q4ResultH.val[3] = vget_high_u8(q8); // d17
vd = v - 2;
write_4x8(vd, pitch, q4ResultH);
}

11
vp8/common/blockd.c
View File

@ -8,15 +8,12 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "blockd.h"
#include "vpx_mem/vpx_mem.h"
const unsigned char vp8_block2left[25] =
{
0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8
const unsigned char vp8_block2left[25] = {
0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8
};
const unsigned char vp8_block2above[25] =
{
0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 4, 5, 6, 7, 6, 7, 8
const unsigned char vp8_block2above[25] = {
0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 4, 5, 6, 7, 6, 7, 8
};

359
vp8/common/blockd.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_BLOCKD_H_
#define VP8_COMMON_BLOCKD_H_
@ -28,280 +27,266 @@ extern "C" {
#define DCPREDSIMTHRESH 0
#define DCPREDCNTTHRESH 3
#define MB_FEATURE_TREE_PROBS 3
#define MAX_MB_SEGMENTS 4
#define MB_FEATURE_TREE_PROBS 3
#define MAX_MB_SEGMENTS 4
#define MAX_REF_LF_DELTAS 4
#define MAX_MODE_LF_DELTAS 4
#define MAX_REF_LF_DELTAS 4
#define MAX_MODE_LF_DELTAS 4
/* Segment Feature Masks */
#define SEGMENT_DELTADATA 0
#define SEGMENT_ABSDATA 1
#define SEGMENT_DELTADATA 0
#define SEGMENT_ABSDATA 1
typedef struct
{
int r, c;
} POS;
#define PLANE_TYPE_Y_NO_DC 0
#define PLANE_TYPE_Y2 1
#define PLANE_TYPE_UV 2
#define PLANE_TYPE_Y_WITH_DC 3
typedef struct { int r, c; } POS;
#define PLANE_TYPE_Y_NO_DC 0
#define PLANE_TYPE_Y2 1
#define PLANE_TYPE_UV 2
#define PLANE_TYPE_Y_WITH_DC 3
typedef char ENTROPY_CONTEXT;
typedef struct
{
ENTROPY_CONTEXT y1[4];
ENTROPY_CONTEXT u[2];
ENTROPY_CONTEXT v[2];
ENTROPY_CONTEXT y2;
typedef struct {
ENTROPY_CONTEXT y1[4];
ENTROPY_CONTEXT u[2];
ENTROPY_CONTEXT v[2];
ENTROPY_CONTEXT y2;
} ENTROPY_CONTEXT_PLANES;
extern const unsigned char vp8_block2left[25];
extern const unsigned char vp8_block2above[25];
#define VP8_COMBINEENTROPYCONTEXTS( Dest, A, B) \
Dest = (A)+(B);
#define VP8_COMBINEENTROPYCONTEXTS(Dest, A, B) Dest = (A) + (B);
typedef enum { KEY_FRAME = 0, INTER_FRAME = 1 } FRAME_TYPE;
typedef enum
{
KEY_FRAME = 0,
INTER_FRAME = 1
} FRAME_TYPE;
typedef enum {
DC_PRED, /* average of above and left pixels */
V_PRED, /* vertical prediction */
H_PRED, /* horizontal prediction */
TM_PRED, /* Truemotion prediction */
B_PRED, /* block based prediction, each block has its own prediction mode */
typedef enum
{
DC_PRED, /* average of above and left pixels */
V_PRED, /* vertical prediction */
H_PRED, /* horizontal prediction */
TM_PRED, /* Truemotion prediction */
B_PRED, /* block based prediction, each block has its own prediction mode */
NEARESTMV,
NEARMV,
ZEROMV,
NEWMV,
SPLITMV,
NEARESTMV,
NEARMV,
ZEROMV,
NEWMV,
SPLITMV,
MB_MODE_COUNT
MB_MODE_COUNT
} MB_PREDICTION_MODE;
/* Macroblock level features */
typedef enum
{
MB_LVL_ALT_Q = 0, /* Use alternate Quantizer .... */
MB_LVL_ALT_LF = 1, /* Use alternate loop filter value... */
MB_LVL_MAX = 2 /* Number of MB level features supported */
typedef enum {
MB_LVL_ALT_Q = 0, /* Use alternate Quantizer .... */
MB_LVL_ALT_LF = 1, /* Use alternate loop filter value... */
MB_LVL_MAX = 2 /* Number of MB level features supported */
} MB_LVL_FEATURES;
/* Segment Feature Masks */
#define SEGMENT_ALTQ 0x01
#define SEGMENT_ALT_LF 0x02
#define SEGMENT_ALTQ 0x01
#define SEGMENT_ALT_LF 0x02
#define VP8_YMODES (B_PRED + 1)
#define VP8_YMODES (B_PRED + 1)
#define VP8_UV_MODES (TM_PRED + 1)
#define VP8_MVREFS (1 + SPLITMV - NEARESTMV)
typedef enum
{
B_DC_PRED, /* average of above and left pixels */
B_TM_PRED,
typedef enum {
B_DC_PRED, /* average of above and left pixels */
B_TM_PRED,
B_VE_PRED, /* vertical prediction */
B_HE_PRED, /* horizontal prediction */
B_VE_PRED, /* vertical prediction */
B_HE_PRED, /* horizontal prediction */
B_LD_PRED,
B_RD_PRED,
B_LD_PRED,
B_RD_PRED,
B_VR_PRED,
B_VL_PRED,
B_HD_PRED,
B_HU_PRED,
B_VR_PRED,
B_VL_PRED,
B_HD_PRED,
B_HU_PRED,
LEFT4X4,
ABOVE4X4,
ZERO4X4,
NEW4X4,
LEFT4X4,
ABOVE4X4,
ZERO4X4,
NEW4X4,
B_MODE_COUNT
B_MODE_COUNT
} B_PREDICTION_MODE;
#define VP8_BINTRAMODES (B_HU_PRED + 1) /* 10 */
#define VP8_BINTRAMODES (B_HU_PRED + 1) /* 10 */
#define VP8_SUBMVREFS (1 + NEW4X4 - LEFT4X4)
/* For keyframes, intra block modes are predicted by the (already decoded)
modes for the Y blocks to the left and above us; for interframes, there
is a single probability table. */
union b_mode_info
{
B_PREDICTION_MODE as_mode;
int_mv mv;
union b_mode_info {
B_PREDICTION_MODE as_mode;
int_mv mv;
};
typedef enum
{
INTRA_FRAME = 0,
LAST_FRAME = 1,
GOLDEN_FRAME = 2,
ALTREF_FRAME = 3,
MAX_REF_FRAMES = 4
typedef enum {
INTRA_FRAME = 0,
LAST_FRAME = 1,
GOLDEN_FRAME = 2,
ALTREF_FRAME = 3,
MAX_REF_FRAMES = 4
} MV_REFERENCE_FRAME;
typedef struct
{
uint8_t mode, uv_mode;
uint8_t ref_frame;
uint8_t is_4x4;
int_mv mv;
typedef struct {
uint8_t mode, uv_mode;
uint8_t ref_frame;
uint8_t is_4x4;
int_mv mv;
uint8_t partitioning;
uint8_t mb_skip_coeff; /* does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens */
uint8_t need_to_clamp_mvs;
uint8_t segment_id; /* Which set of segmentation parameters should be used for this MB */
uint8_t partitioning;
/* does this mb has coefficients at all, 1=no coefficients, 0=need decode
tokens */
uint8_t mb_skip_coeff;
uint8_t need_to_clamp_mvs;
/* Which set of segmentation parameters should be used for this MB */
uint8_t segment_id;
} MB_MODE_INFO;
typedef struct modeinfo
{
MB_MODE_INFO mbmi;
union b_mode_info bmi[16];
typedef struct modeinfo {
MB_MODE_INFO mbmi;
union b_mode_info bmi[16];
} MODE_INFO;
#if CONFIG_MULTI_RES_ENCODING
/* The mb-level information needed to be stored for higher-resolution encoder */
typedef struct
{
MB_PREDICTION_MODE mode;
MV_REFERENCE_FRAME ref_frame;
int_mv mv;
int dissim; /* dissimilarity level of the macroblock */
typedef struct {
MB_PREDICTION_MODE mode;
MV_REFERENCE_FRAME ref_frame;
int_mv mv;
int dissim; /* dissimilarity level of the macroblock */
} LOWER_RES_MB_INFO;
/* The frame-level information needed to be stored for higher-resolution
* encoder */
typedef struct
{
FRAME_TYPE frame_type;
int is_frame_dropped;
// The frame rate for the lowest resolution.
double low_res_framerate;
/* The frame number of each reference frames */
unsigned int low_res_ref_frames[MAX_REF_FRAMES];
// The video frame counter value for the key frame, for lowest resolution.
unsigned int key_frame_counter_value;
LOWER_RES_MB_INFO *mb_info;
typedef struct {
FRAME_TYPE frame_type;
int is_frame_dropped;
// The frame rate for the lowest resolution.
double low_res_framerate;
/* The frame number of each reference frames */
unsigned int low_res_ref_frames[MAX_REF_FRAMES];
// The video frame counter value for the key frame, for lowest resolution.
unsigned int key_frame_counter_value;
LOWER_RES_MB_INFO *mb_info;
} LOWER_RES_FRAME_INFO;
#endif
typedef struct blockd
{
short *qcoeff;
short *dqcoeff;
unsigned char *predictor;
short *dequant;
typedef struct blockd {
short *qcoeff;
short *dqcoeff;
unsigned char *predictor;
short *dequant;
int offset;
char *eob;
int offset;
char *eob;
union b_mode_info bmi;
union b_mode_info bmi;
} BLOCKD;
typedef void (*vp8_subpix_fn_t)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
typedef void (*vp8_subpix_fn_t)(unsigned char *src, int src_pitch, int xofst,
int yofst, unsigned char *dst, int dst_pitch);
typedef struct macroblockd
{
DECLARE_ALIGNED(16, unsigned char, predictor[384]);
DECLARE_ALIGNED(16, short, qcoeff[400]);
DECLARE_ALIGNED(16, short, dqcoeff[400]);
DECLARE_ALIGNED(16, char, eobs[25]);
typedef struct macroblockd {
DECLARE_ALIGNED(16, unsigned char, predictor[384]);
DECLARE_ALIGNED(16, short, qcoeff[400]);
DECLARE_ALIGNED(16, short, dqcoeff[400]);
DECLARE_ALIGNED(16, char, eobs[25]);
DECLARE_ALIGNED(16, short, dequant_y1[16]);
DECLARE_ALIGNED(16, short, dequant_y1_dc[16]);
DECLARE_ALIGNED(16, short, dequant_y2[16]);
DECLARE_ALIGNED(16, short, dequant_uv[16]);
DECLARE_ALIGNED(16, short, dequant_y1[16]);
DECLARE_ALIGNED(16, short, dequant_y1_dc[16]);
DECLARE_ALIGNED(16, short, dequant_y2[16]);
DECLARE_ALIGNED(16, short, dequant_uv[16]);
/* 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. */
BLOCKD block[25];
int fullpixel_mask;
/* 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. */
BLOCKD block[25];
int fullpixel_mask;
YV12_BUFFER_CONFIG pre; /* Filtered copy of previous frame reconstruction */
YV12_BUFFER_CONFIG dst;
YV12_BUFFER_CONFIG pre; /* Filtered copy of previous frame reconstruction */
YV12_BUFFER_CONFIG dst;
MODE_INFO *mode_info_context;
int mode_info_stride;
MODE_INFO *mode_info_context;
int mode_info_stride;
FRAME_TYPE frame_type;
FRAME_TYPE frame_type;
int up_available;
int left_available;
int up_available;
int left_available;
unsigned char *recon_above[3];
unsigned char *recon_left[3];
int recon_left_stride[2];
unsigned char *recon_above[3];
unsigned char *recon_left[3];
int recon_left_stride[2];
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context;
ENTROPY_CONTEXT_PLANES *left_context;
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context;
ENTROPY_CONTEXT_PLANES *left_context;
/* 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active. */
unsigned char segmentation_enabled;
/* 0 indicates segmentation at MB level is not enabled. Otherwise the
* individual bits indicate which features are active. */
unsigned char segmentation_enabled;
/* 0 (do not update) 1 (update) the macroblock segmentation map. */
unsigned char update_mb_segmentation_map;
/* 0 (do not update) 1 (update) the macroblock segmentation map. */
unsigned char update_mb_segmentation_map;
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char update_mb_segmentation_data;
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char update_mb_segmentation_data;
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char mb_segement_abs_delta;
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char mb_segement_abs_delta;
/* Per frame flags that define which MB level features (such as quantizer or loop filter level) */
/* are enabled and when enabled the proabilities used to decode the per MB flags in MB_MODE_INFO */
vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS]; /* Probability Tree used to code Segment number */
/* Per frame flags that define which MB level features (such as quantizer or
* loop filter level) */
/* are enabled and when enabled the proabilities used to decode the per MB
* flags in MB_MODE_INFO */
/* Probability Tree used to code Segment number */
vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS];
/* Segment parameters */
signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; /* Segment parameters */
/* mode_based Loop filter adjustment */
unsigned char mode_ref_lf_delta_enabled;
unsigned char mode_ref_lf_delta_update;
/* mode_based Loop filter adjustment */
unsigned char mode_ref_lf_delta_enabled;
unsigned char mode_ref_lf_delta_update;
/* Delta values have the range +/- MAX_LOOP_FILTER */
signed char
last_ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
/* 0 = BPRED, ZERO_MV, MV, SPLIT */
signed char last_mode_lf_deltas[MAX_MODE_LF_DELTAS];
signed char
mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
/* Delta values have the range +/- MAX_LOOP_FILTER */
signed char last_ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
signed char last_mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
/* Distance of MB away from frame edges */
int mb_to_left_edge;
int mb_to_right_edge;
int mb_to_top_edge;
int mb_to_bottom_edge;
/* Distance of MB away from frame edges */
int mb_to_left_edge;
int mb_to_right_edge;
int mb_to_top_edge;
int mb_to_bottom_edge;
vp8_subpix_fn_t subpixel_predict;
vp8_subpix_fn_t subpixel_predict8x4;
vp8_subpix_fn_t subpixel_predict8x8;
vp8_subpix_fn_t subpixel_predict16x16;
void *current_bc;
vp8_subpix_fn_t subpixel_predict;
vp8_subpix_fn_t subpixel_predict8x4;
vp8_subpix_fn_t subpixel_predict8x8;
vp8_subpix_fn_t subpixel_predict16x16;
void *current_bc;
int corrupted;
int corrupted;
#if ARCH_X86 || ARCH_X86_64
/* This is an intermediate buffer currently used in sub-pixel motion search
* to keep a copy of the reference area. This buffer can be used for other
* purpose.
*/
DECLARE_ALIGNED(32, unsigned char, y_buf[22*32]);
/* This is an intermediate buffer currently used in sub-pixel motion search
* to keep a copy of the reference area. This buffer can be used for other
* purpose.
*/
DECLARE_ALIGNED(32, unsigned char, y_buf[22 * 32]);
#endif
} MACROBLOCKD;
extern void vp8_build_block_doffsets(MACROBLOCKD *x);
extern void vp8_setup_block_dptrs(MACROBLOCKD *x);

342
vp8/common/coefupdateprobs.h
View File

@ -18,177 +18,177 @@ extern "C" {
/* Update probabilities for the nodes in the token entropy tree.
Generated file included by entropy.c */
const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES] =
{
{
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
{249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
{234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255, },
{250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255, },
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
},
{
{
{217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255, },
{234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255, },
},
{
{255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
},
{
{
{186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255, },
{234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255, },
{251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255, },
},
{
{255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
},
{
{
{248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255, },
{248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255, },
},
{
{255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255, },
{248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, },
{250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
{
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, },
},
},
};
const vp8_prob vp8_coef_update_probs
[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][ENTROPY_NODES] = {
{
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
},
{
{
{ 217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255 },
{ 234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255 },
},
{
{ 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
},
{
{
{ 186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255 },
},
{
{ 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
},
{
{
{ 248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255 },
{ 248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
{
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
{ 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
},
},
};
#ifdef __cplusplus
} // extern "C"

24
vp8/common/common.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_COMMON_H_
#define VP8_COMMON_COMMON_H_
@ -24,22 +23,23 @@ extern "C" {
/* Only need this for fixed-size arrays, for structs just assign. */
#define vp8_copy( Dest, Src) { \
assert( sizeof( Dest) == sizeof( Src)); \
memcpy( Dest, Src, sizeof( Src)); \
}
#define vp8_copy(Dest, Src) \
{ \
assert(sizeof(Dest) == sizeof(Src)); \
memcpy(Dest, Src, sizeof(Src)); \
}
/* Use this for variably-sized arrays. */
#define vp8_copy_array( Dest, Src, N) { \
assert( sizeof( *Dest) == sizeof( *Src)); \
memcpy( Dest, Src, N * sizeof( *Src)); \
}
#define vp8_copy_array(Dest, Src, N) \
{ \
assert(sizeof(*Dest) == sizeof(*Src)); \
memcpy(Dest, Src, N * sizeof(*Src)); \
}
#define vp8_zero( Dest) memset( &Dest, 0, sizeof( Dest));
#define vp8_zero_array( Dest, N) memset( Dest, 0, N * sizeof( *Dest));
#define vp8_zero(Dest) memset(&Dest, 0, sizeof(Dest));
#define vp8_zero_array(Dest, N) memset(Dest, 0, N * sizeof(*Dest));
#ifdef __cplusplus
} // extern "C"

1
vp8/common/context.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "entropy.h"
/* *** GENERATED FILE: DO NOT EDIT *** */

19
vp8/common/copy_c.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include <string.h>
#include "./vp8_rtcd.h"
@ -16,17 +15,13 @@
/* Copy 2 macroblocks to a buffer */
void vp8_copy32xn_c(const unsigned char *src_ptr, int src_stride,
unsigned char *dst_ptr, int dst_stride,
int height)
{
int r;
unsigned char *dst_ptr, int dst_stride, int height) {
int r;
for (r = 0; r < height; r++)
{
memcpy(dst_ptr, src_ptr, 32);
for (r = 0; r < height; r++) {
memcpy(dst_ptr, src_ptr, 32);
src_ptr += src_stride;
dst_ptr += dst_stride;
}
src_ptr += src_stride;
dst_ptr += dst_stride;
}
}

182
vp8/common/debugmodes.c
View File

@ -8,148 +8,128 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include <stdio.h>
#include "blockd.h"
void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols,
int frame) {
int mb_row;
int mb_col;
int mb_index = 0;
FILE *mvs = fopen("mvs.stt", "a");
void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int frame)
{
/* print out the macroblock Y modes */
mb_index = 0;
fprintf(mvs, "Mb Modes for Frame %d\n", frame);
int mb_row;
int mb_col;
int mb_index = 0;
FILE *mvs = fopen("mvs.stt", "a");
for (mb_row = 0; mb_row < rows; mb_row++) {
for (mb_col = 0; mb_col < cols; mb_col++) {
fprintf(mvs, "%2d ", mi[mb_index].mbmi.mode);
/* print out the macroblock Y modes */
mb_index = 0;
fprintf(mvs, "Mb Modes for Frame %d\n", frame);
for (mb_row = 0; mb_row < rows; mb_row++)
{
for (mb_col = 0; mb_col < cols; mb_col++)
{
fprintf(mvs, "%2d ", mi[mb_index].mbmi.mode);
mb_index++;
}
fprintf(mvs, "\n");
mb_index++;
mb_index++;
}
fprintf(mvs, "\n");
mb_index++;
}
mb_index = 0;
fprintf(mvs, "Mb mv ref for Frame %d\n", frame);
fprintf(mvs, "\n");
for (mb_row = 0; mb_row < rows; mb_row++)
{
for (mb_col = 0; mb_col < cols; mb_col++)
{
mb_index = 0;
fprintf(mvs, "Mb mv ref for Frame %d\n", frame);
fprintf(mvs, "%2d ", mi[mb_index].mbmi.ref_frame);
for (mb_row = 0; mb_row < rows; mb_row++) {
for (mb_col = 0; mb_col < cols; mb_col++) {
fprintf(mvs, "%2d ", mi[mb_index].mbmi.ref_frame);
mb_index++;
}
fprintf(mvs, "\n");
mb_index++;
mb_index++;
}
fprintf(mvs, "\n");
mb_index++;
}
/* print out the macroblock UV modes */
mb_index = 0;
fprintf(mvs, "UV Modes for Frame %d\n", frame);
fprintf(mvs, "\n");
for (mb_row = 0; mb_row < rows; mb_row++)
{
for (mb_col = 0; mb_col < cols; mb_col++)
{
/* print out the macroblock UV modes */
mb_index = 0;
fprintf(mvs, "UV Modes for Frame %d\n", frame);
fprintf(mvs, "%2d ", mi[mb_index].mbmi.uv_mode);
for (mb_row = 0; mb_row < rows; mb_row++) {
for (mb_col = 0; mb_col < cols; mb_col++) {
fprintf(mvs, "%2d ", mi[mb_index].mbmi.uv_mode);
mb_index++;
}
mb_index++;
fprintf(mvs, "\n");
mb_index++;
}
mb_index++;
fprintf(mvs, "\n");
}
/* print out the block modes */
fprintf(mvs, "Mbs for Frame %d\n", frame);
{
int b_row;
fprintf(mvs, "\n");
for (b_row = 0; b_row < 4 * rows; b_row++)
{
int b_col;
int bindex;
/* print out the block modes */
fprintf(mvs, "Mbs for Frame %d\n", frame);
{
int b_row;
for (b_col = 0; b_col < 4 * cols; b_col++)
{
mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
bindex = (b_row & 3) * 4 + (b_col & 3);
for (b_row = 0; b_row < 4 * rows; b_row++) {
int b_col;
int bindex;
if (mi[mb_index].mbmi.mode == B_PRED)
fprintf(mvs, "%2d ", mi[mb_index].bmi[bindex].as_mode);
else
fprintf(mvs, "xx ");
for (b_col = 0; b_col < 4 * cols; b_col++) {
mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
bindex = (b_row & 3) * 4 + (b_col & 3);
}
if (mi[mb_index].mbmi.mode == B_PRED)
fprintf(mvs, "%2d ", mi[mb_index].bmi[bindex].as_mode);
else
fprintf(mvs, "xx ");
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
/* print out the macroblock mvs */
mb_index = 0;
fprintf(mvs, "MVs for Frame %d\n", frame);
/* print out the macroblock mvs */
mb_index = 0;
fprintf(mvs, "MVs for Frame %d\n", frame);
for (mb_row = 0; mb_row < rows; mb_row++)
{
for (mb_col = 0; mb_col < cols; mb_col++)
{
fprintf(mvs, "%5d:%-5d", mi[mb_index].mbmi.mv.as_mv.row / 2, mi[mb_index].mbmi.mv.as_mv.col / 2);
for (mb_row = 0; mb_row < rows; mb_row++) {
for (mb_col = 0; mb_col < cols; mb_col++) {
fprintf(mvs, "%5d:%-5d", mi[mb_index].mbmi.mv.as_mv.row / 2,
mi[mb_index].mbmi.mv.as_mv.col / 2);
mb_index++;
}
mb_index++;
fprintf(mvs, "\n");
mb_index++;
}
mb_index++;
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
/* print out the block modes */
fprintf(mvs, "MVs for Frame %d\n", frame);
{
int b_row;
/* print out the block modes */
fprintf(mvs, "MVs for Frame %d\n", frame);
{
int b_row;
for (b_row = 0; b_row < 4 * rows; b_row++)
{
int b_col;
int bindex;
for (b_row = 0; b_row < 4 * rows; b_row++) {
int b_col;
int bindex;
for (b_col = 0; b_col < 4 * cols; b_col++)
{
mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
bindex = (b_row & 3) * 4 + (b_col & 3);
fprintf(mvs, "%3d:%-3d ", mi[mb_index].bmi[bindex].mv.as_mv.row, mi[mb_index].bmi[bindex].mv.as_mv.col);
for (b_col = 0; b_col < 4 * cols; b_col++) {
mb_index = (b_row >> 2) * (cols + 1) + (b_col >> 2);
bindex = (b_row & 3) * 4 + (b_col & 3);
fprintf(mvs, "%3d:%-3d ", mi[mb_index].bmi[bindex].mv.as_mv.row,
mi[mb_index].bmi[bindex].mv.as_mv.col);
}
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
}
fprintf(mvs, "\n");
fclose(mvs);
fclose(mvs);
}

246
vp8/common/default_coef_probs.h
View File

@ -17,181 +17,141 @@ extern "C" {
/*Generated file, included by entropy.c*/
static const vp8_prob default_coef_probs [BLOCK_TYPES]
[COEF_BANDS]
[PREV_COEF_CONTEXTS]
[ENTROPY_NODES] =
{
{ /* Block Type ( 0 ) */
static const vp8_prob default_coef_probs
[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][ENTROPY_NODES] = {
{ /* Block Type ( 0 ) */
{ /* Coeff Band ( 0 )*/
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 1 )*/
{ 253, 136, 254, 255, 228, 219, 128, 128, 128, 128, 128 },
{ 189, 129, 242, 255, 227, 213, 255, 219, 128, 128, 128 },
{ 106, 126, 227, 252, 214, 209, 255, 255, 128, 128, 128 }
},
{ 253, 136, 254, 255, 228, 219, 128, 128, 128, 128, 128 },
{ 189, 129, 242, 255, 227, 213, 255, 219, 128, 128, 128 },
{ 106, 126, 227, 252, 214, 209, 255, 255, 128, 128, 128 } },
{ /* Coeff Band ( 2 )*/
{ 1, 98, 248, 255, 236, 226, 255, 255, 128, 128, 128 },
{ 181, 133, 238, 254, 221, 234, 255, 154, 128, 128, 128 },
{ 78, 134, 202, 247, 198, 180, 255, 219, 128, 128, 128 }
},
{ 1, 98, 248, 255, 236, 226, 255, 255, 128, 128, 128 },
{ 181, 133, 238, 254, 221, 234, 255, 154, 128, 128, 128 },
{ 78, 134, 202, 247, 198, 180, 255, 219, 128, 128, 128 } },
{ /* Coeff Band ( 3 )*/
{ 1, 185, 249, 255, 243, 255, 128, 128, 128, 128, 128 },
{ 184, 150, 247, 255, 236, 224, 128, 128, 128, 128, 128 },
{ 77, 110, 216, 255, 236, 230, 128, 128, 128, 128, 128 }
},
{ 1, 185, 249, 255, 243, 255, 128, 128, 128, 128, 128 },
{ 184, 150, 247, 255, 236, 224, 128, 128, 128, 128, 128 },
{ 77, 110, 216, 255, 236, 230, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 4 )*/
{ 1, 101, 251, 255, 241, 255, 128, 128, 128, 128, 128 },
{ 170, 139, 241, 252, 236, 209, 255, 255, 128, 128, 128 },
{ 37, 116, 196, 243, 228, 255, 255, 255, 128, 128, 128 }
},
{ 1, 101, 251, 255, 241, 255, 128, 128, 128, 128, 128 },
{ 170, 139, 241, 252, 236, 209, 255, 255, 128, 128, 128 },
{ 37, 116, 196, 243, 228, 255, 255, 255, 128, 128, 128 } },
{ /* Coeff Band ( 5 )*/
{ 1, 204, 254, 255, 245, 255, 128, 128, 128, 128, 128 },
{ 207, 160, 250, 255, 238, 128, 128, 128, 128, 128, 128 },
{ 102, 103, 231, 255, 211, 171, 128, 128, 128, 128, 128 }
},
{ 1, 204, 254, 255, 245, 255, 128, 128, 128, 128, 128 },
{ 207, 160, 250, 255, 238, 128, 128, 128, 128, 128, 128 },
{ 102, 103, 231, 255, 211, 171, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 6 )*/
{ 1, 152, 252, 255, 240, 255, 128, 128, 128, 128, 128 },
{ 177, 135, 243, 255, 234, 225, 128, 128, 128, 128, 128 },
{ 80, 129, 211, 255, 194, 224, 128, 128, 128, 128, 128 }
},
{ 1, 152, 252, 255, 240, 255, 128, 128, 128, 128, 128 },
{ 177, 135, 243, 255, 234, 225, 128, 128, 128, 128, 128 },
{ 80, 129, 211, 255, 194, 224, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 7 )*/
{ 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 246, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 255, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
}
},
{ /* Block Type ( 1 ) */
{ 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 246, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 255, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 } } },
{ /* Block Type ( 1 ) */
{ /* Coeff Band ( 0 )*/
{ 198, 35, 237, 223, 193, 187, 162, 160, 145, 155, 62 },
{ 131, 45, 198, 221, 172, 176, 220, 157, 252, 221, 1 },
{ 68, 47, 146, 208, 149, 167, 221, 162, 255, 223, 128 }
},
{ 198, 35, 237, 223, 193, 187, 162, 160, 145, 155, 62 },
{ 131, 45, 198, 221, 172, 176, 220, 157, 252, 221, 1 },
{ 68, 47, 146, 208, 149, 167, 221, 162, 255, 223, 128 } },
{ /* Coeff Band ( 1 )*/
{ 1, 149, 241, 255, 221, 224, 255, 255, 128, 128, 128 },
{ 184, 141, 234, 253, 222, 220, 255, 199, 128, 128, 128 },
{ 81, 99, 181, 242, 176, 190, 249, 202, 255, 255, 128 }
},
{ 1, 149, 241, 255, 221, 224, 255, 255, 128, 128, 128 },
{ 184, 141, 234, 253, 222, 220, 255, 199, 128, 128, 128 },
{ 81, 99, 181, 242, 176, 190, 249, 202, 255, 255, 128 } },
{ /* Coeff Band ( 2 )*/
{ 1, 129, 232, 253, 214, 197, 242, 196, 255, 255, 128 },
{ 99, 121, 210, 250, 201, 198, 255, 202, 128, 128, 128 },
{ 23, 91, 163, 242, 170, 187, 247, 210, 255, 255, 128 }
},
{ 1, 129, 232, 253, 214, 197, 242, 196, 255, 255, 128 },
{ 99, 121, 210, 250, 201, 198, 255, 202, 128, 128, 128 },
{ 23, 91, 163, 242, 170, 187, 247, 210, 255, 255, 128 } },
{ /* Coeff Band ( 3 )*/
{ 1, 200, 246, 255, 234, 255, 128, 128, 128, 128, 128 },
{ 109, 178, 241, 255, 231, 245, 255, 255, 128, 128, 128 },
{ 44, 130, 201, 253, 205, 192, 255, 255, 128, 128, 128 }
},
{ 1, 200, 246, 255, 234, 255, 128, 128, 128, 128, 128 },
{ 109, 178, 241, 255, 231, 245, 255, 255, 128, 128, 128 },
{ 44, 130, 201, 253, 205, 192, 255, 255, 128, 128, 128 } },
{ /* Coeff Band ( 4 )*/
{ 1, 132, 239, 251, 219, 209, 255, 165, 128, 128, 128 },
{ 94, 136, 225, 251, 218, 190, 255, 255, 128, 128, 128 },
{ 22, 100, 174, 245, 186, 161, 255, 199, 128, 128, 128 }
},
{ 1, 132, 239, 251, 219, 209, 255, 165, 128, 128, 128 },
{ 94, 136, 225, 251, 218, 190, 255, 255, 128, 128, 128 },
{ 22, 100, 174, 245, 186, 161, 255, 199, 128, 128, 128 } },
{ /* Coeff Band ( 5 )*/
{ 1, 182, 249, 255, 232, 235, 128, 128, 128, 128, 128 },
{ 124, 143, 241, 255, 227, 234, 128, 128, 128, 128, 128 },
{ 35, 77, 181, 251, 193, 211, 255, 205, 128, 128, 128 }
},
{ 1, 182, 249, 255, 232, 235, 128, 128, 128, 128, 128 },
{ 124, 143, 241, 255, 227, 234, 128, 128, 128, 128, 128 },
{ 35, 77, 181, 251, 193, 211, 255, 205, 128, 128, 128 } },
{ /* Coeff Band ( 6 )*/
{ 1, 157, 247, 255, 236, 231, 255, 255, 128, 128, 128 },
{ 121, 141, 235, 255, 225, 227, 255, 255, 128, 128, 128 },
{ 45, 99, 188, 251, 195, 217, 255, 224, 128, 128, 128 }
},
{ 1, 157, 247, 255, 236, 231, 255, 255, 128, 128, 128 },
{ 121, 141, 235, 255, 225, 227, 255, 255, 128, 128, 128 },
{ 45, 99, 188, 251, 195, 217, 255, 224, 128, 128, 128 } },
{ /* Coeff Band ( 7 )*/
{ 1, 1, 251, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 203, 1, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 137, 1, 177, 255, 224, 255, 128, 128, 128, 128, 128 }
}
},
{ /* Block Type ( 2 ) */
{ 1, 1, 251, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 203, 1, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 137, 1, 177, 255, 224, 255, 128, 128, 128, 128, 128 } } },
{ /* Block Type ( 2 ) */
{ /* Coeff Band ( 0 )*/
{ 253, 9, 248, 251, 207, 208, 255, 192, 128, 128, 128 },
{ 175, 13, 224, 243, 193, 185, 249, 198, 255, 255, 128 },
{ 73, 17, 171, 221, 161, 179, 236, 167, 255, 234, 128 }
},
{ 253, 9, 248, 251, 207, 208, 255, 192, 128, 128, 128 },
{ 175, 13, 224, 243, 193, 185, 249, 198, 255, 255, 128 },
{ 73, 17, 171, 221, 161, 179, 236, 167, 255, 234, 128 } },
{ /* Coeff Band ( 1 )*/
{ 1, 95, 247, 253, 212, 183, 255, 255, 128, 128, 128 },
{ 239, 90, 244, 250, 211, 209, 255, 255, 128, 128, 128 },
{ 155, 77, 195, 248, 188, 195, 255, 255, 128, 128, 128 }
},
{ 1, 95, 247, 253, 212, 183, 255, 255, 128, 128, 128 },
{ 239, 90, 244, 250, 211, 209, 255, 255, 128, 128, 128 },
{ 155, 77, 195, 248, 188, 195, 255, 255, 128, 128, 128 } },
{ /* Coeff Band ( 2 )*/
{ 1, 24, 239, 251, 218, 219, 255, 205, 128, 128, 128 },
{ 201, 51, 219, 255, 196, 186, 128, 128, 128, 128, 128 },
{ 69, 46, 190, 239, 201, 218, 255, 228, 128, 128, 128 }
},
{ 1, 24, 239, 251, 218, 219, 255, 205, 128, 128, 128 },
{ 201, 51, 219, 255, 196, 186, 128, 128, 128, 128, 128 },
{ 69, 46, 190, 239, 201, 218, 255, 228, 128, 128, 128 } },
{ /* Coeff Band ( 3 )*/
{ 1, 191, 251, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 223, 165, 249, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 141, 124, 248, 255, 255, 128, 128, 128, 128, 128, 128 }
},
{ 1, 191, 251, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 223, 165, 249, 255, 213, 255, 128, 128, 128, 128, 128 },
{ 141, 124, 248, 255, 255, 128, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 4 )*/
{ 1, 16, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 190, 36, 230, 255, 236, 255, 128, 128, 128, 128, 128 },
{ 149, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ 1, 16, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 190, 36, 230, 255, 236, 255, 128, 128, 128, 128, 128 },
{ 149, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 5 )*/
{ 1, 226, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 247, 192, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 240, 128, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ 1, 226, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 247, 192, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 240, 128, 255, 128, 128, 128, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 6 )*/
{ 1, 134, 252, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 213, 62, 250, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 55, 93, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
},
{ 1, 134, 252, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 213, 62, 250, 255, 255, 128, 128, 128, 128, 128, 128 },
{ 55, 93, 255, 128, 128, 128, 128, 128, 128, 128, 128 } },
{ /* Coeff Band ( 7 )*/
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
}
},
{ /* Block Type ( 3 ) */
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 } } },
{ /* Block Type ( 3 ) */
{ /* Coeff Band ( 0 )*/
{ 202, 24, 213, 235, 186, 191, 220, 160, 240, 175, 255 },
{ 126, 38, 182, 232, 169, 184, 228, 174, 255, 187, 128 },
{ 61, 46, 138, 219, 151, 178, 240, 170, 255, 216, 128 }
},
{ 202, 24, 213, 235, 186, 191, 220, 160, 240, 175, 255 },
{ 126, 38, 182, 232, 169, 184, 228, 174, 255, 187, 128 },
{ 61, 46, 138, 219, 151, 178, 240, 170, 255, 216, 128 } },
{ /* Coeff Band ( 1 )*/
{ 1, 112, 230, 250, 199, 191, 247, 159, 255, 255, 128 },
{ 166, 109, 228, 252, 211, 215, 255, 174, 128, 128, 128 },
{ 39, 77, 162, 232, 172, 180, 245, 178, 255, 255, 128 }
},
{ 1, 112, 230, 250, 199, 191, 247, 159, 255, 255, 128 },
{ 166, 109, 228, 252, 211, 215, 255, 174, 128, 128, 128 },
{ 39, 77, 162, 232, 172, 180, 245, 178, 255, 255, 128 } },
{ /* Coeff Band ( 2 )*/
{ 1, 52, 220, 246, 198, 199, 249, 220, 255, 255, 128 },
{ 124, 74, 191, 243, 183, 193, 250, 221, 255, 255, 128 },
{ 24, 71, 130, 219, 154, 170, 243, 182, 255, 255, 128 }
},
{ 1, 52, 220, 246, 198, 199, 249, 220, 255, 255, 128 },
{ 124, 74, 191, 243, 183, 193, 250, 221, 255, 255, 128 },
{ 24, 71, 130, 219, 154, 170, 243, 182, 255, 255, 128 } },
{ /* Coeff Band ( 3 )*/
{ 1, 182, 225, 249, 219, 240, 255, 224, 128, 128, 128 },
{ 149, 150, 226, 252, 216, 205, 255, 171, 128, 128, 128 },
{ 28, 108, 170, 242, 183, 194, 254, 223, 255, 255, 128 }
},
{ 1, 182, 225, 249, 219, 240, 255, 224, 128, 128, 128 },
{ 149, 150, 226, 252, 216, 205, 255, 171, 128, 128, 128 },
{ 28, 108, 170, 242, 183, 194, 254, 223, 255, 255, 128 } },
{ /* Coeff Band ( 4 )*/
{ 1, 81, 230, 252, 204, 203, 255, 192, 128, 128, 128 },
{ 123, 102, 209, 247, 188, 196, 255, 233, 128, 128, 128 },
{ 20, 95, 153, 243, 164, 173, 255, 203, 128, 128, 128 }
},
{ 1, 81, 230, 252, 204, 203, 255, 192, 128, 128, 128 },
{ 123, 102, 209, 247, 188, 196, 255, 233, 128, 128, 128 },
{ 20, 95, 153, 243, 164, 173, 255, 203, 128, 128, 128 } },
{ /* Coeff Band ( 5 )*/
{ 1, 222, 248, 255, 216, 213, 128, 128, 128, 128, 128 },
{ 168, 175, 246, 252, 235, 205, 255, 255, 128, 128, 128 },
{ 47, 116, 215, 255, 211, 212, 255, 255, 128, 128, 128 }
},
{ 1, 222, 248, 255, 216, 213, 128, 128, 128, 128, 128 },
{ 168, 175, 246, 252, 235, 205, 255, 255, 128, 128, 128 },
{ 47, 116, 215, 255, 211, 212, 255, 255, 128, 128, 128 } },
{ /* Coeff Band ( 6 )*/
{ 1, 121, 236, 253, 212, 214, 255, 255, 128, 128, 128 },
{ 141, 84, 213, 252, 201, 202, 255, 219, 128, 128, 128 },
{ 42, 80, 160, 240, 162, 185, 255, 205, 128, 128, 128 }
},
{ 1, 121, 236, 253, 212, 214, 255, 255, 128, 128, 128 },
{ 141, 84, 213, 252, 201, 202, 255, 219, 128, 128, 128 },
{ 42, 80, 160, 240, 162, 185, 255, 205, 128, 128, 128 } },
{ /* Coeff Band ( 7 )*/
{ 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 244, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 238, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
}
}
};
{ 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 244, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
{ 238, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 } } }
};
#ifdef __cplusplus
} // extern "C"

36
vp8/common/dequantize.c
View File

@ -8,36 +8,30 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#include "vp8/common/blockd.h"
#include "vpx_mem/vpx_mem.h"
void vp8_dequantize_b_c(BLOCKD *d, short *DQC)
{
int i;
short *DQ = d->dqcoeff;
short *Q = d->qcoeff;
void vp8_dequantize_b_c(BLOCKD *d, short *DQC) {
int i;
short *DQ = d->dqcoeff;
short *Q = d->qcoeff;
for (i = 0; i < 16; i++)
{
DQ[i] = Q[i] * DQC[i];
}
for (i = 0; i < 16; i++) {
DQ[i] = Q[i] * DQC[i];
}
}
void vp8_dequant_idct_add_c(short *input, short *dq,
unsigned char *dest, int stride)
{
int i;
void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *dest,
int stride) {
int i;
for (i = 0; i < 16; i++)
{
input[i] = dq[i] * input[i];
}
for (i = 0; i < 16; i++) {
input[i] = dq[i] * input[i];
}
vp8_short_idct4x4llm_c(input, dest, stride, dest, stride);
memset(input, 0, 32);
vp8_short_idct4x4llm_c(input, dest, stride, dest, stride);
memset(input, 0, 32);
}

189
vp8/common/entropy.c
View File

@ -15,47 +15,34 @@
#include "coefupdateprobs.h"
DECLARE_ALIGNED(16, const unsigned char, vp8_norm[256]) =
{
0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
DECLARE_ALIGNED(16, const unsigned char, vp8_norm[256]) = {
0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
3, 3, 3, 3, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]) =
{ 0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7};
DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]) = {
0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7
};
DECLARE_ALIGNED(16, const unsigned char,
vp8_prev_token_class[MAX_ENTROPY_TOKENS]) =
{ 0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 0};
DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]) =
{
0, 1, 4, 8,
5, 2, 3, 6,
9, 12, 13, 10,
7, 11, 14, 15,
vp8_prev_token_class[MAX_ENTROPY_TOKENS]) = {
0, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 0
};
DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]) =
{
1, 2, 6, 7,
3, 5, 8, 13,
4, 9, 12, 14,
10, 11, 15, 16
DECLARE_ALIGNED(16, const int, vp8_default_zig_zag1d[16]) = {
0, 1, 4, 8, 5, 2, 3, 6, 9, 12, 13, 10, 7, 11, 14, 15,
};
DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]) = {
1, 2, 6, 7, 3, 5, 8, 13, 4, 9, 12, 14, 10, 11, 15, 16
};
/* vp8_default_zig_zag_mask generated with:
@ -71,87 +58,69 @@ DECLARE_ALIGNED(16, const short, vp8_default_inv_zig_zag[16]) =
}
*/
DECLARE_ALIGNED(16, const short, vp8_default_zig_zag_mask[16]) =
{
1, 2, 32, 64,
4, 16, 128, 4096,
8, 256, 2048, 8192,
512, 1024, 16384, -32768
DECLARE_ALIGNED(16, const short, vp8_default_zig_zag_mask[16]) = {
1, 2, 32, 64, 4, 16, 128, 4096, 8, 256, 2048, 8192, 512, 1024, 16384, -32768
};
const int vp8_mb_feature_data_bits[MB_LVL_MAX] = {7, 6};
const int vp8_mb_feature_data_bits[MB_LVL_MAX] = { 7, 6 };
/* Array indices are identical to previously-existing CONTEXT_NODE indices */
const vp8_tree_index vp8_coef_tree[ 22] = /* corresponding _CONTEXT_NODEs */
{
-DCT_EOB_TOKEN, 2, /* 0 = EOB */
-ZERO_TOKEN, 4, /* 1 = ZERO */
-ONE_TOKEN, 6, /* 2 = ONE */
8, 12, /* 3 = LOW_VAL */
-TWO_TOKEN, 10, /* 4 = TWO */
-THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */
14, 16, /* 6 = HIGH_LOW */
-DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */
18, 20, /* 8 = CAT_THREEFOUR */
-DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */
-DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */
/* corresponding _CONTEXT_NODEs */
/* clang-format off */
const vp8_tree_index vp8_coef_tree[22] = {
-DCT_EOB_TOKEN, 2, /* 0 = EOB */
-ZERO_TOKEN, 4, /* 1 = ZERO */
-ONE_TOKEN, 6, /* 2 = ONE */
8, 12, /* 3 = LOW_VAL */
-TWO_TOKEN, 10, /* 4 = TWO */
-THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */
14, 16, /* 6 = HIGH_LOW */
-DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */
18, 20, /* 8 = CAT_THREEFOUR */
-DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */
-DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */
};
/* clang-format on */
/* vp8_coef_encodings generated with:
vp8_tokens_from_tree(vp8_coef_encodings, vp8_coef_tree);
*/
vp8_token vp8_coef_encodings[MAX_ENTROPY_TOKENS] =
{
{2, 2},
{6, 3},
{28, 5},
{58, 6},
{59, 6},
{60, 6},
{61, 6},
{124, 7},
{125, 7},
{126, 7},
{127, 7},
{0, 1}
vp8_token vp8_coef_encodings[MAX_ENTROPY_TOKENS] = {
{ 2, 2 }, { 6, 3 }, { 28, 5 }, { 58, 6 }, { 59, 6 }, { 60, 6 },
{ 61, 6 }, { 124, 7 }, { 125, 7 }, { 126, 7 }, { 127, 7 }, { 0, 1 }
};
/* Trees for extra bits. Probabilities are constant and
do not depend on previously encoded bits */
static const vp8_prob Pcat1[] = { 159};
static const vp8_prob Pcat2[] = { 165, 145};
static const vp8_prob Pcat3[] = { 173, 148, 140};
static const vp8_prob Pcat4[] = { 176, 155, 140, 135};
static const vp8_prob Pcat5[] = { 180, 157, 141, 134, 130};
static const vp8_prob Pcat6[] =
{ 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129};
static const vp8_prob Pcat1[] = { 159 };
static const vp8_prob Pcat2[] = { 165, 145 };
static const vp8_prob Pcat3[] = { 173, 148, 140 };
static const vp8_prob Pcat4[] = { 176, 155, 140, 135 };
static const vp8_prob Pcat5[] = { 180, 157, 141, 134, 130 };
static const vp8_prob Pcat6[] = { 254, 254, 243, 230, 196, 177,
153, 140, 133, 130, 129 };
/* tree index tables generated with:
void init_bit_tree(vp8_tree_index *p, int n)
{
int i = 0;
void init_bit_tree(vp8_tree_index *p, int n) {
int i = 0;
while (++i < n)
{
p[0] = p[1] = i << 1;
p += 2;
}
while (++i < n) {
p[0] = p[1] = i << 1;
p += 2;
}
p[0] = p[1] = 0;
p[0] = p[1] = 0;
}
void init_bit_trees()
{
init_bit_tree(cat1, 1);
init_bit_tree(cat2, 2);
init_bit_tree(cat3, 3);
init_bit_tree(cat4, 4);
init_bit_tree(cat5, 5);
init_bit_tree(cat6, 11);
void init_bit_trees() {
init_bit_tree(cat1, 1);
init_bit_tree(cat2, 2);
init_bit_tree(cat3, 3);
init_bit_tree(cat4, 4);
init_bit_tree(cat5, 5);
init_bit_tree(cat6, 11);
}
*/
@ -160,29 +129,19 @@ static const vp8_tree_index cat2[4] = { 2, 2, 0, 0 };
static const vp8_tree_index cat3[6] = { 2, 2, 4, 4, 0, 0 };
static const vp8_tree_index cat4[8] = { 2, 2, 4, 4, 6, 6, 0, 0 };
static const vp8_tree_index cat5[10] = { 2, 2, 4, 4, 6, 6, 8, 8, 0, 0 };
static const vp8_tree_index cat6[22] = { 2, 2, 4, 4, 6, 6, 8, 8, 10, 10, 12, 12,
14, 14, 16, 16, 18, 18, 20, 20, 0, 0 };
static const vp8_tree_index cat6[22] = {
2, 2, 4, 4, 6, 6, 8, 8, 10, 10, 12, 12, 14, 14, 16, 16, 18, 18, 20, 20, 0, 0
};
const vp8_extra_bit_struct vp8_extra_bits[12] =
{
{ 0, 0, 0, 0},
{ 0, 0, 0, 1},
{ 0, 0, 0, 2},
{ 0, 0, 0, 3},
{ 0, 0, 0, 4},
{ cat1, Pcat1, 1, 5},
{ cat2, Pcat2, 2, 7},
{ cat3, Pcat3, 3, 11},
{ cat4, Pcat4, 4, 19},
{ cat5, Pcat5, 5, 35},
{ cat6, Pcat6, 11, 67},
{ 0, 0, 0, 0}
const vp8_extra_bit_struct vp8_extra_bits[12] = {
{ 0, 0, 0, 0 }, { 0, 0, 0, 1 }, { 0, 0, 0, 2 },
{ 0, 0, 0, 3 }, { 0, 0, 0, 4 }, { cat1, Pcat1, 1, 5 },
{ cat2, Pcat2, 2, 7 }, { cat3, Pcat3, 3, 11 }, { cat4, Pcat4, 4, 19 },
{ cat5, Pcat5, 5, 35 }, { cat6, Pcat6, 11, 67 }, { 0, 0, 0, 0 }
};
#include "default_coef_probs.h"
void vp8_default_coef_probs(VP8_COMMON *pc)
{
memcpy(pc->fc.coef_probs, default_coef_probs, sizeof(default_coef_probs));
void vp8_default_coef_probs(VP8_COMMON *pc) {
memcpy(pc->fc.coef_probs, default_coef_probs, sizeof(default_coef_probs));
}

55
vp8/common/entropy.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ENTROPY_H_
#define VP8_COMMON_ENTROPY_H_
@ -21,18 +20,18 @@ extern "C" {
/* Coefficient token alphabet */
#define ZERO_TOKEN 0 /* 0 Extra Bits 0+0 */
#define ONE_TOKEN 1 /* 1 Extra Bits 0+1 */
#define TWO_TOKEN 2 /* 2 Extra Bits 0+1 */
#define THREE_TOKEN 3 /* 3 Extra Bits 0+1 */
#define FOUR_TOKEN 4 /* 4 Extra Bits 0+1 */
#define DCT_VAL_CATEGORY1 5 /* 5-6 Extra Bits 1+1 */
#define DCT_VAL_CATEGORY2 6 /* 7-10 Extra Bits 2+1 */
#define DCT_VAL_CATEGORY3 7 /* 11-18 Extra Bits 3+1 */
#define DCT_VAL_CATEGORY4 8 /* 19-34 Extra Bits 4+1 */
#define DCT_VAL_CATEGORY5 9 /* 35-66 Extra Bits 5+1 */
#define DCT_VAL_CATEGORY6 10 /* 67+ Extra Bits 11+1 */
#define DCT_EOB_TOKEN 11 /* EOB Extra Bits 0+0 */
#define ZERO_TOKEN 0 /* 0 Extra Bits 0+0 */
#define ONE_TOKEN 1 /* 1 Extra Bits 0+1 */
#define TWO_TOKEN 2 /* 2 Extra Bits 0+1 */
#define THREE_TOKEN 3 /* 3 Extra Bits 0+1 */
#define FOUR_TOKEN 4 /* 4 Extra Bits 0+1 */
#define DCT_VAL_CATEGORY1 5 /* 5-6 Extra Bits 1+1 */
#define DCT_VAL_CATEGORY2 6 /* 7-10 Extra Bits 2+1 */
#define DCT_VAL_CATEGORY3 7 /* 11-18 Extra Bits 3+1 */
#define DCT_VAL_CATEGORY4 8 /* 19-34 Extra Bits 4+1 */
#define DCT_VAL_CATEGORY5 9 /* 35-66 Extra Bits 5+1 */
#define DCT_VAL_CATEGORY6 10 /* 67+ Extra Bits 11+1 */
#define DCT_EOB_TOKEN 11 /* EOB Extra Bits 0+0 */
#define MAX_ENTROPY_TOKENS 12
#define ENTROPY_NODES 11
@ -41,21 +40,20 @@ extern const vp8_tree_index vp8_coef_tree[];
extern const struct vp8_token_struct vp8_coef_encodings[MAX_ENTROPY_TOKENS];
typedef struct
{
vp8_tree_p tree;
const vp8_prob *prob;
int Len;
int base_val;
typedef struct {
vp8_tree_p tree;
const vp8_prob *prob;
int Len;
int base_val;
} vp8_extra_bit_struct;
extern const vp8_extra_bit_struct vp8_extra_bits[12]; /* indexed by token value */
extern const vp8_extra_bit_struct
vp8_extra_bits[12]; /* indexed by token value */
#define PROB_UPDATE_BASELINE_COST 7
#define MAX_PROB 255
#define DCT_MAX_VALUE 2048
#define PROB_UPDATE_BASELINE_COST 7
#define MAX_PROB 255
#define DCT_MAX_VALUE 2048
/* Coefficients are predicted via a 3-dimensional probability table. */
@ -86,12 +84,13 @@ extern DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]);
distinct bands). */
/*# define DC_TOKEN_CONTEXTS 3*/ /* 00, 0!0, !0!0 */
# define PREV_COEF_CONTEXTS 3
#define PREV_COEF_CONTEXTS 3
extern DECLARE_ALIGNED(16, const unsigned char, vp8_prev_token_class[MAX_ENTROPY_TOKENS]);
extern const vp8_prob vp8_coef_update_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
extern DECLARE_ALIGNED(16, const unsigned char,
vp8_prev_token_class[MAX_ENTROPY_TOKENS]);
extern const vp8_prob vp8_coef_update_probs[BLOCK_TYPES][COEF_BANDS]
[PREV_COEF_CONTEXTS][ENTROPY_NODES];
struct VP8Common;
void vp8_default_coef_probs(struct VP8Common *);

183
vp8/common/entropymode.c
View File

@ -16,156 +16,97 @@
#include "vp8_entropymodedata.h"
int vp8_mv_cont(const int_mv *l, const int_mv *a)
{
int lez = (l->as_int == 0);
int aez = (a->as_int == 0);
int lea = (l->as_int == a->as_int);
int vp8_mv_cont(const int_mv *l, const int_mv *a) {
int lez = (l->as_int == 0);
int aez = (a->as_int == 0);
int lea = (l->as_int == a->as_int);
if (lea && lez)
return SUBMVREF_LEFT_ABOVE_ZED;
if (lea && lez) return SUBMVREF_LEFT_ABOVE_ZED;
if (lea)
return SUBMVREF_LEFT_ABOVE_SAME;
if (lea) return SUBMVREF_LEFT_ABOVE_SAME;
if (aez)
return SUBMVREF_ABOVE_ZED;
if (aez) return SUBMVREF_ABOVE_ZED;
if (lez)
return SUBMVREF_LEFT_ZED;
if (lez) return SUBMVREF_LEFT_ZED;
return SUBMVREF_NORMAL;
return SUBMVREF_NORMAL;
}
static const vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1] = { 180, 162, 25};
static const vp8_prob sub_mv_ref_prob[VP8_SUBMVREFS - 1] = { 180, 162, 25 };
const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1] =
{
{ 147, 136, 18 },
{ 106, 145, 1 },
{ 179, 121, 1 },
{ 223, 1 , 34 },
{ 208, 1 , 1 }
const vp8_prob vp8_sub_mv_ref_prob2[SUBMVREF_COUNT][VP8_SUBMVREFS - 1] = {
{ 147, 136, 18 },
{ 106, 145, 1 },
{ 179, 121, 1 },
{ 223, 1, 34 },
{ 208, 1, 1 }
};
const vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS] =
{
{
0, 0, 0, 0,
0, 0, 0, 0,
1, 1, 1, 1,
1, 1, 1, 1,
},
{
0, 0, 1, 1,
0, 0, 1, 1,
0, 0, 1, 1,
0, 0, 1, 1,
},
{
0, 0, 1, 1,
0, 0, 1, 1,
2, 2, 3, 3,
2, 2, 3, 3,
},
{
0, 1, 2, 3,
4, 5, 6, 7,
8, 9, 10, 11,
12, 13, 14, 15,
}
const vp8_mbsplit vp8_mbsplits[VP8_NUMMBSPLITS] = {
{
0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1,
},
{
0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1,
},
{
0, 0, 1, 1, 0, 0, 1, 1, 2, 2, 3, 3, 2, 2, 3, 3,
},
{
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
}
};
const int vp8_mbsplit_count [VP8_NUMMBSPLITS] = { 2, 2, 4, 16};
const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1] = { 110, 111, 150};
const int vp8_mbsplit_count[VP8_NUMMBSPLITS] = { 2, 2, 4, 16 };
const vp8_prob vp8_mbsplit_probs[VP8_NUMMBSPLITS - 1] = { 110, 111, 150 };
/* Array indices are identical to previously-existing INTRAMODECONTEXTNODES. */
const vp8_tree_index vp8_bmode_tree[18] = /* INTRAMODECONTEXTNODE value */
{
-B_DC_PRED, 2, /* 0 = DC_NODE */
-B_TM_PRED, 4, /* 1 = TM_NODE */
-B_VE_PRED, 6, /* 2 = VE_NODE */
8, 12, /* 3 = COM_NODE */
-B_HE_PRED, 10, /* 4 = HE_NODE */
-B_RD_PRED, -B_VR_PRED, /* 5 = RD_NODE */
-B_LD_PRED, 14, /* 6 = LD_NODE */
-B_VL_PRED, 16, /* 7 = VL_NODE */
-B_HD_PRED, -B_HU_PRED /* 8 = HD_NODE */
};
const vp8_tree_index vp8_bmode_tree[18] = /* INTRAMODECONTEXTNODE value */
{
-B_DC_PRED, 2, /* 0 = DC_NODE */
-B_TM_PRED, 4, /* 1 = TM_NODE */
-B_VE_PRED, 6, /* 2 = VE_NODE */
8, 12, /* 3 = COM_NODE */
-B_HE_PRED, 10, /* 4 = HE_NODE */
-B_RD_PRED, -B_VR_PRED, /* 5 = RD_NODE */
-B_LD_PRED, 14, /* 6 = LD_NODE */
-B_VL_PRED, 16, /* 7 = VL_NODE */
-B_HD_PRED, -B_HU_PRED /* 8 = HD_NODE */
};
/* Again, these trees use the same probability indices as their
explicitly-programmed predecessors. */
const vp8_tree_index vp8_ymode_tree[8] =
{
-DC_PRED, 2,
4, 6,
-V_PRED, -H_PRED,
-TM_PRED, -B_PRED
const vp8_tree_index vp8_ymode_tree[8] = {
-DC_PRED, 2, 4, 6, -V_PRED, -H_PRED, -TM_PRED, -B_PRED
};
const vp8_tree_index vp8_kf_ymode_tree[8] =
{
-B_PRED, 2,
4, 6,
-DC_PRED, -V_PRED,
-H_PRED, -TM_PRED
const vp8_tree_index vp8_kf_ymode_tree[8] = {
-B_PRED, 2, 4, 6, -DC_PRED, -V_PRED, -H_PRED, -TM_PRED
};
const vp8_tree_index vp8_uv_mode_tree[6] =
{
-DC_PRED, 2,
-V_PRED, 4,
-H_PRED, -TM_PRED
};
const vp8_tree_index vp8_uv_mode_tree[6] = { -DC_PRED, 2, -V_PRED,
4, -H_PRED, -TM_PRED };
const vp8_tree_index vp8_mbsplit_tree[6] =
{
-3, 2,
-2, 4,
-0, -1
};
const vp8_tree_index vp8_mbsplit_tree[6] = { -3, 2, -2, 4, -0, -1 };
const vp8_tree_index vp8_mv_ref_tree[8] =
{
-ZEROMV, 2,
-NEARESTMV, 4,
-NEARMV, 6,
-NEWMV, -SPLITMV
};
const vp8_tree_index vp8_mv_ref_tree[8] = { -ZEROMV, 2, -NEARESTMV, 4,
-NEARMV, 6, -NEWMV, -SPLITMV };
const vp8_tree_index vp8_sub_mv_ref_tree[6] =
{
-LEFT4X4, 2,
-ABOVE4X4, 4,
-ZERO4X4, -NEW4X4
};
const vp8_tree_index vp8_sub_mv_ref_tree[6] = { -LEFT4X4, 2, -ABOVE4X4,
4, -ZERO4X4, -NEW4X4 };
const vp8_tree_index vp8_small_mvtree [14] =
{
2, 8,
4, 6,
-0, -1,
-2, -3,
10, 12,
-4, -5,
-6, -7
};
const vp8_tree_index vp8_small_mvtree[14] = { 2, 8, 4, 6, -0, -1, -2,
-3, 10, 12, -4, -5, -6, -7 };
void vp8_init_mbmode_probs(VP8_COMMON *x)
{
memcpy(x->fc.ymode_prob, vp8_ymode_prob, sizeof(vp8_ymode_prob));
memcpy(x->fc.uv_mode_prob, vp8_uv_mode_prob, sizeof(vp8_uv_mode_prob));
memcpy(x->fc.sub_mv_ref_prob, sub_mv_ref_prob, sizeof(sub_mv_ref_prob));
void vp8_init_mbmode_probs(VP8_COMMON *x) {
memcpy(x->fc.ymode_prob, vp8_ymode_prob, sizeof(vp8_ymode_prob));
memcpy(x->fc.uv_mode_prob, vp8_uv_mode_prob, sizeof(vp8_uv_mode_prob));
memcpy(x->fc.sub_mv_ref_prob, sub_mv_ref_prob, sizeof(sub_mv_ref_prob));
}
void vp8_default_bmode_probs(vp8_prob p [VP8_BINTRAMODES-1])
{
memcpy(p, vp8_bmode_prob, sizeof(vp8_bmode_prob));
void vp8_default_bmode_probs(vp8_prob p[VP8_BINTRAMODES - 1]) {
memcpy(p, vp8_bmode_prob, sizeof(vp8_bmode_prob));
}

55
vp8/common/entropymode.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ENTROPYMODE_H_
#define VP8_COMMON_ENTROPYMODE_H_
@ -19,42 +18,40 @@
extern "C" {
#endif
typedef enum
{
SUBMVREF_NORMAL,
SUBMVREF_LEFT_ZED,
SUBMVREF_ABOVE_ZED,
SUBMVREF_LEFT_ABOVE_SAME,
SUBMVREF_LEFT_ABOVE_ZED
typedef enum {
SUBMVREF_NORMAL,
SUBMVREF_LEFT_ZED,
SUBMVREF_ABOVE_ZED,
SUBMVREF_LEFT_ABOVE_SAME,
SUBMVREF_LEFT_ABOVE_ZED
} sumvfref_t;
typedef int vp8_mbsplit[16];
#define VP8_NUMMBSPLITS 4
extern const vp8_mbsplit vp8_mbsplits [VP8_NUMMBSPLITS];
extern const vp8_mbsplit vp8_mbsplits[VP8_NUMMBSPLITS];
extern const int vp8_mbsplit_count [VP8_NUMMBSPLITS]; /* # of subsets */
extern const int vp8_mbsplit_count[VP8_NUMMBSPLITS]; /* # of subsets */
extern const vp8_prob vp8_mbsplit_probs [VP8_NUMMBSPLITS-1];
extern const vp8_prob vp8_mbsplit_probs[VP8_NUMMBSPLITS - 1];
extern int vp8_mv_cont(const int_mv *l, const int_mv *a);
#define SUBMVREF_COUNT 5
extern const vp8_prob vp8_sub_mv_ref_prob2 [SUBMVREF_COUNT][VP8_SUBMVREFS-1];
extern const unsigned int vp8_kf_default_bmode_counts [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES];
extern const vp8_prob vp8_sub_mv_ref_prob2[SUBMVREF_COUNT][VP8_SUBMVREFS - 1];
extern const unsigned int vp8_kf_default_bmode_counts
[VP8_BINTRAMODES][VP8_BINTRAMODES][VP8_BINTRAMODES];
extern const vp8_tree_index vp8_bmode_tree[];
extern const vp8_tree_index vp8_ymode_tree[];
extern const vp8_tree_index vp8_kf_ymode_tree[];
extern const vp8_tree_index vp8_uv_mode_tree[];
extern const vp8_tree_index vp8_ymode_tree[];
extern const vp8_tree_index vp8_kf_ymode_tree[];
extern const vp8_tree_index vp8_uv_mode_tree[];
extern const vp8_tree_index vp8_mbsplit_tree[];
extern const vp8_tree_index vp8_mv_ref_tree[];
extern const vp8_tree_index vp8_sub_mv_ref_tree[];
extern const vp8_tree_index vp8_mbsplit_tree[];
extern const vp8_tree_index vp8_mv_ref_tree[];
extern const vp8_tree_index vp8_sub_mv_ref_tree[];
extern const struct vp8_token_struct vp8_bmode_encodings[VP8_BINTRAMODES];
extern const struct vp8_token_struct vp8_ymode_encodings[VP8_YMODES];
@ -65,21 +62,23 @@ extern const struct vp8_token_struct vp8_mbsplit_encodings[VP8_NUMMBSPLITS];
/* Inter mode values do not start at zero */
extern const struct vp8_token_struct vp8_mv_ref_encoding_array[VP8_MVREFS];
extern const struct vp8_token_struct vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS];
extern const struct vp8_token_struct
vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS];
extern const vp8_tree_index vp8_small_mvtree[];
extern const struct vp8_token_struct vp8_small_mvencodings[8];
/* Key frame default mode probs */
extern const vp8_prob vp8_kf_bmode_prob[VP8_BINTRAMODES][VP8_BINTRAMODES]
[VP8_BINTRAMODES-1];
extern const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES-1];
extern const vp8_prob vp8_kf_ymode_prob[VP8_YMODES-1];
extern const vp8_prob
vp8_kf_bmode_prob[VP8_BINTRAMODES][VP8_BINTRAMODES][VP8_BINTRAMODES - 1];
extern const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES - 1];
extern const vp8_prob vp8_kf_ymode_prob[VP8_YMODES - 1];
void vp8_init_mbmode_probs(VP8_COMMON *x);
void vp8_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES-1]);
void vp8_kf_default_bmode_probs(vp8_prob dest [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1]);
void vp8_default_bmode_probs(vp8_prob dest[VP8_BINTRAMODES - 1]);
void vp8_kf_default_bmode_probs(
vp8_prob dest[VP8_BINTRAMODES][VP8_BINTRAMODES][VP8_BINTRAMODES - 1]);
#ifdef __cplusplus
} // extern "C"

62
vp8/common/entropymv.c
View File

@ -8,42 +8,40 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "entropymv.h"
const MV_CONTEXT vp8_mv_update_probs[2] =
{
{{
237,
246,
253, 253, 254, 254, 254, 254, 254,
254, 254, 254, 254, 254, 250, 250, 252, 254, 254
}},
{{
231,
243,
245, 253, 254, 254, 254, 254, 254,
254, 254, 254, 254, 254, 251, 251, 254, 254, 254
}}
/* clang-format off */
const MV_CONTEXT vp8_mv_update_probs[2] = {
{ {
237,
246,
253, 253, 254, 254, 254, 254, 254,
254, 254, 254, 254, 254, 250, 250, 252, 254, 254
} },
{ {
231,
243,
245, 253, 254, 254, 254, 254, 254,
254, 254, 254, 254, 254, 251, 251, 254, 254, 254
} }
};
const MV_CONTEXT vp8_default_mv_context[2] =
{
{{
/* row */
162, /* is short */
128, /* sign */
225, 146, 172, 147, 214, 39, 156, /* short tree */
128, 129, 132, 75, 145, 178, 206, 239, 254, 254 /* long bits */
}},
/* clang-format on */
const MV_CONTEXT vp8_default_mv_context[2] = {
{ {
/* row */
162, /* is short */
128, /* sign */
225, 146, 172, 147, 214, 39, 156, /* short tree */
128, 129, 132, 75, 145, 178, 206, 239, 254, 254 /* long bits */
} },
{ {
/* same for column */
164, /* is short */
128, /**/
204, 170, 119, 235, 140, 230, 228, /**/
128, 130, 130, 74, 148, 180, 203, 236, 254, 254 /* long bits */
{{
/* same for column */
164, /* is short */
128,
204, 170, 119, 235, 140, 230, 228,
128, 130, 130, 74, 148, 180, 203, 236, 254, 254 /* long bits */
}}
} }
};

33
vp8/common/entropymv.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ENTROPYMV_H_
#define VP8_COMMON_ENTROPYMV_H_
@ -18,29 +17,27 @@
extern "C" {
#endif
enum
{
mv_max = 1023, /* max absolute value of a MV component */
MVvals = (2 * mv_max) + 1, /* # possible values "" */
mvfp_max = 255, /* max absolute value of a full pixel MV component */
MVfpvals = (2 * mvfp_max) +1, /* # possible full pixel MV values */
enum {
mv_max = 1023, /* max absolute value of a MV component */
MVvals = (2 * mv_max) + 1, /* # possible values "" */
mvfp_max = 255, /* max absolute value of a full pixel MV component */
MVfpvals = (2 * mvfp_max) + 1, /* # possible full pixel MV values */
mvlong_width = 10, /* Large MVs have 9 bit magnitudes */
mvnum_short = 8, /* magnitudes 0 through 7 */
mvlong_width = 10, /* Large MVs have 9 bit magnitudes */
mvnum_short = 8, /* magnitudes 0 through 7 */
/* probability offsets for coding each MV component */
/* probability offsets for coding each MV component */
mvpis_short = 0, /* short (<= 7) vs long (>= 8) */
MVPsign, /* sign for non-zero */
MVPshort, /* 8 short values = 7-position tree */
mvpis_short = 0, /* short (<= 7) vs long (>= 8) */
MVPsign, /* sign for non-zero */
MVPshort, /* 8 short values = 7-position tree */
MVPbits = MVPshort + mvnum_short - 1, /* mvlong_width long value bits */
MVPcount = MVPbits + mvlong_width /* (with independent probabilities) */
MVPbits = MVPshort + mvnum_short - 1, /* mvlong_width long value bits */
MVPcount = MVPbits + mvlong_width /* (with independent probabilities) */
};
typedef struct mv_context
{
vp8_prob prob[MVPcount]; /* often come in row, col pairs */
typedef struct mv_context {
vp8_prob prob[MVPcount]; /* often come in row, col pairs */
} MV_CONTEXT;
extern const MV_CONTEXT vp8_mv_update_probs[2], vp8_default_mv_context[2];

259
vp8/common/extend.c
View File

@ -8,181 +8,146 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "extend.h"
#include "vpx_mem/vpx_mem.h"
static void copy_and_extend_plane(unsigned char *s, /* source */
int sp, /* source pitch */
unsigned char *d, /* destination */
int dp, /* destination pitch */
int h, /* height */
int w, /* width */
int et, /* extend top border */
int el, /* extend left border */
int eb, /* extend bottom border */
int er /* extend right border */
) {
int i;
unsigned char *src_ptr1, *src_ptr2;
unsigned char *dest_ptr1, *dest_ptr2;
int linesize;
static void copy_and_extend_plane
(
unsigned char *s, /* source */
int sp, /* source pitch */
unsigned char *d, /* destination */
int dp, /* destination pitch */
int h, /* height */
int w, /* width */
int et, /* extend top border */
int el, /* extend left border */
int eb, /* extend bottom border */
int er /* extend right border */
)
{
int i;
unsigned char *src_ptr1, *src_ptr2;
unsigned char *dest_ptr1, *dest_ptr2;
int linesize;
/* copy the left and right most columns out */
src_ptr1 = s;
src_ptr2 = s + w - 1;
dest_ptr1 = d - el;
dest_ptr2 = d + w;
/* copy the left and right most columns out */
src_ptr1 = s;
src_ptr2 = s + w - 1;
dest_ptr1 = d - el;
dest_ptr2 = d + w;
for (i = 0; i < h; i++) {
memset(dest_ptr1, src_ptr1[0], el);
memcpy(dest_ptr1 + el, src_ptr1, w);
memset(dest_ptr2, src_ptr2[0], er);
src_ptr1 += sp;
src_ptr2 += sp;
dest_ptr1 += dp;
dest_ptr2 += dp;
}
for (i = 0; i < h; i++)
{
memset(dest_ptr1, src_ptr1[0], el);
memcpy(dest_ptr1 + el, src_ptr1, w);
memset(dest_ptr2, src_ptr2[0], er);
src_ptr1 += sp;
src_ptr2 += sp;
dest_ptr1 += dp;
dest_ptr2 += dp;
}
/* Now copy the top and bottom lines into each line of the respective
* borders
*/
src_ptr1 = d - el;
src_ptr2 = d + dp * (h - 1) - el;
dest_ptr1 = d + dp * (-et) - el;
dest_ptr2 = d + dp * (h)-el;
linesize = el + er + w;
/* Now copy the top and bottom lines into each line of the respective
* borders
*/
src_ptr1 = d - el;
src_ptr2 = d + dp * (h - 1) - el;
dest_ptr1 = d + dp * (-et) - el;
dest_ptr2 = d + dp * (h) - el;
linesize = el + er + w;
for (i = 0; i < et; i++) {
memcpy(dest_ptr1, src_ptr1, linesize);
dest_ptr1 += dp;
}
for (i = 0; i < et; i++)
{
memcpy(dest_ptr1, src_ptr1, linesize);
dest_ptr1 += dp;
}
for (i = 0; i < eb; i++)
{
memcpy(dest_ptr2, src_ptr2, linesize);
dest_ptr2 += dp;
}
for (i = 0; i < eb; i++) {
memcpy(dest_ptr2, src_ptr2, linesize);
dest_ptr2 += dp;
}
}
void vp8_copy_and_extend_frame(YV12_BUFFER_CONFIG *src,
YV12_BUFFER_CONFIG *dst)
{
int et = dst->border;
int el = dst->border;
int eb = dst->border + dst->y_height - src->y_height;
int er = dst->border + dst->y_width - src->y_width;
YV12_BUFFER_CONFIG *dst) {
int et = dst->border;
int el = dst->border;
int eb = dst->border + dst->y_height - src->y_height;
int er = dst->border + dst->y_width - src->y_width;
copy_and_extend_plane(src->y_buffer, src->y_stride,
dst->y_buffer, dst->y_stride,
src->y_height, src->y_width,
et, el, eb, er);
copy_and_extend_plane(src->y_buffer, src->y_stride, dst->y_buffer,
dst->y_stride, src->y_height, src->y_width, et, el, eb,
er);
et = dst->border >> 1;
el = dst->border >> 1;
eb = (dst->border >> 1) + dst->uv_height - src->uv_height;
er = (dst->border >> 1) + dst->uv_width - src->uv_width;
et = dst->border >> 1;
el = dst->border >> 1;
eb = (dst->border >> 1) + dst->uv_height - src->uv_height;
er = (dst->border >> 1) + dst->uv_width - src->uv_width;
copy_and_extend_plane(src->u_buffer, src->uv_stride,
dst->u_buffer, dst->uv_stride,
src->uv_height, src->uv_width,
et, el, eb, er);
copy_and_extend_plane(src->u_buffer, src->uv_stride, dst->u_buffer,
dst->uv_stride, src->uv_height, src->uv_width, et, el,
eb, er);
copy_and_extend_plane(src->v_buffer, src->uv_stride,
dst->v_buffer, dst->uv_stride,
src->uv_height, src->uv_width,
et, el, eb, er);
copy_and_extend_plane(src->v_buffer, src->uv_stride, dst->v_buffer,
dst->uv_stride, src->uv_height, src->uv_width, et, el,
eb, er);
}
void vp8_copy_and_extend_frame_with_rect(YV12_BUFFER_CONFIG *src,
YV12_BUFFER_CONFIG *dst,
int srcy, int srcx,
int srch, int srcw)
{
int et = dst->border;
int el = dst->border;
int eb = dst->border + dst->y_height - src->y_height;
int er = dst->border + dst->y_width - src->y_width;
int src_y_offset = srcy * src->y_stride + srcx;
int dst_y_offset = srcy * dst->y_stride + srcx;
int src_uv_offset = ((srcy * src->uv_stride) >> 1) + (srcx >> 1);
int dst_uv_offset = ((srcy * dst->uv_stride) >> 1) + (srcx >> 1);
YV12_BUFFER_CONFIG *dst, int srcy,
int srcx, int srch, int srcw) {
int et = dst->border;
int el = dst->border;
int eb = dst->border + dst->y_height - src->y_height;
int er = dst->border + dst->y_width - src->y_width;
int src_y_offset = srcy * src->y_stride + srcx;
int dst_y_offset = srcy * dst->y_stride + srcx;
int src_uv_offset = ((srcy * src->uv_stride) >> 1) + (srcx >> 1);
int dst_uv_offset = ((srcy * dst->uv_stride) >> 1) + (srcx >> 1);
/* If the side is not touching the bounder then don't extend. */
if (srcy)
et = 0;
if (srcx)
el = 0;
if (srcy + srch != src->y_height)
eb = 0;
if (srcx + srcw != src->y_width)
er = 0;
/* If the side is not touching the bounder then don't extend. */
if (srcy) et = 0;
if (srcx) el = 0;
if (srcy + srch != src->y_height) eb = 0;
if (srcx + srcw != src->y_width) er = 0;
copy_and_extend_plane(src->y_buffer + src_y_offset,
src->y_stride,
dst->y_buffer + dst_y_offset,
dst->y_stride,
srch, srcw,
et, el, eb, er);
copy_and_extend_plane(src->y_buffer + src_y_offset, src->y_stride,
dst->y_buffer + dst_y_offset, dst->y_stride, srch, srcw,
et, el, eb, er);
et = (et + 1) >> 1;
el = (el + 1) >> 1;
eb = (eb + 1) >> 1;
er = (er + 1) >> 1;
srch = (srch + 1) >> 1;
srcw = (srcw + 1) >> 1;
et = (et + 1) >> 1;
el = (el + 1) >> 1;
eb = (eb + 1) >> 1;
er = (er + 1) >> 1;
srch = (srch + 1) >> 1;
srcw = (srcw + 1) >> 1;
copy_and_extend_plane(src->u_buffer + src_uv_offset,
src->uv_stride,
dst->u_buffer + dst_uv_offset,
dst->uv_stride,
srch, srcw,
et, el, eb, er);
copy_and_extend_plane(src->u_buffer + src_uv_offset, src->uv_stride,
dst->u_buffer + dst_uv_offset, dst->uv_stride, srch,
srcw, et, el, eb, er);
copy_and_extend_plane(src->v_buffer + src_uv_offset,
src->uv_stride,
dst->v_buffer + dst_uv_offset,
dst->uv_stride,
srch, srcw,
et, el, eb, er);
copy_and_extend_plane(src->v_buffer + src_uv_offset, src->uv_stride,
dst->v_buffer + dst_uv_offset, dst->uv_stride, srch,
srcw, et, el, eb, er);
}
/* note the extension is only for the last row, for intra prediction purpose */
void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf,
unsigned char *YPtr,
unsigned char *UPtr,
unsigned char *VPtr)
{
int i;
void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr,
unsigned char *UPtr, unsigned char *VPtr) {
int i;
YPtr += ybf->y_stride * 14;
UPtr += ybf->uv_stride * 6;
VPtr += ybf->uv_stride * 6;
YPtr += ybf->y_stride * 14;
UPtr += ybf->uv_stride * 6;
VPtr += ybf->uv_stride * 6;
for (i = 0; i < 4; i++)
{
YPtr[i] = YPtr[-1];
UPtr[i] = UPtr[-1];
VPtr[i] = VPtr[-1];
}
for (i = 0; i < 4; i++) {
YPtr[i] = YPtr[-1];
UPtr[i] = UPtr[-1];
VPtr[i] = VPtr[-1];
}
YPtr += ybf->y_stride;
UPtr += ybf->uv_stride;
VPtr += ybf->uv_stride;
YPtr += ybf->y_stride;
UPtr += ybf->uv_stride;
VPtr += ybf->uv_stride;
for (i = 0; i < 4; i++)
{
YPtr[i] = YPtr[-1];
UPtr[i] = UPtr[-1];
VPtr[i] = VPtr[-1];
}
for (i = 0; i < 4; i++) {
YPtr[i] = YPtr[-1];
UPtr[i] = UPtr[-1];
VPtr[i] = VPtr[-1];
}
}

9
vp8/common/extend.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_EXTEND_H_
#define VP8_COMMON_EXTEND_H_
@ -18,13 +17,13 @@
extern "C" {
#endif
void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr, unsigned char *UPtr, unsigned char *VPtr);
void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr,
unsigned char *UPtr, unsigned char *VPtr);
void vp8_copy_and_extend_frame(YV12_BUFFER_CONFIG *src,
YV12_BUFFER_CONFIG *dst);
void vp8_copy_and_extend_frame_with_rect(YV12_BUFFER_CONFIG *src,
YV12_BUFFER_CONFIG *dst,
int srcy, int srcx,
int srch, int srcw);
YV12_BUFFER_CONFIG *dst, int srcy,
int srcx, int srch, int srcw);
#ifdef __cplusplus
} // extern "C"

555
vp8/common/filter.c
View File

@ -8,243 +8,186 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "filter.h"
#include "./vp8_rtcd.h"
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters[8][2]) =
{
{ 128, 0 },
{ 112, 16 },
{ 96, 32 },
{ 80, 48 },
{ 64, 64 },
{ 48, 80 },
{ 32, 96 },
{ 16, 112 }
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters[8][2]) = {
{ 128, 0 }, { 112, 16 }, { 96, 32 }, { 80, 48 },
{ 64, 64 }, { 48, 80 }, { 32, 96 }, { 16, 112 }
};
DECLARE_ALIGNED(16, const short, vp8_sub_pel_filters[8][6]) =
{
DECLARE_ALIGNED(16, const short, vp8_sub_pel_filters[8][6]) = {
{ 0, 0, 128, 0, 0, 0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
{ 0, -6, 123, 12, -1, 0 },
{ 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
{ 0, -9, 93, 50, -6, 0 },
{ 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
{ 0, -6, 50, 93, -9, 0 },
{ 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
{ 0, -1, 12, 123, -6, 0 },
{ 0, 0, 128, 0, 0,
0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
{ 0, -6, 123, 12, -1, 0 },
{ 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
{ 0, -9, 93, 50, -6, 0 },
{ 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
{ 0, -6, 50, 93, -9, 0 },
{ 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
{ 0, -1, 12, 123, -6, 0 },
};
static void filter_block2d_first_pass
(
unsigned char *src_ptr,
int *output_ptr,
unsigned int src_pixels_per_line,
unsigned int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter
)
{
unsigned int i, j;
int Temp;
static void filter_block2d_first_pass(unsigned char *src_ptr, int *output_ptr,
unsigned int src_pixels_per_line,
unsigned int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter) {
unsigned int i, j;
int Temp;
for (i = 0; i < output_height; i++)
{
for (j = 0; j < output_width; j++)
{
Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
((int)src_ptr[0] * vp8_filter[2]) +
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
for (i = 0; i < output_height; i++) {
for (j = 0; j < output_width; j++) {
Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
((int)src_ptr[0] * vp8_filter[2]) +
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2 * pixel_step] * vp8_filter[4]) +
((int)src_ptr[3 * pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
if (Temp < 0)
Temp = 0;
else if (Temp > 255)
Temp = 255;
if (Temp < 0)
Temp = 0;
else if (Temp > 255)
Temp = 255;
output_ptr[j] = Temp;
src_ptr++;
}
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
output_ptr[j] = Temp;
src_ptr++;
}
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
}
static void filter_block2d_second_pass
(
int *src_ptr,
unsigned char *output_ptr,
int output_pitch,
unsigned int src_pixels_per_line,
unsigned int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter
)
{
unsigned int i, j;
int Temp;
static void filter_block2d_second_pass(int *src_ptr, unsigned char *output_ptr,
int output_pitch,
unsigned int src_pixels_per_line,
unsigned int pixel_step,
unsigned int output_height,
unsigned int output_width,
const short *vp8_filter) {
unsigned int i, j;
int Temp;
for (i = 0; i < output_height; i++)
{
for (j = 0; j < output_width; j++)
{
/* Apply filter */
Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
((int)src_ptr[0] * vp8_filter[2]) +
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
for (i = 0; i < output_height; i++) {
for (j = 0; j < output_width; j++) {
/* Apply filter */
Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
((int)src_ptr[0] * vp8_filter[2]) +
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2 * pixel_step] * vp8_filter[4]) +
((int)src_ptr[3 * pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
if (Temp < 0)
Temp = 0;
else if (Temp > 255)
Temp = 255;
if (Temp < 0)
Temp = 0;
else if (Temp > 255)
Temp = 255;
output_ptr[j] = (unsigned char)Temp;
src_ptr++;
}
/* Start next row */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_pitch;
output_ptr[j] = (unsigned char)Temp;
src_ptr++;
}
/* Start next row */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_pitch;
}
}
static void filter_block2d(unsigned char *src_ptr, unsigned char *output_ptr,
unsigned int src_pixels_per_line, int output_pitch,
const short *HFilter, const short *VFilter) {
int FData[9 * 4]; /* Temp data buffer used in filtering */
static void filter_block2d
(
unsigned char *src_ptr,
unsigned char *output_ptr,
unsigned int src_pixels_per_line,
int output_pitch,
const short *HFilter,
const short *VFilter
)
{
int FData[9*4]; /* Temp data buffer used in filtering */
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData,
src_pixels_per_line, 1, 9, 4, HFilter);
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 4, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 8, output_ptr, output_pitch, 4, 4, 4, 4, VFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 8, output_ptr, output_pitch, 4, 4, 4, 4,
VFilter);
}
void vp8_sixtap_predict4x4_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
void vp8_sixtap_predict4x4_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
filter_block2d(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter);
filter_block2d(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter,
VFilter);
}
void vp8_sixtap_predict8x8_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
int FData[13*16]; /* Temp data buffer used in filtering */
void vp8_sixtap_predict8x8_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
int FData[13 * 16]; /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 13, 8, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 8, 8, VFilter);
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData,
src_pixels_per_line, 1, 13, 8, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 8, 8,
VFilter);
}
void vp8_sixtap_predict8x4_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
int FData[13*16]; /* Temp data buffer used in filtering */
void vp8_sixtap_predict8x4_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
int FData[13 * 16]; /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 8, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 4, 8, VFilter);
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData,
src_pixels_per_line, 1, 9, 8, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 4, 8,
VFilter);
}
void vp8_sixtap_predict16x16_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
int FData[21*24]; /* Temp data buffer used in filtering */
void vp8_sixtap_predict16x16_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
int FData[21 * 24]; /* Temp data buffer used in filtering */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
HFilter = vp8_sub_pel_filters[xoffset]; /* 6 tap */
VFilter = vp8_sub_pel_filters[yoffset]; /* 6 tap */
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 21, 16, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 32, dst_ptr, dst_pitch, 16, 16, 16, 16, VFilter);
/* First filter 1-D horizontally... */
filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData,
src_pixels_per_line, 1, 21, 16, HFilter);
/* then filter verticaly... */
filter_block2d_second_pass(FData + 32, dst_ptr, dst_pitch, 16, 16, 16, 16,
VFilter);
}
/****************************************************************************
*
* ROUTINE : filter_block2d_bil_first_pass
@ -267,33 +210,25 @@ void vp8_sixtap_predict16x16_c
* Two filter taps should sum to VP8_FILTER_WEIGHT.
*
****************************************************************************/
static void filter_block2d_bil_first_pass
(
unsigned char *src_ptr,
unsigned short *dst_ptr,
unsigned int src_stride,
unsigned int height,
unsigned int width,
const short *vp8_filter
)
{
unsigned int i, j;
static void filter_block2d_bil_first_pass(
unsigned char *src_ptr, unsigned short *dst_ptr, unsigned int src_stride,
unsigned int height, unsigned int width, const short *vp8_filter) {
unsigned int i, j;
for (i = 0; i < height; i++)
{
for (j = 0; j < width; j++)
{
/* Apply bilinear filter */
dst_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[1] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
src_ptr++;
}
/* Next row... */
src_ptr += src_stride - width;
dst_ptr += width;
for (i = 0; i < height; i++) {
for (j = 0; j < width; j++) {
/* Apply bilinear filter */
dst_ptr[j] =
(((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[1] * vp8_filter[1]) + (VP8_FILTER_WEIGHT / 2)) >>
VP8_FILTER_SHIFT;
src_ptr++;
}
/* Next row... */
src_ptr += src_stride - width;
dst_ptr += width;
}
}
/****************************************************************************
@ -312,42 +247,35 @@ static void filter_block2d_bil_first_pass
*
* FUNCTION : Applies a 1-D 2-tap bi-linear filter to the source block
* in the vertical direction to produce the filtered output
* block. Used to implement second-pass of 2-D separable filter.
* block. Used to implement second-pass of 2-D separable
* filter.
*
* SPECIAL NOTES : Requires 32-bit input as produced by filter_block2d_bil_first_pass.
* SPECIAL NOTES : Requires 32-bit input as produced by
* filter_block2d_bil_first_pass.
* Two filter taps should sum to VP8_FILTER_WEIGHT.
*
****************************************************************************/
static void filter_block2d_bil_second_pass
(
unsigned short *src_ptr,
unsigned char *dst_ptr,
int dst_pitch,
unsigned int height,
unsigned int width,
const short *vp8_filter
)
{
unsigned int i, j;
int Temp;
static void filter_block2d_bil_second_pass(unsigned short *src_ptr,
unsigned char *dst_ptr,
int dst_pitch, unsigned int height,
unsigned int width,
const short *vp8_filter) {
unsigned int i, j;
int Temp;
for (i = 0; i < height; i++)
{
for (j = 0; j < width; j++)
{
/* Apply filter */
Temp = ((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[width] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2);
dst_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
src_ptr++;
}
/* Next row... */
dst_ptr += dst_pitch;
for (i = 0; i < height; i++) {
for (j = 0; j < width; j++) {
/* Apply filter */
Temp = ((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[width] * vp8_filter[1]) + (VP8_FILTER_WEIGHT / 2);
dst_ptr[j] = (unsigned int)(Temp >> VP8_FILTER_SHIFT);
src_ptr++;
}
}
/* Next row... */
dst_ptr += dst_pitch;
}
}
/****************************************************************************
*
@ -356,7 +284,8 @@ static void filter_block2d_bil_second_pass
* INPUTS : UINT8 *src_ptr : Pointer to source block.
* UINT32 src_pitch : Stride of source block.
* UINT32 dst_pitch : Stride of destination block.
* INT32 *HFilter : Array of 2 horizontal filter taps.
* INT32 *HFilter : Array of 2 horizontal filter
* taps.
* INT32 *VFilter : Array of 2 vertical filter taps.
* INT32 Width : Block width
* INT32 Height : Block height
@ -372,44 +301,29 @@ static void filter_block2d_bil_second_pass
* SPECIAL NOTES : The largest block size can be handled here is 16x16
*
****************************************************************************/
static void filter_block2d_bil
(
unsigned char *src_ptr,
unsigned char *dst_ptr,
unsigned int src_pitch,
unsigned int dst_pitch,
const short *HFilter,
const short *VFilter,
int Width,
int Height
)
{
static void filter_block2d_bil(unsigned char *src_ptr, unsigned char *dst_ptr,
unsigned int src_pitch, unsigned int dst_pitch,
const short *HFilter, const short *VFilter,
int Width, int Height) {
unsigned short FData[17 * 16]; /* Temp data buffer used in filtering */
unsigned short FData[17*16]; /* Temp data buffer used in filtering */
/* First filter 1-D horizontally... */
filter_block2d_bil_first_pass(src_ptr, FData, src_pitch, Height + 1, Width,
HFilter);
/* First filter 1-D horizontally... */
filter_block2d_bil_first_pass(src_ptr, FData, src_pitch, Height + 1, Width, HFilter);
/* then 1-D vertically... */
filter_block2d_bil_second_pass(FData, dst_ptr, dst_pitch, Height, Width, VFilter);
/* then 1-D vertically... */
filter_block2d_bil_second_pass(FData, dst_ptr, dst_pitch, Height, Width,
VFilter);
}
void vp8_bilinear_predict4x4_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict4x4_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
#if 0
{
int i;
@ -429,65 +343,46 @@ void vp8_bilinear_predict4x4_c
}
}
#endif
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 4, 4);
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter,
VFilter, 4, 4);
}
void vp8_bilinear_predict8x8_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict8x8_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 8);
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter,
VFilter, 8, 8);
}
void vp8_bilinear_predict8x4_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict8x4_c(unsigned char *src_ptr, int src_pixels_per_line,
int xoffset, int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 8, 4);
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter,
VFilter, 8, 4);
}
void vp8_bilinear_predict16x16_c
(
unsigned char *src_ptr,
int src_pixels_per_line,
int xoffset,
int yoffset,
unsigned char *dst_ptr,
int dst_pitch
)
{
const short *HFilter;
const short *VFilter;
void vp8_bilinear_predict16x16_c(unsigned char *src_ptr,
int src_pixels_per_line, int xoffset,
int yoffset, unsigned char *dst_ptr,
int dst_pitch) {
const short *HFilter;
const short *VFilter;
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
HFilter = vp8_bilinear_filters[xoffset];
VFilter = vp8_bilinear_filters[yoffset];
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter, 16, 16);
filter_block2d_bil(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter,
VFilter, 16, 16);
}

3
vp8/common/filter.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_FILTER_H_
#define VP8_COMMON_FILTER_H_
@ -20,7 +19,7 @@ extern "C" {
#define BLOCK_HEIGHT_WIDTH 4
#define VP8_FILTER_WEIGHT 128
#define VP8_FILTER_SHIFT 7
#define VP8_FILTER_SHIFT 7
extern DECLARE_ALIGNED(16, const short, vp8_bilinear_filters[8][2]);
extern DECLARE_ALIGNED(16, const short, vp8_sub_pel_filters[8][6]);

259
vp8/common/findnearmv.c
View File

@ -8,186 +8,147 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "findnearmv.h"
const unsigned char vp8_mbsplit_offset[4][16] = {
{ 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 2, 8, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}
{ 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 2, 8, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 }
};
/* Predict motion vectors using those from already-decoded nearby blocks.
Note that we only consider one 4x4 subblock from each candidate 16x16
macroblock. */
void vp8_find_near_mvs
(
MACROBLOCKD *xd,
const MODE_INFO *here,
int_mv *nearest,
int_mv *nearby,
int_mv *best_mv,
int cnt[4],
int refframe,
int *ref_frame_sign_bias
)
{
const MODE_INFO *above = here - xd->mode_info_stride;
const MODE_INFO *left = here - 1;
const MODE_INFO *aboveleft = above - 1;
int_mv near_mvs[4];
int_mv *mv = near_mvs;
int *cntx = cnt;
enum {CNT_INTRA, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV};
void vp8_find_near_mvs(MACROBLOCKD *xd, const MODE_INFO *here, int_mv *nearest,
int_mv *nearby, int_mv *best_mv, int cnt[4],
int refframe, int *ref_frame_sign_bias) {
const MODE_INFO *above = here - xd->mode_info_stride;
const MODE_INFO *left = here - 1;
const MODE_INFO *aboveleft = above - 1;
int_mv near_mvs[4];
int_mv *mv = near_mvs;
int *cntx = cnt;
enum { CNT_INTRA, CNT_NEAREST, CNT_NEAR, CNT_SPLITMV };
/* Zero accumulators */
mv[0].as_int = mv[1].as_int = mv[2].as_int = 0;
cnt[0] = cnt[1] = cnt[2] = cnt[3] = 0;
/* Zero accumulators */
mv[0].as_int = mv[1].as_int = mv[2].as_int = 0;
cnt[0] = cnt[1] = cnt[2] = cnt[3] = 0;
/* Process above */
if (above->mbmi.ref_frame != INTRA_FRAME)
{
if (above->mbmi.mv.as_int)
{
(++mv)->as_int = above->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, mv, ref_frame_sign_bias);
++cntx;
}
*cntx += 2;
/* Process above */
if (above->mbmi.ref_frame != INTRA_FRAME) {
if (above->mbmi.mv.as_int) {
(++mv)->as_int = above->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[above->mbmi.ref_frame], refframe, mv,
ref_frame_sign_bias);
++cntx;
}
/* Process left */
if (left->mbmi.ref_frame != INTRA_FRAME)
{
if (left->mbmi.mv.as_int)
{
int_mv this_mv;
*cntx += 2;
}
this_mv.as_int = left->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
/* Process left */
if (left->mbmi.ref_frame != INTRA_FRAME) {
if (left->mbmi.mv.as_int) {
int_mv this_mv;
if (this_mv.as_int != mv->as_int)
{
(++mv)->as_int = this_mv.as_int;
++cntx;
}
this_mv.as_int = left->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[left->mbmi.ref_frame], refframe, &this_mv,
ref_frame_sign_bias);
*cntx += 2;
}
else
cnt[CNT_INTRA] += 2;
}
if (this_mv.as_int != mv->as_int) {
(++mv)->as_int = this_mv.as_int;
++cntx;
}
/* Process above left */
if (aboveleft->mbmi.ref_frame != INTRA_FRAME)
{
if (aboveleft->mbmi.mv.as_int)
{
int_mv this_mv;
*cntx += 2;
} else
cnt[CNT_INTRA] += 2;
}
this_mv.as_int = aboveleft->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe, &this_mv, ref_frame_sign_bias);
/* Process above left */
if (aboveleft->mbmi.ref_frame != INTRA_FRAME) {
if (aboveleft->mbmi.mv.as_int) {
int_mv this_mv;
if (this_mv.as_int != mv->as_int)
{
(++mv)->as_int = this_mv.as_int;
++cntx;
}
this_mv.as_int = aboveleft->mbmi.mv.as_int;
mv_bias(ref_frame_sign_bias[aboveleft->mbmi.ref_frame], refframe,
&this_mv, ref_frame_sign_bias);
*cntx += 1;
}
else
cnt[CNT_INTRA] += 1;
}
if (this_mv.as_int != mv->as_int) {
(++mv)->as_int = this_mv.as_int;
++cntx;
}
/* If we have three distinct MV's ... */
if (cnt[CNT_SPLITMV])
{
/* See if above-left MV can be merged with NEAREST */
if (mv->as_int == near_mvs[CNT_NEAREST].as_int)
cnt[CNT_NEAREST] += 1;
}
*cntx += 1;
} else
cnt[CNT_INTRA] += 1;
}
cnt[CNT_SPLITMV] = ((above->mbmi.mode == SPLITMV)
+ (left->mbmi.mode == SPLITMV)) * 2
+ (aboveleft->mbmi.mode == SPLITMV);
/* If we have three distinct MV's ... */
if (cnt[CNT_SPLITMV]) {
/* See if above-left MV can be merged with NEAREST */
if (mv->as_int == near_mvs[CNT_NEAREST].as_int) cnt[CNT_NEAREST] += 1;
}
/* Swap near and nearest if necessary */
if (cnt[CNT_NEAR] > cnt[CNT_NEAREST])
{
int tmp;
tmp = cnt[CNT_NEAREST];
cnt[CNT_NEAREST] = cnt[CNT_NEAR];
cnt[CNT_NEAR] = tmp;
tmp = near_mvs[CNT_NEAREST].as_int;
near_mvs[CNT_NEAREST].as_int = near_mvs[CNT_NEAR].as_int;
near_mvs[CNT_NEAR].as_int = tmp;
}
cnt[CNT_SPLITMV] =
((above->mbmi.mode == SPLITMV) + (left->mbmi.mode == SPLITMV)) * 2 +
(aboveleft->mbmi.mode == SPLITMV);
/* Use near_mvs[0] to store the "best" MV */
if (cnt[CNT_NEAREST] >= cnt[CNT_INTRA])
near_mvs[CNT_INTRA] = near_mvs[CNT_NEAREST];
/* Swap near and nearest if necessary */
if (cnt[CNT_NEAR] > cnt[CNT_NEAREST]) {
int tmp;
tmp = cnt[CNT_NEAREST];
cnt[CNT_NEAREST] = cnt[CNT_NEAR];
cnt[CNT_NEAR] = tmp;
tmp = near_mvs[CNT_NEAREST].as_int;
near_mvs[CNT_NEAREST].as_int = near_mvs[CNT_NEAR].as_int;
near_mvs[CNT_NEAR].as_int = tmp;
}
/* Set up return values */
best_mv->as_int = near_mvs[0].as_int;
nearest->as_int = near_mvs[CNT_NEAREST].as_int;
nearby->as_int = near_mvs[CNT_NEAR].as_int;
/* Use near_mvs[0] to store the "best" MV */
if (cnt[CNT_NEAREST] >= cnt[CNT_INTRA])
near_mvs[CNT_INTRA] = near_mvs[CNT_NEAREST];
/* Set up return values */
best_mv->as_int = near_mvs[0].as_int;
nearest->as_int = near_mvs[CNT_NEAREST].as_int;
nearby->as_int = near_mvs[CNT_NEAR].as_int;
}
static void invert_and_clamp_mvs(int_mv *inv, int_mv *src, MACROBLOCKD *xd)
{
inv->as_mv.row = src->as_mv.row * -1;
inv->as_mv.col = src->as_mv.col * -1;
vp8_clamp_mv2(inv, xd);
vp8_clamp_mv2(src, xd);
static void invert_and_clamp_mvs(int_mv *inv, int_mv *src, MACROBLOCKD *xd) {
inv->as_mv.row = src->as_mv.row * -1;
inv->as_mv.col = src->as_mv.col * -1;
vp8_clamp_mv2(inv, xd);
vp8_clamp_mv2(src, xd);
}
int vp8_find_near_mvs_bias(MACROBLOCKD *xd, const MODE_INFO *here,
int_mv mode_mv_sb[2][MB_MODE_COUNT],
int_mv best_mv_sb[2], int cnt[4], int refframe,
int *ref_frame_sign_bias) {
int sign_bias = ref_frame_sign_bias[refframe];
int vp8_find_near_mvs_bias
(
MACROBLOCKD *xd,
const MODE_INFO *here,
int_mv mode_mv_sb[2][MB_MODE_COUNT],
int_mv best_mv_sb[2],
int cnt[4],
int refframe,
int *ref_frame_sign_bias
)
{
int sign_bias = ref_frame_sign_bias[refframe];
vp8_find_near_mvs(xd, here, &mode_mv_sb[sign_bias][NEARESTMV],
&mode_mv_sb[sign_bias][NEARMV], &best_mv_sb[sign_bias], cnt,
refframe, ref_frame_sign_bias);
vp8_find_near_mvs(xd,
here,
&mode_mv_sb[sign_bias][NEARESTMV],
&mode_mv_sb[sign_bias][NEARMV],
&best_mv_sb[sign_bias],
cnt,
refframe,
ref_frame_sign_bias);
invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARESTMV],
&mode_mv_sb[sign_bias][NEARESTMV], xd);
invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARMV],
&mode_mv_sb[sign_bias][NEARMV], xd);
invert_and_clamp_mvs(&best_mv_sb[!sign_bias], &best_mv_sb[sign_bias], xd);
invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARESTMV],
&mode_mv_sb[sign_bias][NEARESTMV], xd);
invert_and_clamp_mvs(&mode_mv_sb[!sign_bias][NEARMV],
&mode_mv_sb[sign_bias][NEARMV], xd);
invert_and_clamp_mvs(&best_mv_sb[!sign_bias],
&best_mv_sb[sign_bias], xd);
return sign_bias;
return sign_bias;
}
vp8_prob *vp8_mv_ref_probs(
vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
)
{
p[0] = vp8_mode_contexts [near_mv_ref_ct[0]] [0];
p[1] = vp8_mode_contexts [near_mv_ref_ct[1]] [1];
p[2] = vp8_mode_contexts [near_mv_ref_ct[2]] [2];
p[3] = vp8_mode_contexts [near_mv_ref_ct[3]] [3];
/*p[3] = vp8_mode_contexts [near_mv_ref_ct[1] + near_mv_ref_ct[2] + near_mv_ref_ct[3]] [3];*/
return p;
vp8_prob *vp8_mv_ref_probs(vp8_prob p[VP8_MVREFS - 1],
const int near_mv_ref_ct[4]) {
p[0] = vp8_mode_contexts[near_mv_ref_ct[0]][0];
p[1] = vp8_mode_contexts[near_mv_ref_ct[1]][1];
p[2] = vp8_mode_contexts[near_mv_ref_ct[2]][2];
p[3] = vp8_mode_contexts[near_mv_ref_ct[3]][3];
/* p[3] = vp8_mode_contexts[near_mv_ref_ct[1] + near_mv_ref_ct[2] +
near_mv_ref_ct[3]][3]; */
return p;
}

210
vp8/common/findnearmv.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_FINDNEARMV_H_
#define VP8_COMMON_FINDNEARMV_H_
@ -22,170 +21,125 @@
extern "C" {
#endif
static INLINE void mv_bias(int refmb_ref_frame_sign_bias, int refframe,
int_mv *mvp, const int *ref_frame_sign_bias)
{
if (refmb_ref_frame_sign_bias != ref_frame_sign_bias[refframe])
{
mvp->as_mv.row *= -1;
mvp->as_mv.col *= -1;
}
int_mv *mvp, const int *ref_frame_sign_bias) {
if (refmb_ref_frame_sign_bias != ref_frame_sign_bias[refframe]) {
mvp->as_mv.row *= -1;
mvp->as_mv.col *= -1;
}
}
#define LEFT_TOP_MARGIN (16 << 3)
#define RIGHT_BOTTOM_MARGIN (16 << 3)
static INLINE void vp8_clamp_mv2(int_mv *mv, const MACROBLOCKD *xd)
{
if (mv->as_mv.col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN))
mv->as_mv.col = xd->mb_to_left_edge - LEFT_TOP_MARGIN;
else if (mv->as_mv.col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)
mv->as_mv.col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN;
static INLINE void vp8_clamp_mv2(int_mv *mv, const MACROBLOCKD *xd) {
if (mv->as_mv.col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN))
mv->as_mv.col = xd->mb_to_left_edge - LEFT_TOP_MARGIN;
else if (mv->as_mv.col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)
mv->as_mv.col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN;
if (mv->as_mv.row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN))
mv->as_mv.row = xd->mb_to_top_edge - LEFT_TOP_MARGIN;
else if (mv->as_mv.row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)
mv->as_mv.row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN;
if (mv->as_mv.row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN))
mv->as_mv.row = xd->mb_to_top_edge - LEFT_TOP_MARGIN;
else if (mv->as_mv.row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)
mv->as_mv.row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN;
}
static INLINE void vp8_clamp_mv(int_mv *mv, int mb_to_left_edge,
int mb_to_right_edge, int mb_to_top_edge,
int mb_to_bottom_edge)
{
mv->as_mv.col = (mv->as_mv.col < mb_to_left_edge) ?
mb_to_left_edge : mv->as_mv.col;
mv->as_mv.col = (mv->as_mv.col > mb_to_right_edge) ?
mb_to_right_edge : mv->as_mv.col;
mv->as_mv.row = (mv->as_mv.row < mb_to_top_edge) ?
mb_to_top_edge : mv->as_mv.row;
mv->as_mv.row = (mv->as_mv.row > mb_to_bottom_edge) ?
mb_to_bottom_edge : mv->as_mv.row;
int mb_to_bottom_edge) {
mv->as_mv.col =
(mv->as_mv.col < mb_to_left_edge) ? mb_to_left_edge : mv->as_mv.col;
mv->as_mv.col =
(mv->as_mv.col > mb_to_right_edge) ? mb_to_right_edge : mv->as_mv.col;
mv->as_mv.row =
(mv->as_mv.row < mb_to_top_edge) ? mb_to_top_edge : mv->as_mv.row;
mv->as_mv.row =
(mv->as_mv.row > mb_to_bottom_edge) ? mb_to_bottom_edge : mv->as_mv.row;
}
static INLINE unsigned int vp8_check_mv_bounds(int_mv *mv, int mb_to_left_edge,
int mb_to_right_edge,
int mb_to_top_edge,
int mb_to_bottom_edge)
{
unsigned int need_to_clamp;
need_to_clamp = (mv->as_mv.col < mb_to_left_edge);
need_to_clamp |= (mv->as_mv.col > mb_to_right_edge);
need_to_clamp |= (mv->as_mv.row < mb_to_top_edge);
need_to_clamp |= (mv->as_mv.row > mb_to_bottom_edge);
return need_to_clamp;
int mb_to_bottom_edge) {
unsigned int need_to_clamp;
need_to_clamp = (mv->as_mv.col < mb_to_left_edge);
need_to_clamp |= (mv->as_mv.col > mb_to_right_edge);
need_to_clamp |= (mv->as_mv.row < mb_to_top_edge);
need_to_clamp |= (mv->as_mv.row > mb_to_bottom_edge);
return need_to_clamp;
}
void vp8_find_near_mvs
(
MACROBLOCKD *xd,
const MODE_INFO *here,
int_mv *nearest, int_mv *nearby, int_mv *best,
int near_mv_ref_cts[4],
int refframe,
int *ref_frame_sign_bias
);
void vp8_find_near_mvs(MACROBLOCKD *xd, const MODE_INFO *here, int_mv *nearest,
int_mv *nearby, int_mv *best, int near_mv_ref_cts[4],
int refframe, int *ref_frame_sign_bias);
int vp8_find_near_mvs_bias(MACROBLOCKD *xd, const MODE_INFO *here,
int_mv mode_mv_sb[2][MB_MODE_COUNT],
int_mv best_mv_sb[2], int cnt[4], int refframe,
int *ref_frame_sign_bias);
int vp8_find_near_mvs_bias
(
MACROBLOCKD *xd,
const MODE_INFO *here,
int_mv mode_mv_sb[2][MB_MODE_COUNT],
int_mv best_mv_sb[2],
int cnt[4],
int refframe,
int *ref_frame_sign_bias
);
vp8_prob *vp8_mv_ref_probs(
vp8_prob p[VP8_MVREFS-1], const int near_mv_ref_ct[4]
);
vp8_prob *vp8_mv_ref_probs(vp8_prob p[VP8_MVREFS - 1],
const int near_mv_ref_ct[4]);
extern const unsigned char vp8_mbsplit_offset[4][16];
static INLINE uint32_t left_block_mv(const MODE_INFO *cur_mb, int b) {
if (!(b & 3)) {
/* On L edge, get from MB to left of us */
--cur_mb;
static INLINE uint32_t left_block_mv(const MODE_INFO *cur_mb, int b)
{
if (!(b & 3))
{
/* On L edge, get from MB to left of us */
--cur_mb;
if (cur_mb->mbmi.mode != SPLITMV) return cur_mb->mbmi.mv.as_int;
b += 4;
}
if(cur_mb->mbmi.mode != SPLITMV)
return cur_mb->mbmi.mv.as_int;
b += 4;
}
return (cur_mb->bmi + b - 1)->mv.as_int;
return (cur_mb->bmi + b - 1)->mv.as_int;
}
static INLINE uint32_t above_block_mv(const MODE_INFO *cur_mb, int b,
int mi_stride)
{
if (!(b >> 2))
{
/* On top edge, get from MB above us */
cur_mb -= mi_stride;
int mi_stride) {
if (!(b >> 2)) {
/* On top edge, get from MB above us */
cur_mb -= mi_stride;
if(cur_mb->mbmi.mode != SPLITMV)
return cur_mb->mbmi.mv.as_int;
b += 16;
}
if (cur_mb->mbmi.mode != SPLITMV) return cur_mb->mbmi.mv.as_int;
b += 16;
}
return (cur_mb->bmi + (b - 4))->mv.as_int;
return (cur_mb->bmi + (b - 4))->mv.as_int;
}
static INLINE B_PREDICTION_MODE left_block_mode(const MODE_INFO *cur_mb, int b)
{
if (!(b & 3))
{
/* On L edge, get from MB to left of us */
--cur_mb;
switch (cur_mb->mbmi.mode)
{
case B_PRED:
return (cur_mb->bmi + b + 3)->as_mode;
case DC_PRED:
return B_DC_PRED;
case V_PRED:
return B_VE_PRED;
case H_PRED:
return B_HE_PRED;
case TM_PRED:
return B_TM_PRED;
default:
return B_DC_PRED;
}
static INLINE B_PREDICTION_MODE left_block_mode(const MODE_INFO *cur_mb,
int b) {
if (!(b & 3)) {
/* On L edge, get from MB to left of us */
--cur_mb;
switch (cur_mb->mbmi.mode) {
case B_PRED: return (cur_mb->bmi + b + 3)->as_mode;
case DC_PRED: return B_DC_PRED;
case V_PRED: return B_VE_PRED;
case H_PRED: return B_HE_PRED;
case TM_PRED: return B_TM_PRED;
default: return B_DC_PRED;
}
}
return (cur_mb->bmi + b - 1)->as_mode;
return (cur_mb->bmi + b - 1)->as_mode;
}
static INLINE B_PREDICTION_MODE above_block_mode(const MODE_INFO *cur_mb, int b,
int mi_stride)
{
if (!(b >> 2))
{
/* On top edge, get from MB above us */
cur_mb -= mi_stride;
int mi_stride) {
if (!(b >> 2)) {
/* On top edge, get from MB above us */
cur_mb -= mi_stride;
switch (cur_mb->mbmi.mode)
{
case B_PRED:
return (cur_mb->bmi + b + 12)->as_mode;
case DC_PRED:
return B_DC_PRED;
case V_PRED:
return B_VE_PRED;
case H_PRED:
return B_HE_PRED;
case TM_PRED:
return B_TM_PRED;
default:
return B_DC_PRED;
}
switch (cur_mb->mbmi.mode) {
case B_PRED: return (cur_mb->bmi + b + 12)->as_mode;
case DC_PRED: return B_DC_PRED;
case V_PRED: return B_VE_PRED;
case H_PRED: return B_HE_PRED;
case TM_PRED: return B_TM_PRED;
default: return B_DC_PRED;
}
}
return (cur_mb->bmi + b - 4)->as_mode;
return (cur_mb->bmi + b - 4)->as_mode;
}
#ifdef __cplusplus

78
vp8/common/generic/systemdependent.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#if ARCH_ARM
@ -24,7 +23,7 @@
#include <unistd.h>
#elif defined(_WIN32)
#include <windows.h>
typedef void (WINAPI *PGNSI)(LPSYSTEM_INFO);
typedef void(WINAPI *PGNSI)(LPSYSTEM_INFO);
#elif defined(__OS2__)
#define INCL_DOS
#define INCL_DOSSPINLOCK
@ -33,74 +32,69 @@ typedef void (WINAPI *PGNSI)(LPSYSTEM_INFO);
#endif
#if CONFIG_MULTITHREAD
static int get_cpu_count()
{
int core_count = 16;
static int get_cpu_count() {
int core_count = 16;
#if HAVE_UNISTD_H && !defined(__OS2__)
#if defined(_SC_NPROCESSORS_ONLN)
core_count = sysconf(_SC_NPROCESSORS_ONLN);
core_count = sysconf(_SC_NPROCESSORS_ONLN);
#elif defined(_SC_NPROC_ONLN)
core_count = sysconf(_SC_NPROC_ONLN);
core_count = sysconf(_SC_NPROC_ONLN);
#endif
#elif defined(_WIN32)
{
{
#if _WIN32_WINNT >= 0x0501
SYSTEM_INFO sysinfo;
GetNativeSystemInfo(&sysinfo);
SYSTEM_INFO sysinfo;
GetNativeSystemInfo(&sysinfo);
#else
PGNSI pGNSI;
SYSTEM_INFO sysinfo;
PGNSI pGNSI;
SYSTEM_INFO sysinfo;
/* Call GetNativeSystemInfo if supported or
* GetSystemInfo otherwise. */
/* Call GetNativeSystemInfo if supported or
* GetSystemInfo otherwise. */
pGNSI = (PGNSI) GetProcAddress(
GetModuleHandle(TEXT("kernel32.dll")), "GetNativeSystemInfo");
if (pGNSI != NULL)
pGNSI(&sysinfo);
else
GetSystemInfo(&sysinfo);
pGNSI = (PGNSI)GetProcAddress(GetModuleHandle(TEXT("kernel32.dll")),
"GetNativeSystemInfo");
if (pGNSI != NULL)
pGNSI(&sysinfo);
else
GetSystemInfo(&sysinfo);
#endif
core_count = sysinfo.dwNumberOfProcessors;
}
core_count = sysinfo.dwNumberOfProcessors;
}
#elif defined(__OS2__)
{
ULONG proc_id;
ULONG status;
{
ULONG proc_id;
ULONG status;
core_count = 0;
for (proc_id = 1; ; proc_id++)
{
if (DosGetProcessorStatus(proc_id, &status))
break;
core_count = 0;
for (proc_id = 1;; proc_id++) {
if (DosGetProcessorStatus(proc_id, &status)) break;
if (status == PROC_ONLINE)
core_count++;
}
if (status == PROC_ONLINE) core_count++;
}
}
#else
/* other platforms */
/* other platforms */
#endif
return core_count > 0 ? core_count : 1;
return core_count > 0 ? core_count : 1;
}
#endif
void vp8_clear_system_state_c() {};
void vp8_clear_system_state_c(){};
void vp8_machine_specific_config(VP8_COMMON *ctx)
{
void vp8_machine_specific_config(VP8_COMMON *ctx) {
#if CONFIG_MULTITHREAD
ctx->processor_core_count = get_cpu_count();
ctx->processor_core_count = get_cpu_count();
#else
(void)ctx;
(void)ctx;
#endif /* CONFIG_MULTITHREAD */
#if ARCH_ARM
ctx->cpu_caps = arm_cpu_caps();
ctx->cpu_caps = arm_cpu_caps();
#elif ARCH_X86 || ARCH_X86_64
ctx->cpu_caps = x86_simd_caps();
ctx->cpu_caps = x86_simd_caps();
#endif
}

25
vp8/common/header.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_HEADER_H_
#define VP8_COMMON_HEADER_H_
@ -17,22 +16,21 @@ extern "C" {
#endif
/* 24 bits total */
typedef struct
{
unsigned int type: 1;
unsigned int version: 3;
unsigned int show_frame: 1;
typedef struct {
unsigned int type : 1;
unsigned int version : 3;
unsigned int show_frame : 1;
/* Allow 2^20 bytes = 8 megabits for first partition */
/* Allow 2^20 bytes = 8 megabits for first partition */
unsigned int first_partition_length_in_bytes: 19;
unsigned int first_partition_length_in_bytes : 19;
#ifdef PACKET_TESTING
unsigned int frame_number;
unsigned int update_gold: 1;
unsigned int uses_gold: 1;
unsigned int update_last: 1;
unsigned int uses_last: 1;
unsigned int frame_number;
unsigned int update_gold : 1;
unsigned int uses_gold : 1;
unsigned int update_last : 1;
unsigned int uses_last : 1;
#endif
} VP8_HEADER;
@ -43,7 +41,6 @@ typedef struct
#define VP8_HEADER_SIZE 3
#endif
#ifdef __cplusplus
} // extern "C"
#endif

108
vp8/common/idct_blk.c
View File

@ -12,79 +12,67 @@
#include "vp8_rtcd.h"
#include "vpx_mem/vpx_mem.h"
void vp8_dequant_idct_add_c(short *input, short *dq,
unsigned char *dest, int stride);
void vp8_dc_only_idct_add_c(short input_dc, unsigned char * pred,
void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *dest,
int stride);
void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred,
int pred_stride, unsigned char *dst_ptr,
int dst_stride);
void vp8_dequant_idct_add_y_block_c
(short *q, short *dq,
unsigned char *dst, int stride, char *eobs)
{
int i, j;
void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i, j;
for (i = 0; i < 4; i++)
{
for (j = 0; j < 4; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_c (q, dq, dst, stride);
else
{
vp8_dc_only_idct_add_c (q[0]*dq[0], dst, stride, dst, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
for (i = 0; i < 4; i++) {
for (j = 0; j < 4; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_c(q, dq, dst, stride);
else {
vp8_dc_only_idct_add_c(q[0] * dq[0], dst, stride, dst, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
q += 16;
dst += 4;
}
dst += 4*stride - 16;
q += 16;
dst += 4;
}
dst += 4 * stride - 16;
}
}
void vp8_dequant_idct_add_uv_block_c
(short *q, short *dq,
unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
{
int i, j;
void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dstu,
unsigned char *dstv, int stride,
char *eobs) {
int i, j;
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_c (q, dq, dstu, stride);
else
{
vp8_dc_only_idct_add_c (q[0]*dq[0], dstu, stride, dstu, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_c(q, dq, dstu, stride);
else {
vp8_dc_only_idct_add_c(q[0] * dq[0], dstu, stride, dstu, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
q += 16;
dstu += 4;
}
dstu += 4*stride - 8;
q += 16;
dstu += 4;
}
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_c (q, dq, dstv, stride);
else
{
vp8_dc_only_idct_add_c (q[0]*dq[0], dstv, stride, dstv, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
dstu += 4 * stride - 8;
}
q += 16;
dstv += 4;
}
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_c(q, dq, dstv, stride);
else {
vp8_dc_only_idct_add_c(q[0] * dq[0], dstv, stride, dstv, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
dstv += 4*stride - 8;
q += 16;
dstv += 4;
}
dstv += 4 * stride - 8;
}
}

270
vp8/common/idctllm.c
View File

@ -24,182 +24,162 @@
* x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
**************************************************************************/
static const int cospi8sqrt2minus1 = 20091;
static const int sinpi8sqrt2 = 35468;
static const int sinpi8sqrt2 = 35468;
void vp8_short_idct4x4llm_c(short *input, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride)
{
int i;
int r, c;
int a1, b1, c1, d1;
short output[16];
short *ip = input;
short *op = output;
int temp1, temp2;
int shortpitch = 4;
int dst_stride) {
int i;
int r, c;
int a1, b1, c1, d1;
short output[16];
short *ip = input;
short *op = output;
int temp1, temp2;
int shortpitch = 4;
for (i = 0; i < 4; i++)
{
a1 = ip[0] + ip[8];
b1 = ip[0] - ip[8];
for (i = 0; i < 4; i++) {
a1 = ip[0] + ip[8];
b1 = ip[0] - ip[8];
temp1 = (ip[4] * sinpi8sqrt2) >> 16;
temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = (ip[4] * sinpi8sqrt2) >> 16;
temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[12] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[12] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
op[shortpitch*0] = a1 + d1;
op[shortpitch*3] = a1 - d1;
op[shortpitch * 0] = a1 + d1;
op[shortpitch * 3] = a1 - d1;
op[shortpitch*1] = b1 + c1;
op[shortpitch*2] = b1 - c1;
op[shortpitch * 1] = b1 + c1;
op[shortpitch * 2] = b1 - c1;
ip++;
op++;
}
ip = output;
op = output;
for (i = 0; i < 4; i++)
{
a1 = ip[0] + ip[2];
b1 = ip[0] - ip[2];
temp1 = (ip[1] * sinpi8sqrt2) >> 16;
temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[3] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
op[0] = (a1 + d1 + 4) >> 3;
op[3] = (a1 - d1 + 4) >> 3;
op[1] = (b1 + c1 + 4) >> 3;
op[2] = (b1 - c1 + 4) >> 3;
ip += shortpitch;
op += shortpitch;
}
ip = output;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
int a = ip[c] + pred_ptr[c] ;
if (a < 0)
a = 0;
if (a > 255)
a = 255;
dst_ptr[c] = (unsigned char) a ;
}
ip += 4;
dst_ptr += dst_stride;
pred_ptr += pred_stride;
ip++;
op++;
}
ip = output;
op = output;
for (i = 0; i < 4; i++) {
a1 = ip[0] + ip[2];
b1 = ip[0] - ip[2];
temp1 = (ip[1] * sinpi8sqrt2) >> 16;
temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[3] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
op[0] = (a1 + d1 + 4) >> 3;
op[3] = (a1 - d1 + 4) >> 3;
op[1] = (b1 + c1 + 4) >> 3;
op[2] = (b1 - c1 + 4) >> 3;
ip += shortpitch;
op += shortpitch;
}
ip = output;
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++) {
int a = ip[c] + pred_ptr[c];
if (a < 0) a = 0;
if (a > 255) a = 255;
dst_ptr[c] = (unsigned char)a;
}
ip += 4;
dst_ptr += dst_stride;
pred_ptr += pred_stride;
}
}
void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride)
{
int a1 = ((input_dc + 4) >> 3);
int r, c;
int dst_stride) {
int a1 = ((input_dc + 4) >> 3);
int r, c;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
int a = a1 + pred_ptr[c] ;
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++) {
int a = a1 + pred_ptr[c];
if (a < 0)
a = 0;
if (a < 0) a = 0;
if (a > 255)
a = 255;
if (a > 255) a = 255;
dst_ptr[c] = (unsigned char) a ;
}
dst_ptr += dst_stride;
pred_ptr += pred_stride;
dst_ptr[c] = (unsigned char)a;
}
dst_ptr += dst_stride;
pred_ptr += pred_stride;
}
}
void vp8_short_inv_walsh4x4_c(short *input, short *mb_dqcoeff)
{
short output[16];
int i;
int a1, b1, c1, d1;
int a2, b2, c2, d2;
short *ip = input;
short *op = output;
void vp8_short_inv_walsh4x4_c(short *input, short *mb_dqcoeff) {
short output[16];
int i;
int a1, b1, c1, d1;
int a2, b2, c2, d2;
short *ip = input;
short *op = output;
for (i = 0; i < 4; i++)
{
a1 = ip[0] + ip[12];
b1 = ip[4] + ip[8];
c1 = ip[4] - ip[8];
d1 = ip[0] - ip[12];
for (i = 0; i < 4; i++) {
a1 = ip[0] + ip[12];
b1 = ip[4] + ip[8];
c1 = ip[4] - ip[8];
d1 = ip[0] - ip[12];
op[0] = a1 + b1;
op[4] = c1 + d1;
op[8] = a1 - b1;
op[12] = d1 - c1;
ip++;
op++;
}
op[0] = a1 + b1;
op[4] = c1 + d1;
op[8] = a1 - b1;
op[12] = d1 - c1;
ip++;
op++;
}
ip = output;
op = output;
ip = output;
op = output;
for (i = 0; i < 4; i++)
{
a1 = ip[0] + ip[3];
b1 = ip[1] + ip[2];
c1 = ip[1] - ip[2];
d1 = ip[0] - ip[3];
for (i = 0; i < 4; i++) {
a1 = ip[0] + ip[3];
b1 = ip[1] + ip[2];
c1 = ip[1] - ip[2];
d1 = ip[0] - ip[3];
a2 = a1 + b1;
b2 = c1 + d1;
c2 = a1 - b1;
d2 = d1 - c1;
a2 = a1 + b1;
b2 = c1 + d1;
c2 = a1 - b1;
d2 = d1 - c1;
op[0] = (a2 + 3) >> 3;
op[1] = (b2 + 3) >> 3;
op[2] = (c2 + 3) >> 3;
op[3] = (d2 + 3) >> 3;
op[0] = (a2 + 3) >> 3;
op[1] = (b2 + 3) >> 3;
op[2] = (c2 + 3) >> 3;
op[3] = (d2 + 3) >> 3;
ip += 4;
op += 4;
}
ip += 4;
op += 4;
}
for(i = 0; i < 16; i++)
{
mb_dqcoeff[i * 16] = output[i];
}
for (i = 0; i < 16; i++) {
mb_dqcoeff[i * 16] = output[i];
}
}
void vp8_short_inv_walsh4x4_1_c(short *input, short *mb_dqcoeff)
{
int i;
int a1;
void vp8_short_inv_walsh4x4_1_c(short *input, short *mb_dqcoeff) {
int i;
int a1;
a1 = ((input[0] + 3) >> 3);
for(i = 0; i < 16; i++)
{
mb_dqcoeff[i * 16] = a1;
}
a1 = ((input[0] + 3) >> 3);
for (i = 0; i < 16; i++) {
mb_dqcoeff[i * 16] = a1;
}
}

55
vp8/common/invtrans.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_INVTRANS_H_
#define VP8_COMMON_INVTRANS_H_
@ -25,43 +24,31 @@
extern "C" {
#endif
static void eob_adjust(char *eobs, short *diff)
{
/* eob adjust.... the idct can only skip if both the dc and eob are zero */
int js;
for(js = 0; js < 16; js++)
{
if((eobs[js] == 0) && (diff[0] != 0))
eobs[js]++;
diff+=16;
}
static void eob_adjust(char *eobs, short *diff) {
/* eob adjust.... the idct can only skip if both the dc and eob are zero */
int js;
for (js = 0; js < 16; js++) {
if ((eobs[js] == 0) && (diff[0] != 0)) eobs[js]++;
diff += 16;
}
}
static INLINE void vp8_inverse_transform_mby(MACROBLOCKD *xd)
{
short *DQC = xd->dequant_y1;
static INLINE void vp8_inverse_transform_mby(MACROBLOCKD *xd) {
short *DQC = xd->dequant_y1;
if (xd->mode_info_context->mbmi.mode != SPLITMV)
{
/* do 2nd order transform on the dc block */
if (xd->eobs[24] > 1)
{
vp8_short_inv_walsh4x4
(&xd->block[24].dqcoeff[0], xd->qcoeff);
}
else
{
vp8_short_inv_walsh4x4_1
(&xd->block[24].dqcoeff[0], xd->qcoeff);
}
eob_adjust(xd->eobs, xd->qcoeff);
DQC = xd->dequant_y1_dc;
if (xd->mode_info_context->mbmi.mode != SPLITMV) {
/* do 2nd order transform on the dc block */
if (xd->eobs[24] > 1) {
vp8_short_inv_walsh4x4(&xd->block[24].dqcoeff[0], xd->qcoeff);
} else {
vp8_short_inv_walsh4x4_1(&xd->block[24].dqcoeff[0], xd->qcoeff);
}
vp8_dequant_idct_add_y_block
(xd->qcoeff, DQC,
xd->dst.y_buffer,
xd->dst.y_stride, xd->eobs);
eob_adjust(xd->eobs, xd->qcoeff);
DQC = xd->dequant_y1_dc;
}
vp8_dequant_idct_add_y_block(xd->qcoeff, DQC, xd->dst.y_buffer,
xd->dst.y_stride, xd->eobs);
}
#ifdef __cplusplus
} // extern "C"

72
vp8/common/loopfilter.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_LOOPFILTER_H_
#define VP8_COMMON_LOOPFILTER_H_
@ -20,16 +19,12 @@
extern "C" {
#endif
#define MAX_LOOP_FILTER 63
#define MAX_LOOP_FILTER 63
/* fraction of total macroblock rows to be used in fast filter level picking */
/* has to be > 2 */
#define PARTIAL_FRAME_FRACTION 8
#define PARTIAL_FRAME_FRACTION 8
typedef enum
{
NORMAL_LOOPFILTER = 0,
SIMPLE_LOOPFILTER = 1
} LOOPFILTERTYPE;
typedef enum { NORMAL_LOOPFILTER = 0, SIMPLE_LOOPFILTER = 1 } LOOPFILTERTYPE;
#if ARCH_ARM
#define SIMD_WIDTH 1
@ -40,35 +35,32 @@ typedef enum
/* Need to align this structure so when it is declared and
* passed it can be loaded into vector registers.
*/
typedef struct
{
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, mblim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, blim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, lim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, hev_thr[4][SIMD_WIDTH]);
unsigned char lvl[4][4][4];
unsigned char hev_thr_lut[2][MAX_LOOP_FILTER + 1];
unsigned char mode_lf_lut[10];
typedef struct {
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char,
mblim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char,
blim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char,
lim[MAX_LOOP_FILTER + 1][SIMD_WIDTH]);
DECLARE_ALIGNED(SIMD_WIDTH, unsigned char, hev_thr[4][SIMD_WIDTH]);
unsigned char lvl[4][4][4];
unsigned char hev_thr_lut[2][MAX_LOOP_FILTER + 1];
unsigned char mode_lf_lut[10];
} loop_filter_info_n;
typedef struct loop_filter_info
{
const unsigned char * mblim;
const unsigned char * blim;
const unsigned char * lim;
const unsigned char * hev_thr;
typedef struct loop_filter_info {
const unsigned char *mblim;
const unsigned char *blim;
const unsigned char *lim;
const unsigned char *hev_thr;
} loop_filter_info;
typedef void loop_filter_uvfunction
(
unsigned char *u, /* source pointer */
int p, /* pitch */
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
unsigned char *v
);
typedef void loop_filter_uvfunction(unsigned char *u, /* source pointer */
int p, /* pitch */
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
unsigned char *v);
/* assorted loopfilter functions which get used elsewhere */
struct VP8Common;
@ -77,8 +69,7 @@ struct modeinfo;
void vp8_loop_filter_init(struct VP8Common *cm);
void vp8_loop_filter_frame_init(struct VP8Common *cm,
struct macroblockd *mbd,
void vp8_loop_filter_frame_init(struct VP8Common *cm, struct macroblockd *mbd,
int default_filt_lvl);
void vp8_loop_filter_frame(struct VP8Common *cm, struct macroblockd *mbd,
@ -88,22 +79,21 @@ void vp8_loop_filter_partial_frame(struct VP8Common *cm,
struct macroblockd *mbd,
int default_filt_lvl);
void vp8_loop_filter_frame_yonly(struct VP8Common *cm,
struct macroblockd *mbd,
void vp8_loop_filter_frame_yonly(struct VP8Common *cm, struct macroblockd *mbd,
int default_filt_lvl);
void vp8_loop_filter_update_sharpness(loop_filter_info_n *lfi,
int sharpness_lvl);
void vp8_loop_filter_row_normal(struct VP8Common *cm,
struct modeinfo *mode_info_context,
int mb_row, int post_ystride, int post_uvstride,
struct modeinfo *mode_info_context, int mb_row,
int post_ystride, int post_uvstride,
unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr);
void vp8_loop_filter_row_simple(struct VP8Common *cm,
struct modeinfo *mode_info_context,
int mb_row, int post_ystride, int post_uvstride,
struct modeinfo *mode_info_context, int mb_row,
int post_ystride, int post_uvstride,
unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr);
#ifdef __cplusplus

589
vp8/common/loopfilter_filters.c
View File

@ -8,423 +8,374 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include <stdlib.h>
#include "loopfilter.h"
#include "onyxc_int.h"
typedef unsigned char uc;
static signed char vp8_signed_char_clamp(int t)
{
t = (t < -128 ? -128 : t);
t = (t > 127 ? 127 : t);
return (signed char) t;
static signed char vp8_signed_char_clamp(int t) {
t = (t < -128 ? -128 : t);
t = (t > 127 ? 127 : t);
return (signed char)t;
}
/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
static signed char vp8_filter_mask(uc limit, uc blimit,
uc p3, uc p2, uc p1, uc p0,
uc q0, uc q1, uc q2, uc q3)
{
signed char mask = 0;
mask |= (abs(p3 - p2) > limit);
mask |= (abs(p2 - p1) > limit);
mask |= (abs(p1 - p0) > limit);
mask |= (abs(q1 - q0) > limit);
mask |= (abs(q2 - q1) > limit);
mask |= (abs(q3 - q2) > limit);
mask |= (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > blimit);
return mask - 1;
static signed char vp8_filter_mask(uc limit, uc blimit, uc p3, uc p2, uc p1,
uc p0, uc q0, uc q1, uc q2, uc q3) {
signed char mask = 0;
mask |= (abs(p3 - p2) > limit);
mask |= (abs(p2 - p1) > limit);
mask |= (abs(p1 - p0) > limit);
mask |= (abs(q1 - q0) > limit);
mask |= (abs(q2 - q1) > limit);
mask |= (abs(q3 - q2) > limit);
mask |= (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 > blimit);
return mask - 1;
}
/* is there high variance internal edge ( 11111111 yes, 00000000 no) */
static signed char vp8_hevmask(uc thresh, uc p1, uc p0, uc q0, uc q1)
{
signed char hev = 0;
hev |= (abs(p1 - p0) > thresh) * -1;
hev |= (abs(q1 - q0) > thresh) * -1;
return hev;
static signed char vp8_hevmask(uc thresh, uc p1, uc p0, uc q0, uc q1) {
signed char hev = 0;
hev |= (abs(p1 - p0) > thresh) * -1;
hev |= (abs(q1 - q0) > thresh) * -1;
return hev;
}
static void vp8_filter(signed char mask, uc hev, uc *op1,
uc *op0, uc *oq0, uc *oq1)
static void vp8_filter(signed char mask, uc hev, uc *op1, uc *op0, uc *oq0,
uc *oq1) {
signed char ps0, qs0;
signed char ps1, qs1;
signed char filter_value, Filter1, Filter2;
signed char u;
{
signed char ps0, qs0;
signed char ps1, qs1;
signed char filter_value, Filter1, Filter2;
signed char u;
ps1 = (signed char)*op1 ^ 0x80;
ps0 = (signed char)*op0 ^ 0x80;
qs0 = (signed char)*oq0 ^ 0x80;
qs1 = (signed char)*oq1 ^ 0x80;
ps1 = (signed char) * op1 ^ 0x80;
ps0 = (signed char) * op0 ^ 0x80;
qs0 = (signed char) * oq0 ^ 0x80;
qs1 = (signed char) * oq1 ^ 0x80;
/* add outer taps if we have high edge variance */
filter_value = vp8_signed_char_clamp(ps1 - qs1);
filter_value &= hev;
/* add outer taps if we have high edge variance */
filter_value = vp8_signed_char_clamp(ps1 - qs1);
filter_value &= hev;
/* inner taps */
filter_value = vp8_signed_char_clamp(filter_value + 3 * (qs0 - ps0));
filter_value &= mask;
/* inner taps */
filter_value = vp8_signed_char_clamp(filter_value + 3 * (qs0 - ps0));
filter_value &= mask;
/* save bottom 3 bits so that we round one side +4 and the other +3
* if it equals 4 we'll set to adjust by -1 to account for the fact
* we'd round 3 the other way
*/
Filter1 = vp8_signed_char_clamp(filter_value + 4);
Filter2 = vp8_signed_char_clamp(filter_value + 3);
Filter1 >>= 3;
Filter2 >>= 3;
u = vp8_signed_char_clamp(qs0 - Filter1);
*oq0 = u ^ 0x80;
u = vp8_signed_char_clamp(ps0 + Filter2);
*op0 = u ^ 0x80;
filter_value = Filter1;
/* save bottom 3 bits so that we round one side +4 and the other +3
* if it equals 4 we'll set to adjust by -1 to account for the fact
* we'd round 3 the other way
*/
Filter1 = vp8_signed_char_clamp(filter_value + 4);
Filter2 = vp8_signed_char_clamp(filter_value + 3);
Filter1 >>= 3;
Filter2 >>= 3;
u = vp8_signed_char_clamp(qs0 - Filter1);
*oq0 = u ^ 0x80;
u = vp8_signed_char_clamp(ps0 + Filter2);
*op0 = u ^ 0x80;
filter_value = Filter1;
/* outer tap adjustments */
filter_value += 1;
filter_value >>= 1;
filter_value &= ~hev;
u = vp8_signed_char_clamp(qs1 - filter_value);
*oq1 = u ^ 0x80;
u = vp8_signed_char_clamp(ps1 + filter_value);
*op1 = u ^ 0x80;
/* outer tap adjustments */
filter_value += 1;
filter_value >>= 1;
filter_value &= ~hev;
u = vp8_signed_char_clamp(qs1 - filter_value);
*oq1 = u ^ 0x80;
u = vp8_signed_char_clamp(ps1 + filter_value);
*op1 = u ^ 0x80;
}
void vp8_loop_filter_horizontal_edge_c
(
unsigned char *s,
int p, /* pitch */
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
int count
)
{
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
void vp8_loop_filter_horizontal_edge_c(unsigned char *s, int p, /* pitch */
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh, int count) {
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
mask = vp8_filter_mask(limit[0], blimit[0],
s[-4*p], s[-3*p], s[-2*p], s[-1*p],
s[0*p], s[1*p], s[2*p], s[3*p]);
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do {
mask = vp8_filter_mask(limit[0], blimit[0], s[-4 * p], s[-3 * p], s[-2 * p],
s[-1 * p], s[0 * p], s[1 * p], s[2 * p], s[3 * p]);
hev = vp8_hevmask(thresh[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
hev = vp8_hevmask(thresh[0], s[-2 * p], s[-1 * p], s[0 * p], s[1 * p]);
vp8_filter(mask, hev, s - 2 * p, s - 1 * p, s, s + 1 * p);
vp8_filter(mask, hev, s - 2 * p, s - 1 * p, s, s + 1 * p);
++s;
}
while (++i < count * 8);
++s;
} while (++i < count * 8);
}
void vp8_loop_filter_vertical_edge_c
(
unsigned char *s,
int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
int count
)
{
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
void vp8_loop_filter_vertical_edge_c(unsigned char *s, int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh, int count) {
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
mask = vp8_filter_mask(limit[0], blimit[0],
s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do {
mask = vp8_filter_mask(limit[0], blimit[0], s[-4], s[-3], s[-2], s[-1],
s[0], s[1], s[2], s[3]);
hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
vp8_filter(mask, hev, s - 2, s - 1, s, s + 1);
vp8_filter(mask, hev, s - 2, s - 1, s, s + 1);
s += p;
}
while (++i < count * 8);
s += p;
} while (++i < count * 8);
}
static void vp8_mbfilter(signed char mask, uc hev,
uc *op2, uc *op1, uc *op0, uc *oq0, uc *oq1, uc *oq2)
{
signed char s, u;
signed char filter_value, Filter1, Filter2;
signed char ps2 = (signed char) * op2 ^ 0x80;
signed char ps1 = (signed char) * op1 ^ 0x80;
signed char ps0 = (signed char) * op0 ^ 0x80;
signed char qs0 = (signed char) * oq0 ^ 0x80;
signed char qs1 = (signed char) * oq1 ^ 0x80;
signed char qs2 = (signed char) * oq2 ^ 0x80;
static void vp8_mbfilter(signed char mask, uc hev, uc *op2, uc *op1, uc *op0,
uc *oq0, uc *oq1, uc *oq2) {
signed char s, u;
signed char filter_value, Filter1, Filter2;
signed char ps2 = (signed char)*op2 ^ 0x80;
signed char ps1 = (signed char)*op1 ^ 0x80;
signed char ps0 = (signed char)*op0 ^ 0x80;
signed char qs0 = (signed char)*oq0 ^ 0x80;
signed char qs1 = (signed char)*oq1 ^ 0x80;
signed char qs2 = (signed char)*oq2 ^ 0x80;
/* add outer taps if we have high edge variance */
filter_value = vp8_signed_char_clamp(ps1 - qs1);
filter_value = vp8_signed_char_clamp(filter_value + 3 * (qs0 - ps0));
filter_value &= mask;
/* add outer taps if we have high edge variance */
filter_value = vp8_signed_char_clamp(ps1 - qs1);
filter_value = vp8_signed_char_clamp(filter_value + 3 * (qs0 - ps0));
filter_value &= mask;
Filter2 = filter_value;
Filter2 &= hev;
Filter2 = filter_value;
Filter2 &= hev;
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(Filter2 + 4);
Filter2 = vp8_signed_char_clamp(Filter2 + 3);
Filter1 >>= 3;
Filter2 >>= 3;
qs0 = vp8_signed_char_clamp(qs0 - Filter1);
ps0 = vp8_signed_char_clamp(ps0 + Filter2);
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(Filter2 + 4);
Filter2 = vp8_signed_char_clamp(Filter2 + 3);
Filter1 >>= 3;
Filter2 >>= 3;
qs0 = vp8_signed_char_clamp(qs0 - Filter1);
ps0 = vp8_signed_char_clamp(ps0 + Filter2);
/* only apply wider filter if not high edge variance */
filter_value &= ~hev;
Filter2 = filter_value;
/* only apply wider filter if not high edge variance */
filter_value &= ~hev;
Filter2 = filter_value;
/* roughly 3/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 27) >> 7);
s = vp8_signed_char_clamp(qs0 - u);
*oq0 = s ^ 0x80;
s = vp8_signed_char_clamp(ps0 + u);
*op0 = s ^ 0x80;
/* roughly 3/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 27) >> 7);
s = vp8_signed_char_clamp(qs0 - u);
*oq0 = s ^ 0x80;
s = vp8_signed_char_clamp(ps0 + u);
*op0 = s ^ 0x80;
/* roughly 2/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 18) >> 7);
s = vp8_signed_char_clamp(qs1 - u);
*oq1 = s ^ 0x80;
s = vp8_signed_char_clamp(ps1 + u);
*op1 = s ^ 0x80;
/* roughly 2/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 18) >> 7);
s = vp8_signed_char_clamp(qs1 - u);
*oq1 = s ^ 0x80;
s = vp8_signed_char_clamp(ps1 + u);
*op1 = s ^ 0x80;
/* roughly 1/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7);
s = vp8_signed_char_clamp(qs2 - u);
*oq2 = s ^ 0x80;
s = vp8_signed_char_clamp(ps2 + u);
*op2 = s ^ 0x80;
/* roughly 1/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7);
s = vp8_signed_char_clamp(qs2 - u);
*oq2 = s ^ 0x80;
s = vp8_signed_char_clamp(ps2 + u);
*op2 = s ^ 0x80;
}
void vp8_mbloop_filter_horizontal_edge_c
(
unsigned char *s,
int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
int count
)
{
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
void vp8_mbloop_filter_horizontal_edge_c(unsigned char *s, int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
int count) {
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do {
mask = vp8_filter_mask(limit[0], blimit[0], s[-4 * p], s[-3 * p], s[-2 * p],
s[-1 * p], s[0 * p], s[1 * p], s[2 * p], s[3 * p]);
mask = vp8_filter_mask(limit[0], blimit[0],
s[-4*p], s[-3*p], s[-2*p], s[-1*p],
s[0*p], s[1*p], s[2*p], s[3*p]);
hev = vp8_hevmask(thresh[0], s[-2 * p], s[-1 * p], s[0 * p], s[1 * p]);
hev = vp8_hevmask(thresh[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
vp8_mbfilter(mask, hev, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p, s + 2 * p);
++s;
}
while (++i < count * 8);
vp8_mbfilter(mask, hev, s - 3 * p, s - 2 * p, s - 1 * p, s, s + 1 * p,
s + 2 * p);
++s;
} while (++i < count * 8);
}
void vp8_mbloop_filter_vertical_edge_c(unsigned char *s, int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh, int count) {
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
void vp8_mbloop_filter_vertical_edge_c
(
unsigned char *s,
int p,
const unsigned char *blimit,
const unsigned char *limit,
const unsigned char *thresh,
int count
)
{
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
do {
mask = vp8_filter_mask(limit[0], blimit[0], s[-4], s[-3], s[-2], s[-1],
s[0], s[1], s[2], s[3]);
do
{
hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
mask = vp8_filter_mask(limit[0], blimit[0],
s[-4], s[-3], s[-2], s[-1], s[0], s[1], s[2], s[3]);
hev = vp8_hevmask(thresh[0], s[-2], s[-1], s[0], s[1]);
vp8_mbfilter(mask, hev, s - 3, s - 2, s - 1, s, s + 1, s + 2);
s += p;
}
while (++i < count * 8);
vp8_mbfilter(mask, hev, s - 3, s - 2, s - 1, s, s + 1, s + 2);
s += p;
} while (++i < count * 8);
}
/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
static signed char vp8_simple_filter_mask(uc blimit, uc p1, uc p0, uc q0, uc q1)
{
/* Why does this cause problems for win32?
* error C2143: syntax error : missing ';' before 'type'
* (void) limit;
*/
signed char mask = (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 <= blimit) * -1;
return mask;
static signed char vp8_simple_filter_mask(uc blimit, uc p1, uc p0, uc q0,
uc q1) {
/* Why does this cause problems for win32?
* error C2143: syntax error : missing ';' before 'type'
* (void) limit;
*/
signed char mask = (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 <= blimit) * -1;
return mask;
}
static void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0, uc *oq1)
{
signed char filter_value, Filter1, Filter2;
signed char p1 = (signed char) * op1 ^ 0x80;
signed char p0 = (signed char) * op0 ^ 0x80;
signed char q0 = (signed char) * oq0 ^ 0x80;
signed char q1 = (signed char) * oq1 ^ 0x80;
signed char u;
static void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0,
uc *oq1) {
signed char filter_value, Filter1, Filter2;
signed char p1 = (signed char)*op1 ^ 0x80;
signed char p0 = (signed char)*op0 ^ 0x80;
signed char q0 = (signed char)*oq0 ^ 0x80;
signed char q1 = (signed char)*oq1 ^ 0x80;
signed char u;
filter_value = vp8_signed_char_clamp(p1 - q1);
filter_value = vp8_signed_char_clamp(filter_value + 3 * (q0 - p0));
filter_value &= mask;
filter_value = vp8_signed_char_clamp(p1 - q1);
filter_value = vp8_signed_char_clamp(filter_value + 3 * (q0 - p0));
filter_value &= mask;
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(filter_value + 4);
Filter1 >>= 3;
u = vp8_signed_char_clamp(q0 - Filter1);
*oq0 = u ^ 0x80;
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(filter_value + 4);
Filter1 >>= 3;
u = vp8_signed_char_clamp(q0 - Filter1);
*oq0 = u ^ 0x80;
Filter2 = vp8_signed_char_clamp(filter_value + 3);
Filter2 >>= 3;
u = vp8_signed_char_clamp(p0 + Filter2);
*op0 = u ^ 0x80;
Filter2 = vp8_signed_char_clamp(filter_value + 3);
Filter2 >>= 3;
u = vp8_signed_char_clamp(p0 + Filter2);
*op0 = u ^ 0x80;
}
void vp8_loop_filter_simple_horizontal_edge_c
(
unsigned char *s,
int p,
const unsigned char *blimit
)
{
signed char mask = 0;
int i = 0;
void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *s, int p,
const unsigned char *blimit) {
signed char mask = 0;
int i = 0;
do
{
mask = vp8_simple_filter_mask(blimit[0], s[-2*p], s[-1*p], s[0*p], s[1*p]);
vp8_simple_filter(mask, s - 2 * p, s - 1 * p, s, s + 1 * p);
++s;
}
while (++i < 16);
do {
mask = vp8_simple_filter_mask(blimit[0], s[-2 * p], s[-1 * p], s[0 * p],
s[1 * p]);
vp8_simple_filter(mask, s - 2 * p, s - 1 * p, s, s + 1 * p);
++s;
} while (++i < 16);
}
void vp8_loop_filter_simple_vertical_edge_c
(
unsigned char *s,
int p,
const unsigned char *blimit
)
{
signed char mask = 0;
int i = 0;
do
{
mask = vp8_simple_filter_mask(blimit[0], s[-2], s[-1], s[0], s[1]);
vp8_simple_filter(mask, s - 2, s - 1, s, s + 1);
s += p;
}
while (++i < 16);
void vp8_loop_filter_simple_vertical_edge_c(unsigned char *s, int p,
const unsigned char *blimit) {
signed char mask = 0;
int i = 0;
do {
mask = vp8_simple_filter_mask(blimit[0], s[-2], s[-1], s[0], s[1]);
vp8_simple_filter(mask, s - 2, s - 1, s, s + 1);
s += p;
} while (++i < 16);
}
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi)
{
vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
loop_filter_info *lfi) {
vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi)
{
vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
loop_filter_info *lfi) {
vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi)
{
vp8_loop_filter_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
loop_filter_info *lfi) {
vp8_loop_filter_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
}
void vp8_loop_filter_bhs_c(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit)
{
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, blimit);
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 4 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 8 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride,
blimit);
}
/* Vertical B Filtering */
void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi)
{
vp8_loop_filter_vertical_edge_c(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
loop_filter_info *lfi) {
vp8_loop_filter_vertical_edge_c(y_ptr + 4, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 8, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 1);
}
void vp8_loop_filter_bvs_c(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit)
{
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 12, y_stride, blimit);
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_c(y_ptr + 12, y_stride, blimit);
}

77
vp8/common/mbpitch.c
View File

@ -8,61 +8,50 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "blockd.h"
void vp8_setup_block_dptrs(MACROBLOCKD *x)
{
int r, c;
void vp8_setup_block_dptrs(MACROBLOCKD *x) {
int r, c;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
x->block[r*4+c].predictor = x->predictor + r * 4 * 16 + c * 4;
}
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++) {
x->block[r * 4 + c].predictor = x->predictor + r * 4 * 16 + c * 4;
}
}
for (r = 0; r < 2; r++)
{
for (c = 0; c < 2; c++)
{
x->block[16+r*2+c].predictor = x->predictor + 256 + r * 4 * 8 + c * 4;
}
for (r = 0; r < 2; r++) {
for (c = 0; c < 2; c++) {
x->block[16 + r * 2 + c].predictor =
x->predictor + 256 + r * 4 * 8 + c * 4;
}
}
for (r = 0; r < 2; r++)
{
for (c = 0; c < 2; c++)
{
x->block[20+r*2+c].predictor = x->predictor + 320 + r * 4 * 8 + c * 4;
}
for (r = 0; r < 2; r++) {
for (c = 0; c < 2; c++) {
x->block[20 + r * 2 + c].predictor =
x->predictor + 320 + r * 4 * 8 + c * 4;
}
}
for (r = 0; r < 25; r++)
{
x->block[r].qcoeff = x->qcoeff + r * 16;
x->block[r].dqcoeff = x->dqcoeff + r * 16;
x->block[r].eob = x->eobs + r;
}
for (r = 0; r < 25; r++) {
x->block[r].qcoeff = x->qcoeff + r * 16;
x->block[r].dqcoeff = x->dqcoeff + r * 16;
x->block[r].eob = x->eobs + r;
}
}
void vp8_build_block_doffsets(MACROBLOCKD *x)
{
int block;
void vp8_build_block_doffsets(MACROBLOCKD *x) {
int block;
for (block = 0; block < 16; block++) /* y blocks */
{
x->block[block].offset =
(block >> 2) * 4 * x->dst.y_stride + (block & 3) * 4;
}
for (block = 0; block < 16; block++) /* y blocks */
{
x->block[block].offset =
(block >> 2) * 4 * x->dst.y_stride + (block & 3) * 4;
}
for (block = 16; block < 20; block++) /* U and V blocks */
{
x->block[block+4].offset =
x->block[block].offset =
((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4;
}
for (block = 16; block < 20; block++) /* U and V blocks */
{
x->block[block + 4].offset = x->block[block].offset =
((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4;
}
}

579
vp8/common/mfqe.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
/* MFQE: Multiframe Quality Enhancement
* In rate limited situations keyframes may cause significant visual artifacts
* commonly referred to as "popping." This file implements a postproccesing
@ -28,359 +27,299 @@
#include <stdlib.h>
static void filter_by_weight(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride,
int block_size, int src_weight)
{
int dst_weight = (1 << MFQE_PRECISION) - src_weight;
int rounding_bit = 1 << (MFQE_PRECISION - 1);
int r, c;
unsigned char *dst, int dst_stride, int block_size,
int src_weight) {
int dst_weight = (1 << MFQE_PRECISION) - src_weight;
int rounding_bit = 1 << (MFQE_PRECISION - 1);
int r, c;
for (r = 0; r < block_size; r++)
{
for (c = 0; c < block_size; c++)
{
dst[c] = (src[c] * src_weight +
dst[c] * dst_weight +
rounding_bit) >> MFQE_PRECISION;
}
src += src_stride;
dst += dst_stride;
for (r = 0; r < block_size; r++) {
for (c = 0; c < block_size; c++) {
dst[c] = (src[c] * src_weight + dst[c] * dst_weight + rounding_bit) >>
MFQE_PRECISION;
}
src += src_stride;
dst += dst_stride;
}
}
void vp8_filter_by_weight16x16_c(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride,
int src_weight)
{
filter_by_weight(src, src_stride, dst, dst_stride, 16, src_weight);
int src_weight) {
filter_by_weight(src, src_stride, dst, dst_stride, 16, src_weight);
}
void vp8_filter_by_weight8x8_c(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride,
int src_weight)
{
filter_by_weight(src, src_stride, dst, dst_stride, 8, src_weight);
int src_weight) {
filter_by_weight(src, src_stride, dst, dst_stride, 8, src_weight);
}
void vp8_filter_by_weight4x4_c(unsigned char *src, int src_stride,
unsigned char *dst, int dst_stride,
int src_weight)
{
filter_by_weight(src, src_stride, dst, dst_stride, 4, src_weight);
int src_weight) {
filter_by_weight(src, src_stride, dst, dst_stride, 4, src_weight);
}
static void apply_ifactor(unsigned char *y_src,
int y_src_stride,
unsigned char *y_dst,
int y_dst_stride,
unsigned char *u_src,
unsigned char *v_src,
int uv_src_stride,
unsigned char *u_dst,
unsigned char *v_dst,
int uv_dst_stride,
int block_size,
int src_weight)
{
if (block_size == 16)
{
vp8_filter_by_weight16x16(y_src, y_src_stride, y_dst, y_dst_stride, src_weight);
vp8_filter_by_weight8x8(u_src, uv_src_stride, u_dst, uv_dst_stride, src_weight);
vp8_filter_by_weight8x8(v_src, uv_src_stride, v_dst, uv_dst_stride, src_weight);
}
else /* if (block_size == 8) */
{
vp8_filter_by_weight8x8(y_src, y_src_stride, y_dst, y_dst_stride, src_weight);
vp8_filter_by_weight4x4(u_src, uv_src_stride, u_dst, uv_dst_stride, src_weight);
vp8_filter_by_weight4x4(v_src, uv_src_stride, v_dst, uv_dst_stride, src_weight);
}
static void apply_ifactor(unsigned char *y_src, int y_src_stride,
unsigned char *y_dst, int y_dst_stride,
unsigned char *u_src, unsigned char *v_src,
int uv_src_stride, unsigned char *u_dst,
unsigned char *v_dst, int uv_dst_stride,
int block_size, int src_weight) {
if (block_size == 16) {
vp8_filter_by_weight16x16(y_src, y_src_stride, y_dst, y_dst_stride,
src_weight);
vp8_filter_by_weight8x8(u_src, uv_src_stride, u_dst, uv_dst_stride,
src_weight);
vp8_filter_by_weight8x8(v_src, uv_src_stride, v_dst, uv_dst_stride,
src_weight);
} else /* if (block_size == 8) */
{
vp8_filter_by_weight8x8(y_src, y_src_stride, y_dst, y_dst_stride,
src_weight);
vp8_filter_by_weight4x4(u_src, uv_src_stride, u_dst, uv_dst_stride,
src_weight);
vp8_filter_by_weight4x4(v_src, uv_src_stride, v_dst, uv_dst_stride,
src_weight);
}
}
static unsigned int int_sqrt(unsigned int x)
{
unsigned int y = x;
unsigned int guess;
int p = 1;
while (y>>=1) p++;
p>>=1;
static unsigned int int_sqrt(unsigned int x) {
unsigned int y = x;
unsigned int guess;
int p = 1;
while (y >>= 1) p++;
p >>= 1;
guess=0;
while (p>=0)
{
guess |= (1<<p);
if (x<guess*guess)
guess -= (1<<p);
p--;
}
/* choose between guess or guess+1 */
return guess+(guess*guess+guess+1<=x);
guess = 0;
while (p >= 0) {
guess |= (1 << p);
if (x < guess * guess) guess -= (1 << p);
p--;
}
/* choose between guess or guess+1 */
return guess + (guess * guess + guess + 1 <= x);
}
#define USE_SSD
static void multiframe_quality_enhance_block
(
static void multiframe_quality_enhance_block(
int blksize, /* Currently only values supported are 16, 8 */
int qcurr,
int qprev,
unsigned char *y,
unsigned char *u,
unsigned char *v,
int y_stride,
int uv_stride,
unsigned char *yd,
unsigned char *ud,
unsigned char *vd,
int yd_stride,
int uvd_stride
)
{
static const unsigned char VP8_ZEROS[16]=
int qcurr, int qprev, unsigned char *y, unsigned char *u, unsigned char *v,
int y_stride, int uv_stride, unsigned char *yd, unsigned char *ud,
unsigned char *vd, int yd_stride, int uvd_stride) {
static const unsigned char VP8_ZEROS[16] = { 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0 };
int uvblksize = blksize >> 1;
int qdiff = qcurr - qprev;
int i;
unsigned char *up;
unsigned char *udp;
unsigned char *vp;
unsigned char *vdp;
unsigned int act, actd, sad, usad, vsad, sse, thr, thrsq, actrisk;
if (blksize == 16) {
actd = (vpx_variance16x16(yd, yd_stride, VP8_ZEROS, 0, &sse) + 128) >> 8;
act = (vpx_variance16x16(y, y_stride, VP8_ZEROS, 0, &sse) + 128) >> 8;
#ifdef USE_SSD
vpx_variance16x16(y, y_stride, yd, yd_stride, &sse);
sad = (sse + 128) >> 8;
vpx_variance8x8(u, uv_stride, ud, uvd_stride, &sse);
usad = (sse + 32) >> 6;
vpx_variance8x8(v, uv_stride, vd, uvd_stride, &sse);
vsad = (sse + 32) >> 6;
#else
sad = (vpx_sad16x16(y, y_stride, yd, yd_stride) + 128) >> 8;
usad = (vpx_sad8x8(u, uv_stride, ud, uvd_stride) + 32) >> 6;
vsad = (vpx_sad8x8(v, uv_stride, vd, uvd_stride) + 32) >> 6;
#endif
} else /* if (blksize == 8) */
{
actd = (vpx_variance8x8(yd, yd_stride, VP8_ZEROS, 0, &sse) + 32) >> 6;
act = (vpx_variance8x8(y, y_stride, VP8_ZEROS, 0, &sse) + 32) >> 6;
#ifdef USE_SSD
vpx_variance8x8(y, y_stride, yd, yd_stride, &sse);
sad = (sse + 32) >> 6;
vpx_variance4x4(u, uv_stride, ud, uvd_stride, &sse);
usad = (sse + 8) >> 4;
vpx_variance4x4(v, uv_stride, vd, uvd_stride, &sse);
vsad = (sse + 8) >> 4;
#else
sad = (vpx_sad8x8(y, y_stride, yd, yd_stride) + 32) >> 6;
usad = (vpx_sad4x4(u, uv_stride, ud, uvd_stride) + 8) >> 4;
vsad = (vpx_sad4x4(v, uv_stride, vd, uvd_stride) + 8) >> 4;
#endif
}
actrisk = (actd > act * 5);
/* thr = qdiff/16 + log2(act) + log4(qprev) */
thr = (qdiff >> 4);
while (actd >>= 1) thr++;
while (qprev >>= 2) thr++;
#ifdef USE_SSD
thrsq = thr * thr;
if (sad < thrsq &&
/* additional checks for color mismatch and excessive addition of
* high-frequencies */
4 * usad < thrsq && 4 * vsad < thrsq && !actrisk)
#else
if (sad < thr &&
/* additional checks for color mismatch and excessive addition of
* high-frequencies */
2 * usad < thr && 2 * vsad < thr && !actrisk)
#endif
{
int ifactor;
#ifdef USE_SSD
/* TODO: optimize this later to not need sqr root */
sad = int_sqrt(sad);
#endif
ifactor = (sad << MFQE_PRECISION) / thr;
ifactor >>= (qdiff >> 5);
if (ifactor) {
apply_ifactor(y, y_stride, yd, yd_stride, u, v, uv_stride, ud, vd,
uvd_stride, blksize, ifactor);
}
} else /* else implicitly copy from previous frame */
{
if (blksize == 16) {
vp8_copy_mem16x16(y, y_stride, yd, yd_stride);
vp8_copy_mem8x8(u, uv_stride, ud, uvd_stride);
vp8_copy_mem8x8(v, uv_stride, vd, uvd_stride);
} else /* if (blksize == 8) */
{
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
vp8_copy_mem8x8(y, y_stride, yd, yd_stride);
for (up = u, udp = ud, i = 0; i < uvblksize;
++i, up += uv_stride, udp += uvd_stride)
memcpy(udp, up, uvblksize);
for (vp = v, vdp = vd, i = 0; i < uvblksize;
++i, vp += uv_stride, vdp += uvd_stride)
memcpy(vdp, vp, uvblksize);
}
}
}
static int qualify_inter_mb(const MODE_INFO *mode_info_context, int *map) {
if (mode_info_context->mbmi.mb_skip_coeff)
map[0] = map[1] = map[2] = map[3] = 1;
else if (mode_info_context->mbmi.mode == SPLITMV) {
static int ndx[4][4] = {
{ 0, 1, 4, 5 }, { 2, 3, 6, 7 }, { 8, 9, 12, 13 }, { 10, 11, 14, 15 }
};
int uvblksize = blksize >> 1;
int qdiff = qcurr - qprev;
int i;
unsigned char *up;
unsigned char *udp;
unsigned char *vp;
unsigned char *vdp;
unsigned int act, actd, sad, usad, vsad, sse, thr, thrsq, actrisk;
if (blksize == 16)
{
actd = (vpx_variance16x16(yd, yd_stride, VP8_ZEROS, 0, &sse)+128)>>8;
act = (vpx_variance16x16(y, y_stride, VP8_ZEROS, 0, &sse)+128)>>8;
#ifdef USE_SSD
vpx_variance16x16(y, y_stride, yd, yd_stride, &sse);
sad = (sse + 128)>>8;
vpx_variance8x8(u, uv_stride, ud, uvd_stride, &sse);
usad = (sse + 32)>>6;
vpx_variance8x8(v, uv_stride, vd, uvd_stride, &sse);
vsad = (sse + 32)>>6;
#else
sad = (vpx_sad16x16(y, y_stride, yd, yd_stride) + 128) >> 8;
usad = (vpx_sad8x8(u, uv_stride, ud, uvd_stride) + 32) >> 6;
vsad = (vpx_sad8x8(v, uv_stride, vd, uvd_stride)+ 32) >> 6;
#endif
}
else /* if (blksize == 8) */
{
actd = (vpx_variance8x8(yd, yd_stride, VP8_ZEROS, 0, &sse)+32)>>6;
act = (vpx_variance8x8(y, y_stride, VP8_ZEROS, 0, &sse)+32)>>6;
#ifdef USE_SSD
vpx_variance8x8(y, y_stride, yd, yd_stride, &sse);
sad = (sse + 32)>>6;
vpx_variance4x4(u, uv_stride, ud, uvd_stride, &sse);
usad = (sse + 8)>>4;
vpx_variance4x4(v, uv_stride, vd, uvd_stride, &sse);
vsad = (sse + 8)>>4;
#else
sad = (vpx_sad8x8(y, y_stride, yd, yd_stride) + 32) >> 6;
usad = (vpx_sad4x4(u, uv_stride, ud, uvd_stride) + 8) >> 4;
vsad = (vpx_sad4x4(v, uv_stride, vd, uvd_stride) + 8) >> 4;
#endif
}
actrisk = (actd > act * 5);
/* thr = qdiff/16 + log2(act) + log4(qprev) */
thr = (qdiff >> 4);
while (actd >>= 1) thr++;
while (qprev >>= 2) thr++;
#ifdef USE_SSD
thrsq = thr * thr;
if (sad < thrsq &&
/* additional checks for color mismatch and excessive addition of
* high-frequencies */
4 * usad < thrsq && 4 * vsad < thrsq && !actrisk)
#else
if (sad < thr &&
/* additional checks for color mismatch and excessive addition of
* high-frequencies */
2 * usad < thr && 2 * vsad < thr && !actrisk)
#endif
{
int ifactor;
#ifdef USE_SSD
/* TODO: optimize this later to not need sqr root */
sad = int_sqrt(sad);
#endif
ifactor = (sad << MFQE_PRECISION) / thr;
ifactor >>= (qdiff >> 5);
if (ifactor)
{
apply_ifactor(y, y_stride, yd, yd_stride,
u, v, uv_stride,
ud, vd, uvd_stride,
blksize, ifactor);
}
}
else /* else implicitly copy from previous frame */
{
if (blksize == 16)
{
vp8_copy_mem16x16(y, y_stride, yd, yd_stride);
vp8_copy_mem8x8(u, uv_stride, ud, uvd_stride);
vp8_copy_mem8x8(v, uv_stride, vd, uvd_stride);
}
else /* if (blksize == 8) */
{
vp8_copy_mem8x8(y, y_stride, yd, yd_stride);
for (up = u, udp = ud, i = 0; i < uvblksize; ++i, up += uv_stride, udp += uvd_stride)
memcpy(udp, up, uvblksize);
for (vp = v, vdp = vd, i = 0; i < uvblksize; ++i, vp += uv_stride, vdp += uvd_stride)
memcpy(vdp, vp, uvblksize);
}
int i, j;
for (i = 0; i < 4; ++i) {
map[i] = 1;
for (j = 0; j < 4 && map[j]; ++j)
map[i] &= (mode_info_context->bmi[ndx[i][j]].mv.as_mv.row <= 2 &&
mode_info_context->bmi[ndx[i][j]].mv.as_mv.col <= 2);
}
} else {
map[0] = map[1] = map[2] = map[3] =
(mode_info_context->mbmi.mode > B_PRED &&
abs(mode_info_context->mbmi.mv.as_mv.row) <= 2 &&
abs(mode_info_context->mbmi.mv.as_mv.col) <= 2);
}
return (map[0] + map[1] + map[2] + map[3]);
}
static int qualify_inter_mb(const MODE_INFO *mode_info_context, int *map)
{
if (mode_info_context->mbmi.mb_skip_coeff)
map[0] = map[1] = map[2] = map[3] = 1;
else if (mode_info_context->mbmi.mode==SPLITMV)
{
static int ndx[4][4] =
{
{0, 1, 4, 5},
{2, 3, 6, 7},
{8, 9, 12, 13},
{10, 11, 14, 15}
};
int i, j;
for (i=0; i<4; ++i)
{
map[i] = 1;
for (j=0; j<4 && map[j]; ++j)
map[i] &= (mode_info_context->bmi[ndx[i][j]].mv.as_mv.row <= 2 &&
mode_info_context->bmi[ndx[i][j]].mv.as_mv.col <= 2);
}
}
else
{
map[0] = map[1] = map[2] = map[3] =
(mode_info_context->mbmi.mode > B_PRED &&
abs(mode_info_context->mbmi.mv.as_mv.row) <= 2 &&
abs(mode_info_context->mbmi.mv.as_mv.col) <= 2);
}
return (map[0]+map[1]+map[2]+map[3]);
}
void vp8_multiframe_quality_enhance(VP8_COMMON *cm) {
YV12_BUFFER_CONFIG *show = cm->frame_to_show;
YV12_BUFFER_CONFIG *dest = &cm->post_proc_buffer;
void vp8_multiframe_quality_enhance
(
VP8_COMMON *cm
)
{
YV12_BUFFER_CONFIG *show = cm->frame_to_show;
YV12_BUFFER_CONFIG *dest = &cm->post_proc_buffer;
FRAME_TYPE frame_type = cm->frame_type;
/* Point at base of Mb MODE_INFO list has motion vectors etc */
const MODE_INFO *mode_info_context = cm->show_frame_mi;
int mb_row;
int mb_col;
int totmap, map[4];
int qcurr = cm->base_qindex;
int qprev = cm->postproc_state.last_base_qindex;
FRAME_TYPE frame_type = cm->frame_type;
/* Point at base of Mb MODE_INFO list has motion vectors etc */
const MODE_INFO *mode_info_context = cm->show_frame_mi;
int mb_row;
int mb_col;
int totmap, map[4];
int qcurr = cm->base_qindex;
int qprev = cm->postproc_state.last_base_qindex;
unsigned char *y_ptr, *u_ptr, *v_ptr;
unsigned char *yd_ptr, *ud_ptr, *vd_ptr;
unsigned char *y_ptr, *u_ptr, *v_ptr;
unsigned char *yd_ptr, *ud_ptr, *vd_ptr;
/* Set up the buffer pointers */
y_ptr = show->y_buffer;
u_ptr = show->u_buffer;
v_ptr = show->v_buffer;
yd_ptr = dest->y_buffer;
ud_ptr = dest->u_buffer;
vd_ptr = dest->v_buffer;
/* Set up the buffer pointers */
y_ptr = show->y_buffer;
u_ptr = show->u_buffer;
v_ptr = show->v_buffer;
yd_ptr = dest->y_buffer;
ud_ptr = dest->u_buffer;
vd_ptr = dest->v_buffer;
/* postprocess each macro block */
for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
{
for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
{
/* if motion is high there will likely be no benefit */
if (frame_type == INTER_FRAME) totmap = qualify_inter_mb(mode_info_context, map);
else totmap = (frame_type == KEY_FRAME ? 4 : 0);
if (totmap)
{
if (totmap < 4)
{
int i, j;
for (i=0; i<2; ++i)
for (j=0; j<2; ++j)
{
if (map[i*2+j])
{
multiframe_quality_enhance_block(8, qcurr, qprev,
y_ptr + 8*(i*show->y_stride+j),
u_ptr + 4*(i*show->uv_stride+j),
v_ptr + 4*(i*show->uv_stride+j),
show->y_stride,
show->uv_stride,
yd_ptr + 8*(i*dest->y_stride+j),
ud_ptr + 4*(i*dest->uv_stride+j),
vd_ptr + 4*(i*dest->uv_stride+j),
dest->y_stride,
dest->uv_stride);
}
else
{
/* copy a 8x8 block */
int k;
unsigned char *up = u_ptr + 4*(i*show->uv_stride+j);
unsigned char *udp = ud_ptr + 4*(i*dest->uv_stride+j);
unsigned char *vp = v_ptr + 4*(i*show->uv_stride+j);
unsigned char *vdp = vd_ptr + 4*(i*dest->uv_stride+j);
vp8_copy_mem8x8(y_ptr + 8*(i*show->y_stride+j), show->y_stride,
yd_ptr + 8*(i*dest->y_stride+j), dest->y_stride);
for (k = 0; k < 4; ++k, up += show->uv_stride, udp += dest->uv_stride,
vp += show->uv_stride, vdp += dest->uv_stride)
{
memcpy(udp, up, 4);
memcpy(vdp, vp, 4);
}
}
}
}
else /* totmap = 4 */
{
multiframe_quality_enhance_block(16, qcurr, qprev, y_ptr,
u_ptr, v_ptr,
show->y_stride,
show->uv_stride,
yd_ptr, ud_ptr, vd_ptr,
dest->y_stride,
dest->uv_stride);
/* postprocess each macro block */
for (mb_row = 0; mb_row < cm->mb_rows; mb_row++) {
for (mb_col = 0; mb_col < cm->mb_cols; mb_col++) {
/* if motion is high there will likely be no benefit */
if (frame_type == INTER_FRAME)
totmap = qualify_inter_mb(mode_info_context, map);
else
totmap = (frame_type == KEY_FRAME ? 4 : 0);
if (totmap) {
if (totmap < 4) {
int i, j;
for (i = 0; i < 2; ++i)
for (j = 0; j < 2; ++j) {
if (map[i * 2 + j]) {
multiframe_quality_enhance_block(
8, qcurr, qprev, y_ptr + 8 * (i * show->y_stride + j),
u_ptr + 4 * (i * show->uv_stride + j),
v_ptr + 4 * (i * show->uv_stride + j), show->y_stride,
show->uv_stride, yd_ptr + 8 * (i * dest->y_stride + j),
ud_ptr + 4 * (i * dest->uv_stride + j),
vd_ptr + 4 * (i * dest->uv_stride + j), dest->y_stride,
dest->uv_stride);
} else {
/* copy a 8x8 block */
int k;
unsigned char *up = u_ptr + 4 * (i * show->uv_stride + j);
unsigned char *udp = ud_ptr + 4 * (i * dest->uv_stride + j);
unsigned char *vp = v_ptr + 4 * (i * show->uv_stride + j);
unsigned char *vdp = vd_ptr + 4 * (i * dest->uv_stride + j);
vp8_copy_mem8x8(
y_ptr + 8 * (i * show->y_stride + j), show->y_stride,
yd_ptr + 8 * (i * dest->y_stride + j), dest->y_stride);
for (k = 0; k < 4; ++k, up += show->uv_stride,
udp += dest->uv_stride, vp += show->uv_stride,
vdp += dest->uv_stride) {
memcpy(udp, up, 4);
memcpy(vdp, vp, 4);
}
}
}
else
{
vp8_copy_mem16x16(y_ptr, show->y_stride, yd_ptr, dest->y_stride);
vp8_copy_mem8x8(u_ptr, show->uv_stride, ud_ptr, dest->uv_stride);
vp8_copy_mem8x8(v_ptr, show->uv_stride, vd_ptr, dest->uv_stride);
}
y_ptr += 16;
u_ptr += 8;
v_ptr += 8;
yd_ptr += 16;
ud_ptr += 8;
vd_ptr += 8;
mode_info_context++; /* step to next MB */
} else /* totmap = 4 */
{
multiframe_quality_enhance_block(
16, qcurr, qprev, y_ptr, u_ptr, v_ptr, show->y_stride,
show->uv_stride, yd_ptr, ud_ptr, vd_ptr, dest->y_stride,
dest->uv_stride);
}
y_ptr += show->y_stride * 16 - 16 * cm->mb_cols;
u_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
v_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
yd_ptr += dest->y_stride * 16 - 16 * cm->mb_cols;
ud_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
vd_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
mode_info_context++; /* Skip border mb */
} else {
vp8_copy_mem16x16(y_ptr, show->y_stride, yd_ptr, dest->y_stride);
vp8_copy_mem8x8(u_ptr, show->uv_stride, ud_ptr, dest->uv_stride);
vp8_copy_mem8x8(v_ptr, show->uv_stride, vd_ptr, dest->uv_stride);
}
y_ptr += 16;
u_ptr += 8;
v_ptr += 8;
yd_ptr += 16;
ud_ptr += 8;
vd_ptr += 8;
mode_info_context++; /* step to next MB */
}
y_ptr += show->y_stride * 16 - 16 * cm->mb_cols;
u_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
v_ptr += show->uv_stride * 8 - 8 * cm->mb_cols;
yd_ptr += dest->y_stride * 16 - 16 * cm->mb_cols;
ud_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
vd_ptr += dest->uv_stride * 8 - 8 * cm->mb_cols;
mode_info_context++; /* Skip border mb */
}
}

20
vp8/common/mips/dspr2/dequantize_dspr2.c
View File

@ -8,26 +8,22 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#include "vpx_mem/vpx_mem.h"
#if HAVE_DSPR2
void vp8_dequant_idct_add_dspr2(short *input, short *dq,
unsigned char *dest, int stride)
{
int i;
void vp8_dequant_idct_add_dspr2(short *input, short *dq, unsigned char *dest,
int stride) {
int i;
for (i = 0; i < 16; i++)
{
input[i] = dq[i] * input[i];
}
for (i = 0; i < 16; i++) {
input[i] = dq[i] * input[i];
}
vp8_short_idct4x4llm_dspr2(input, dest, stride, dest, stride);
memset(input, 0, 32);
vp8_short_idct4x4llm_dspr2(input, dest, stride, dest, stride);
memset(input, 0, 32);
}
#endif

5193
vp8/common/mips/dspr2/filter_dspr2.c
View File

File diff suppressed because it is too large Load Diff

104
vp8/common/mips/dspr2/idct_blk_dspr2.c
View File

@ -13,76 +13,64 @@
#if HAVE_DSPR2
void vp8_dequant_idct_add_y_block_dspr2
(short *q, short *dq,
unsigned char *dst, int stride, char *eobs)
{
int i, j;
void vp8_dequant_idct_add_y_block_dspr2(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i, j;
for (i = 0; i < 4; i++)
{
for (j = 0; j < 4; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dst, stride);
else
{
vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dst, stride, dst, stride);
((int *)q)[0] = 0;
}
for (i = 0; i < 4; i++) {
for (j = 0; j < 4; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dst, stride);
else {
vp8_dc_only_idct_add_dspr2(q[0] * dq[0], dst, stride, dst, stride);
((int *)q)[0] = 0;
}
q += 16;
dst += 4;
}
dst += 4 * stride - 16;
q += 16;
dst += 4;
}
dst += 4 * stride - 16;
}
}
void vp8_dequant_idct_add_uv_block_dspr2
(short *q, short *dq,
unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
{
int i, j;
void vp8_dequant_idct_add_uv_block_dspr2(short *q, short *dq,
unsigned char *dstu,
unsigned char *dstv, int stride,
char *eobs) {
int i, j;
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dstu, stride);
else
{
vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dstu, stride, dstu, stride);
((int *)q)[0] = 0;
}
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dstu, stride);
else {
vp8_dc_only_idct_add_dspr2(q[0] * dq[0], dstu, stride, dstu, stride);
((int *)q)[0] = 0;
}
q += 16;
dstu += 4;
}
dstu += 4 * stride - 8;
q += 16;
dstu += 4;
}
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dstv, stride);
else
{
vp8_dc_only_idct_add_dspr2(q[0]*dq[0], dstv, stride, dstv, stride);
((int *)q)[0] = 0;
}
dstu += 4 * stride - 8;
}
q += 16;
dstv += 4;
}
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
if (*eobs++ > 1)
vp8_dequant_idct_add_dspr2(q, dq, dstv, stride);
else {
vp8_dc_only_idct_add_dspr2(q[0] * dq[0], dstv, stride, dstv, stride);
((int *)q)[0] = 0;
}
dstv += 4 * stride - 8;
q += 16;
dstv += 4;
}
dstv += 4 * stride - 8;
}
}
#endif

509
vp8/common/mips/dspr2/idctllm_dspr2.c
View File

@ -28,342 +28,319 @@
****************************************************************************/
extern unsigned char ff_cropTbl[256 + 2 * CROP_WIDTH];
static const int cospi8sqrt2minus1 = 20091;
static const int sinpi8sqrt2 = 35468;
static const int sinpi8sqrt2 = 35468;
inline void prefetch_load_short(short *src)
{
__asm__ __volatile__ (
"pref 0, 0(%[src]) \n\t"
:
: [src] "r" (src)
);
inline void prefetch_load_short(short *src) {
__asm__ __volatile__("pref 0, 0(%[src]) \n\t" : : [src] "r"(src));
}
void vp8_short_idct4x4llm_dspr2(short *input, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride)
{
int r, c;
int a1, b1, c1, d1;
short output[16];
short *ip = input;
short *op = output;
int temp1, temp2;
int shortpitch = 4;
int dst_stride) {
int r, c;
int a1, b1, c1, d1;
short output[16];
short *ip = input;
short *op = output;
int temp1, temp2;
int shortpitch = 4;
int c2, d2;
int temp3, temp4;
unsigned char *cm = ff_cropTbl + CROP_WIDTH;
int c2, d2;
int temp3, temp4;
unsigned char *cm = ff_cropTbl + CROP_WIDTH;
/* prepare data for load */
prefetch_load_short(ip + 8);
/* prepare data for load */
prefetch_load_short(ip + 8);
/* first loop is unrolled */
a1 = ip[0] + ip[8];
b1 = ip[0] - ip[8];
/* first loop is unrolled */
a1 = ip[0] + ip[8];
b1 = ip[0] - ip[8];
temp1 = (ip[4] * sinpi8sqrt2) >> 16;
temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = (ip[4] * sinpi8sqrt2) >> 16;
temp2 = ip[12] + ((ip[12] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[12] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp1 = ip[4] + ((ip[4] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[12] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp3 = (ip[5] * sinpi8sqrt2) >> 16;
temp4 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = (ip[5] * sinpi8sqrt2) >> 16;
temp4 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[13] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[13] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
op[0] = a1 + d1;
op[12] = a1 - d1;
op[4] = b1 + c1;
op[8] = b1 - c1;
op[0] = a1 + d1;
op[12] = a1 - d1;
op[4] = b1 + c1;
op[8] = b1 - c1;
a1 = ip[1] + ip[9];
b1 = ip[1] - ip[9];
a1 = ip[1] + ip[9];
b1 = ip[1] - ip[9];
op[1] = a1 + d2;
op[13] = a1 - d2;
op[5] = b1 + c2;
op[9] = b1 - c2;
op[1] = a1 + d2;
op[13] = a1 - d2;
op[5] = b1 + c2;
op[9] = b1 - c2;
a1 = ip[2] + ip[10];
b1 = ip[2] - ip[10];
a1 = ip[2] + ip[10];
b1 = ip[2] - ip[10];
temp1 = (ip[6] * sinpi8sqrt2) >> 16;
temp2 = ip[14] + ((ip[14] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = (ip[6] * sinpi8sqrt2) >> 16;
temp2 = ip[14] + ((ip[14] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[6] + ((ip[6] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[14] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp1 = ip[6] + ((ip[6] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[14] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp3 = (ip[7] * sinpi8sqrt2) >> 16;
temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = (ip[7] * sinpi8sqrt2) >> 16;
temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[15] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
temp3 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[15] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
op[2] = a1 + d1;
op[14] = a1 - d1;
op[6] = b1 + c1;
op[10] = b1 - c1;
op[2] = a1 + d1;
op[14] = a1 - d1;
op[6] = b1 + c1;
op[10] = b1 - c1;
a1 = ip[3] + ip[11];
b1 = ip[3] - ip[11];
a1 = ip[3] + ip[11];
b1 = ip[3] - ip[11];
op[3] = a1 + d2;
op[15] = a1 - d2;
op[7] = b1 + c2;
op[11] = b1 - c2;
op[3] = a1 + d2;
op[15] = a1 - d2;
op[7] = b1 + c2;
op[11] = b1 - c2;
ip = output;
ip = output;
/* prepare data for load */
prefetch_load_short(ip + shortpitch);
/* prepare data for load */
prefetch_load_short(ip + shortpitch);
/* second loop is unrolled */
a1 = ip[0] + ip[2];
b1 = ip[0] - ip[2];
/* second loop is unrolled */
a1 = ip[0] + ip[2];
b1 = ip[0] - ip[2];
temp1 = (ip[1] * sinpi8sqrt2) >> 16;
temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = (ip[1] * sinpi8sqrt2) >> 16;
temp2 = ip[3] + ((ip[3] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[3] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp1 = ip[1] + ((ip[1] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[3] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp3 = (ip[5] * sinpi8sqrt2) >> 16;
temp4 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = (ip[5] * sinpi8sqrt2) >> 16;
temp4 = ip[7] + ((ip[7] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[7] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
temp3 = ip[5] + ((ip[5] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[7] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
op[0] = (a1 + d1 + 4) >> 3;
op[3] = (a1 - d1 + 4) >> 3;
op[1] = (b1 + c1 + 4) >> 3;
op[2] = (b1 - c1 + 4) >> 3;
op[0] = (a1 + d1 + 4) >> 3;
op[3] = (a1 - d1 + 4) >> 3;
op[1] = (b1 + c1 + 4) >> 3;
op[2] = (b1 - c1 + 4) >> 3;
a1 = ip[4] + ip[6];
b1 = ip[4] - ip[6];
a1 = ip[4] + ip[6];
b1 = ip[4] - ip[6];
op[4] = (a1 + d2 + 4) >> 3;
op[7] = (a1 - d2 + 4) >> 3;
op[5] = (b1 + c2 + 4) >> 3;
op[6] = (b1 - c2 + 4) >> 3;
op[4] = (a1 + d2 + 4) >> 3;
op[7] = (a1 - d2 + 4) >> 3;
op[5] = (b1 + c2 + 4) >> 3;
op[6] = (b1 - c2 + 4) >> 3;
a1 = ip[8] + ip[10];
b1 = ip[8] - ip[10];
a1 = ip[8] + ip[10];
b1 = ip[8] - ip[10];
temp1 = (ip[9] * sinpi8sqrt2) >> 16;
temp2 = ip[11] + ((ip[11] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = (ip[9] * sinpi8sqrt2) >> 16;
temp2 = ip[11] + ((ip[11] * cospi8sqrt2minus1) >> 16);
c1 = temp1 - temp2;
temp1 = ip[9] + ((ip[9] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[11] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp1 = ip[9] + ((ip[9] * cospi8sqrt2minus1) >> 16);
temp2 = (ip[11] * sinpi8sqrt2) >> 16;
d1 = temp1 + temp2;
temp3 = (ip[13] * sinpi8sqrt2) >> 16;
temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = (ip[13] * sinpi8sqrt2) >> 16;
temp4 = ip[15] + ((ip[15] * cospi8sqrt2minus1) >> 16);
c2 = temp3 - temp4;
temp3 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[15] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
temp3 = ip[13] + ((ip[13] * cospi8sqrt2minus1) >> 16);
temp4 = (ip[15] * sinpi8sqrt2) >> 16;
d2 = temp3 + temp4;
op[8] = (a1 + d1 + 4) >> 3;
op[11] = (a1 - d1 + 4) >> 3;
op[9] = (b1 + c1 + 4) >> 3;
op[10] = (b1 - c1 + 4) >> 3;
op[8] = (a1 + d1 + 4) >> 3;
op[11] = (a1 - d1 + 4) >> 3;
op[9] = (b1 + c1 + 4) >> 3;
op[10] = (b1 - c1 + 4) >> 3;
a1 = ip[12] + ip[14];
b1 = ip[12] - ip[14];
a1 = ip[12] + ip[14];
b1 = ip[12] - ip[14];
op[12] = (a1 + d2 + 4) >> 3;
op[15] = (a1 - d2 + 4) >> 3;
op[13] = (b1 + c2 + 4) >> 3;
op[14] = (b1 - c2 + 4) >> 3;
op[12] = (a1 + d2 + 4) >> 3;
op[15] = (a1 - d2 + 4) >> 3;
op[13] = (b1 + c2 + 4) >> 3;
op[14] = (b1 - c2 + 4) >> 3;
ip = output;
ip = output;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
short a = ip[c] + pred_ptr[c] ;
dst_ptr[c] = cm[a] ;
}
ip += 4;
dst_ptr += dst_stride;
pred_ptr += pred_stride;
for (r = 0; r < 4; r++) {
for (c = 0; c < 4; c++) {
short a = ip[c] + pred_ptr[c];
dst_ptr[c] = cm[a];
}
ip += 4;
dst_ptr += dst_stride;
pred_ptr += pred_stride;
}
}
void vp8_dc_only_idct_add_dspr2(short input_dc, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride)
{
int a1;
int i, absa1;
int t2, vector_a1, vector_a;
void vp8_dc_only_idct_add_dspr2(short input_dc, unsigned char *pred_ptr,
int pred_stride, unsigned char *dst_ptr,
int dst_stride) {
int a1;
int i, absa1;
int t2, vector_a1, vector_a;
/* a1 = ((input_dc + 4) >> 3); */
__asm__ __volatile__ (
"addi %[a1], %[input_dc], 4 \n\t"
"sra %[a1], %[a1], 3 \n\t"
: [a1] "=r" (a1)
: [input_dc] "r" (input_dc)
);
/* a1 = ((input_dc + 4) >> 3); */
__asm__ __volatile__(
"addi %[a1], %[input_dc], 4 \n\t"
"sra %[a1], %[a1], 3 \n\t"
: [a1] "=r"(a1)
: [input_dc] "r"(input_dc));
if (a1 < 0)
{
/* use quad-byte
* input and output memory are four byte aligned
*/
__asm__ __volatile__ (
"abs %[absa1], %[a1] \n\t"
"replv.qb %[vector_a1], %[absa1] \n\t"
: [absa1] "=r" (absa1), [vector_a1] "=r" (vector_a1)
: [a1] "r" (a1)
);
if (a1 < 0) {
/* use quad-byte
* input and output memory are four byte aligned
*/
__asm__ __volatile__(
"abs %[absa1], %[a1] \n\t"
"replv.qb %[vector_a1], %[absa1] \n\t"
: [absa1] "=r"(absa1), [vector_a1] "=r"(vector_a1)
: [a1] "r"(a1));
/* use (a1 - predptr[c]) instead a1 + predptr[c] */
for (i = 4; i--;)
{
__asm__ __volatile__ (
"lw %[t2], 0(%[pred_ptr]) \n\t"
"add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
"subu_s.qb %[vector_a], %[t2], %[vector_a1] \n\t"
"sw %[vector_a], 0(%[dst_ptr]) \n\t"
"add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
: [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
[dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
: [dst_stride] "r" (dst_stride), [pred_stride] "r" (pred_stride), [vector_a1] "r" (vector_a1)
);
}
/* use (a1 - predptr[c]) instead a1 + predptr[c] */
for (i = 4; i--;) {
__asm__ __volatile__(
"lw %[t2], 0(%[pred_ptr]) \n\t"
"add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
"subu_s.qb %[vector_a], %[t2], %[vector_a1] \n\t"
"sw %[vector_a], 0(%[dst_ptr]) \n\t"
"add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
: [t2] "=&r"(t2), [vector_a] "=&r"(vector_a),
[dst_ptr] "+&r"(dst_ptr), [pred_ptr] "+&r"(pred_ptr)
: [dst_stride] "r"(dst_stride), [pred_stride] "r"(pred_stride),
[vector_a1] "r"(vector_a1));
}
else
{
/* use quad-byte
* input and output memory are four byte aligned
*/
__asm__ __volatile__ (
"replv.qb %[vector_a1], %[a1] \n\t"
: [vector_a1] "=r" (vector_a1)
: [a1] "r" (a1)
);
} else {
/* use quad-byte
* input and output memory are four byte aligned
*/
__asm__ __volatile__("replv.qb %[vector_a1], %[a1] \n\t"
: [vector_a1] "=r"(vector_a1)
: [a1] "r"(a1));
for (i = 4; i--;)
{
__asm__ __volatile__ (
"lw %[t2], 0(%[pred_ptr]) \n\t"
"add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
"addu_s.qb %[vector_a], %[vector_a1], %[t2] \n\t"
"sw %[vector_a], 0(%[dst_ptr]) \n\t"
"add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
: [t2] "=&r" (t2), [vector_a] "=&r" (vector_a),
[dst_ptr] "+&r" (dst_ptr), [pred_ptr] "+&r" (pred_ptr)
: [dst_stride] "r" (dst_stride), [pred_stride] "r" (pred_stride), [vector_a1] "r" (vector_a1)
);
}
for (i = 4; i--;) {
__asm__ __volatile__(
"lw %[t2], 0(%[pred_ptr]) \n\t"
"add %[pred_ptr], %[pred_ptr], %[pred_stride] \n\t"
"addu_s.qb %[vector_a], %[vector_a1], %[t2] \n\t"
"sw %[vector_a], 0(%[dst_ptr]) \n\t"
"add %[dst_ptr], %[dst_ptr], %[dst_stride] \n\t"
: [t2] "=&r"(t2), [vector_a] "=&r"(vector_a),
[dst_ptr] "+&r"(dst_ptr), [pred_ptr] "+&r"(pred_ptr)
: [dst_stride] "r"(dst_stride), [pred_stride] "r"(pred_stride),
[vector_a1] "r"(vector_a1));
}
}
}
void vp8_short_inv_walsh4x4_dspr2(short *input, short *mb_dqcoeff)
{
short output[16];
int i;
int a1, b1, c1, d1;
int a2, b2, c2, d2;
short *ip = input;
short *op = output;
void vp8_short_inv_walsh4x4_dspr2(short *input, short *mb_dqcoeff) {
short output[16];
int i;
int a1, b1, c1, d1;
int a2, b2, c2, d2;
short *ip = input;
short *op = output;
prefetch_load_short(ip);
prefetch_load_short(ip);
for (i = 4; i--;)
{
a1 = ip[0] + ip[12];
b1 = ip[4] + ip[8];
c1 = ip[4] - ip[8];
d1 = ip[0] - ip[12];
for (i = 4; i--;) {
a1 = ip[0] + ip[12];
b1 = ip[4] + ip[8];
c1 = ip[4] - ip[8];
d1 = ip[0] - ip[12];
op[0] = a1 + b1;
op[4] = c1 + d1;
op[8] = a1 - b1;
op[12] = d1 - c1;
op[0] = a1 + b1;
op[4] = c1 + d1;
op[8] = a1 - b1;
op[12] = d1 - c1;
ip++;
op++;
}
ip++;
op++;
}
ip = output;
op = output;
ip = output;
op = output;
prefetch_load_short(ip);
prefetch_load_short(ip);
for (i = 4; i--;)
{
a1 = ip[0] + ip[3] + 3;
b1 = ip[1] + ip[2];
c1 = ip[1] - ip[2];
d1 = ip[0] - ip[3] + 3;
for (i = 4; i--;) {
a1 = ip[0] + ip[3] + 3;
b1 = ip[1] + ip[2];
c1 = ip[1] - ip[2];
d1 = ip[0] - ip[3] + 3;
a2 = a1 + b1;
b2 = d1 + c1;
c2 = a1 - b1;
d2 = d1 - c1;
a2 = a1 + b1;
b2 = d1 + c1;
c2 = a1 - b1;
d2 = d1 - c1;
op[0] = a2 >> 3;
op[1] = b2 >> 3;
op[2] = c2 >> 3;
op[3] = d2 >> 3;
op[0] = a2 >> 3;
op[1] = b2 >> 3;
op[2] = c2 >> 3;
op[3] = d2 >> 3;
ip += 4;
op += 4;
}
ip += 4;
op += 4;
}
for (i = 0; i < 16; i++)
{
mb_dqcoeff[i * 16] = output[i];
}
for (i = 0; i < 16; i++) {
mb_dqcoeff[i * 16] = output[i];
}
}
void vp8_short_inv_walsh4x4_1_dspr2(short *input, short *mb_dqcoeff)
{
int a1;
void vp8_short_inv_walsh4x4_1_dspr2(short *input, short *mb_dqcoeff) {
int a1;
a1 = ((input[0] + 3) >> 3);
a1 = ((input[0] + 3) >> 3);
__asm__ __volatile__ (
"sh %[a1], 0(%[mb_dqcoeff]) \n\t"
"sh %[a1], 32(%[mb_dqcoeff]) \n\t"
"sh %[a1], 64(%[mb_dqcoeff]) \n\t"
"sh %[a1], 96(%[mb_dqcoeff]) \n\t"
"sh %[a1], 128(%[mb_dqcoeff]) \n\t"
"sh %[a1], 160(%[mb_dqcoeff]) \n\t"
"sh %[a1], 192(%[mb_dqcoeff]) \n\t"
"sh %[a1], 224(%[mb_dqcoeff]) \n\t"
"sh %[a1], 256(%[mb_dqcoeff]) \n\t"
"sh %[a1], 288(%[mb_dqcoeff]) \n\t"
"sh %[a1], 320(%[mb_dqcoeff]) \n\t"
"sh %[a1], 352(%[mb_dqcoeff]) \n\t"
"sh %[a1], 384(%[mb_dqcoeff]) \n\t"
"sh %[a1], 416(%[mb_dqcoeff]) \n\t"
"sh %[a1], 448(%[mb_dqcoeff]) \n\t"
"sh %[a1], 480(%[mb_dqcoeff]) \n\t"
__asm__ __volatile__(
"sh %[a1], 0(%[mb_dqcoeff]) \n\t"
"sh %[a1], 32(%[mb_dqcoeff]) \n\t"
"sh %[a1], 64(%[mb_dqcoeff]) \n\t"
"sh %[a1], 96(%[mb_dqcoeff]) \n\t"
"sh %[a1], 128(%[mb_dqcoeff]) \n\t"
"sh %[a1], 160(%[mb_dqcoeff]) \n\t"
"sh %[a1], 192(%[mb_dqcoeff]) \n\t"
"sh %[a1], 224(%[mb_dqcoeff]) \n\t"
"sh %[a1], 256(%[mb_dqcoeff]) \n\t"
"sh %[a1], 288(%[mb_dqcoeff]) \n\t"
"sh %[a1], 320(%[mb_dqcoeff]) \n\t"
"sh %[a1], 352(%[mb_dqcoeff]) \n\t"
"sh %[a1], 384(%[mb_dqcoeff]) \n\t"
"sh %[a1], 416(%[mb_dqcoeff]) \n\t"
"sh %[a1], 448(%[mb_dqcoeff]) \n\t"
"sh %[a1], 480(%[mb_dqcoeff]) \n\t"
:
: [a1] "r" (a1), [mb_dqcoeff] "r" (mb_dqcoeff)
);
:
: [a1] "r"(a1), [mb_dqcoeff] "r"(mb_dqcoeff));
}
#endif

156
vp8/common/mips/dspr2/reconinter_dspr2.c
View File

@ -8,114 +8,90 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8_rtcd.h"
#include "vpx/vpx_integer.h"
#if HAVE_DSPR2
inline void prefetch_load_int(unsigned char *src)
{
__asm__ __volatile__ (
"pref 0, 0(%[src]) \n\t"
:
: [src] "r" (src)
);
inline void prefetch_load_int(unsigned char *src) {
__asm__ __volatile__("pref 0, 0(%[src]) \n\t" : : [src] "r"(src));
}
__inline void vp8_copy_mem16x16_dspr2(unsigned char *RESTRICT src,
int src_stride,
unsigned char *RESTRICT dst,
int dst_stride) {
int r;
unsigned int a0, a1, a2, a3;
__inline void vp8_copy_mem16x16_dspr2(
unsigned char *RESTRICT src,
int src_stride,
unsigned char *RESTRICT dst,
int dst_stride)
{
int r;
unsigned int a0, a1, a2, a3;
for (r = 16; r--;)
{
/* load src data in cache memory */
prefetch_load_int(src + src_stride);
/* use unaligned memory load and store */
__asm__ __volatile__ (
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"ulw %[a2], 8(%[src]) \n\t"
"ulw %[a3], 12(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
"sw %[a2], 8(%[dst]) \n\t"
"sw %[a3], 12(%[dst]) \n\t"
: [a0] "=&r" (a0), [a1] "=&r" (a1),
[a2] "=&r" (a2), [a3] "=&r" (a3)
: [src] "r" (src), [dst] "r" (dst)
);
src += src_stride;
dst += dst_stride;
}
}
__inline void vp8_copy_mem8x8_dspr2(
unsigned char *RESTRICT src,
int src_stride,
unsigned char *RESTRICT dst,
int dst_stride)
{
int r;
unsigned int a0, a1;
for (r = 16; r--;) {
/* load src data in cache memory */
prefetch_load_int(src + src_stride);
for (r = 8; r--;)
{
/* use unaligned memory load and store */
__asm__ __volatile__ (
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
: [a0] "=&r" (a0), [a1] "=&r" (a1)
: [src] "r" (src), [dst] "r" (dst)
);
/* use unaligned memory load and store */
__asm__ __volatile__(
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"ulw %[a2], 8(%[src]) \n\t"
"ulw %[a3], 12(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
"sw %[a2], 8(%[dst]) \n\t"
"sw %[a3], 12(%[dst]) \n\t"
: [a0] "=&r"(a0), [a1] "=&r"(a1), [a2] "=&r"(a2), [a3] "=&r"(a3)
: [src] "r"(src), [dst] "r"(dst));
src += src_stride;
dst += dst_stride;
}
src += src_stride;
dst += dst_stride;
}
}
__inline void vp8_copy_mem8x8_dspr2(unsigned char *RESTRICT src, int src_stride,
unsigned char *RESTRICT dst,
int dst_stride) {
int r;
unsigned int a0, a1;
__inline void vp8_copy_mem8x4_dspr2(
unsigned char *RESTRICT src,
int src_stride,
unsigned char *RESTRICT dst,
int dst_stride)
{
int r;
unsigned int a0, a1;
/* load src data in cache memory */
prefetch_load_int(src + src_stride);
/* load src data in cache memory */
prefetch_load_int(src + src_stride);
for (r = 8; r--;) {
/* use unaligned memory load and store */
__asm__ __volatile__(
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
: [a0] "=&r"(a0), [a1] "=&r"(a1)
: [src] "r"(src), [dst] "r"(dst));
for (r = 4; r--;)
{
/* use unaligned memory load and store */
__asm__ __volatile__ (
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
: [a0] "=&r" (a0), [a1] "=&r" (a1)
: [src] "r" (src), [dst] "r" (dst)
);
src += src_stride;
dst += dst_stride;
}
}
src += src_stride;
dst += dst_stride;
}
__inline void vp8_copy_mem8x4_dspr2(unsigned char *RESTRICT src, int src_stride,
unsigned char *RESTRICT dst,
int dst_stride) {
int r;
unsigned int a0, a1;
/* load src data in cache memory */
prefetch_load_int(src + src_stride);
for (r = 4; r--;) {
/* use unaligned memory load and store */
__asm__ __volatile__(
"ulw %[a0], 0(%[src]) \n\t"
"ulw %[a1], 4(%[src]) \n\t"
"sw %[a0], 0(%[dst]) \n\t"
"sw %[a1], 4(%[dst]) \n\t"
: [a0] "=&r"(a0), [a1] "=&r"(a1)
: [src] "r"(src), [dst] "r"(dst));
src += src_stride;
dst += dst_stride;
}
}
#endif

4310
vp8/common/mips/dspr2/vp8_loopfilter_filters_dspr2.c
View File

File diff suppressed because it is too large Load Diff

1202
vp8/common/mips/msa/bilinear_filter_msa.c
View File

File diff suppressed because it is too large Load Diff

74
vp8/common/mips/msa/copymem_msa.c
View File

@ -11,60 +11,52 @@
#include "./vp8_rtcd.h"
#include "vp8/common/mips/msa/vp8_macros_msa.h"
static void copy_8x4_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
uint64_t src0, src1, src2, src3;
static void copy_8x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
uint64_t src0, src1, src2, src3;
LD4(src, src_stride, src0, src1, src2, src3);
SD4(src0, src1, src2, src3, dst, dst_stride);
LD4(src, src_stride, src0, src1, src2, src3);
SD4(src0, src1, src2, src3, dst, dst_stride);
}
static void copy_8x8_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
uint64_t src0, src1, src2, src3;
static void copy_8x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
uint64_t src0, src1, src2, src3;
LD4(src, src_stride, src0, src1, src2, src3);
src += (4 * src_stride);
SD4(src0, src1, src2, src3, dst, dst_stride);
dst += (4 * dst_stride);
LD4(src, src_stride, src0, src1, src2, src3);
src += (4 * src_stride);
SD4(src0, src1, src2, src3, dst, dst_stride);
dst += (4 * dst_stride);
LD4(src, src_stride, src0, src1, src2, src3);
SD4(src0, src1, src2, src3, dst, dst_stride);
LD4(src, src_stride, src0, src1, src2, src3);
SD4(src0, src1, src2, src3, dst, dst_stride);
}
static void copy_16x16_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
v16u8 src8, src9, src10, src11, src12, src13, src14, src15;
static void copy_16x16_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
v16u8 src8, src9, src10, src11, src12, src13, src14, src15;
LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
src += (8 * src_stride);
LD_UB8(src, src_stride, src8, src9, src10, src11, src12, src13, src14,
src15);
LD_UB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
src += (8 * src_stride);
LD_UB8(src, src_stride, src8, src9, src10, src11, src12, src13, src14, src15);
ST_UB8(src0, src1, src2, src3, src4, src5, src6, src7, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(src8, src9, src10, src11, src12, src13, src14, src15, dst,
dst_stride);
ST_UB8(src0, src1, src2, src3, src4, src5, src6, src7, dst, dst_stride);
dst += (8 * dst_stride);
ST_UB8(src8, src9, src10, src11, src12, src13, src14, src15, dst, dst_stride);
}
void vp8_copy_mem16x16_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
copy_16x16_msa(src, src_stride, dst, dst_stride);
void vp8_copy_mem16x16_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
copy_16x16_msa(src, src_stride, dst, dst_stride);
}
void vp8_copy_mem8x8_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
copy_8x8_msa(src, src_stride, dst, dst_stride);
void vp8_copy_mem8x8_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
copy_8x8_msa(src, src_stride, dst, dst_stride);
}
void vp8_copy_mem8x4_msa(uint8_t *src, int32_t src_stride,
uint8_t *dst, int32_t dst_stride)
{
copy_8x4_msa(src, src_stride, dst, dst_stride);
void vp8_copy_mem8x4_msa(uint8_t *src, int32_t src_stride, uint8_t *dst,
int32_t dst_stride) {
copy_8x4_msa(src, src_stride, dst, dst_stride);
}

732
vp8/common/mips/msa/idct_msa.c
View File

@ -15,443 +15,401 @@
static const int32_t cospi8sqrt2minus1 = 20091;
static const int32_t sinpi8sqrt2 = 35468;
#define TRANSPOSE_TWO_4x4_H(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v8i16 s4_m, s5_m, s6_m, s7_m; \
\
TRANSPOSE8X4_SH_SH(in0, in1, in2, in3, s4_m, s5_m, s6_m, s7_m); \
ILVR_D2_SH(s6_m, s4_m, s7_m, s5_m, out0, out2); \
out1 = (v8i16)__msa_ilvl_d((v2i64)s6_m, (v2i64)s4_m); \
out3 = (v8i16)__msa_ilvl_d((v2i64)s7_m, (v2i64)s5_m); \
}
#define TRANSPOSE_TWO_4x4_H(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v8i16 s4_m, s5_m, s6_m, s7_m; \
\
TRANSPOSE8X4_SH_SH(in0, in1, in2, in3, s4_m, s5_m, s6_m, s7_m); \
ILVR_D2_SH(s6_m, s4_m, s7_m, s5_m, out0, out2); \
out1 = (v8i16)__msa_ilvl_d((v2i64)s6_m, (v2i64)s4_m); \
out3 = (v8i16)__msa_ilvl_d((v2i64)s7_m, (v2i64)s5_m); \
}
#define EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in) \
({ \
v8i16 out_m; \
v8i16 zero_m = { 0 }; \
v4i32 tmp1_m, tmp2_m; \
v4i32 sinpi8_sqrt2_m = __msa_fill_w(sinpi8sqrt2); \
\
ILVRL_H2_SW(in, zero_m, tmp1_m, tmp2_m); \
tmp1_m >>= 16; \
tmp2_m >>= 16; \
tmp1_m = (tmp1_m * sinpi8_sqrt2_m) >> 16; \
tmp2_m = (tmp2_m * sinpi8_sqrt2_m) >> 16; \
out_m = __msa_pckev_h((v8i16)tmp2_m, (v8i16)tmp1_m); \
\
out_m; \
})
#define EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in) \
({ \
v8i16 out_m; \
v8i16 zero_m = { 0 }; \
v4i32 tmp1_m, tmp2_m; \
v4i32 sinpi8_sqrt2_m = __msa_fill_w(sinpi8sqrt2); \
\
ILVRL_H2_SW(in, zero_m, tmp1_m, tmp2_m); \
tmp1_m >>= 16; \
tmp2_m >>= 16; \
tmp1_m = (tmp1_m * sinpi8_sqrt2_m) >> 16; \
tmp2_m = (tmp2_m * sinpi8_sqrt2_m) >> 16; \
out_m = __msa_pckev_h((v8i16)tmp2_m, (v8i16)tmp1_m); \
\
out_m; \
})
#define VP8_IDCT_1D_H(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v8i16 a1_m, b1_m, c1_m, d1_m; \
v8i16 c_tmp1_m, c_tmp2_m, d_tmp1_m, d_tmp2_m; \
v8i16 const_cospi8sqrt2minus1_m; \
\
const_cospi8sqrt2minus1_m = __msa_fill_h(cospi8sqrt2minus1); \
a1_m = in0 + in2; \
b1_m = in0 - in2; \
c_tmp1_m = EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in1); \
c_tmp2_m = __msa_mul_q_h(in3, const_cospi8sqrt2minus1_m); \
c_tmp2_m = c_tmp2_m >> 1; \
c_tmp2_m = in3 + c_tmp2_m; \
c1_m = c_tmp1_m - c_tmp2_m; \
d_tmp1_m = __msa_mul_q_h(in1, const_cospi8sqrt2minus1_m); \
d_tmp1_m = d_tmp1_m >> 1; \
d_tmp1_m = in1 + d_tmp1_m; \
d_tmp2_m = EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in3); \
d1_m = d_tmp1_m + d_tmp2_m; \
BUTTERFLY_4(a1_m, b1_m, c1_m, d1_m, out0, out1, out2, out3); \
}
#define VP8_IDCT_1D_H(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v8i16 a1_m, b1_m, c1_m, d1_m; \
v8i16 c_tmp1_m, c_tmp2_m, d_tmp1_m, d_tmp2_m; \
v8i16 const_cospi8sqrt2minus1_m; \
\
const_cospi8sqrt2minus1_m = __msa_fill_h(cospi8sqrt2minus1); \
a1_m = in0 + in2; \
b1_m = in0 - in2; \
c_tmp1_m = EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in1); \
c_tmp2_m = __msa_mul_q_h(in3, const_cospi8sqrt2minus1_m); \
c_tmp2_m = c_tmp2_m >> 1; \
c_tmp2_m = in3 + c_tmp2_m; \
c1_m = c_tmp1_m - c_tmp2_m; \
d_tmp1_m = __msa_mul_q_h(in1, const_cospi8sqrt2minus1_m); \
d_tmp1_m = d_tmp1_m >> 1; \
d_tmp1_m = in1 + d_tmp1_m; \
d_tmp2_m = EXPAND_TO_H_MULTIPLY_SINPI8SQRT2_PCK_TO_W(in3); \
d1_m = d_tmp1_m + d_tmp2_m; \
BUTTERFLY_4(a1_m, b1_m, c1_m, d1_m, out0, out1, out2, out3); \
}
#define VP8_IDCT_1D_W(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v4i32 a1_m, b1_m, c1_m, d1_m; \
v4i32 c_tmp1_m, c_tmp2_m, d_tmp1_m, d_tmp2_m; \
v4i32 const_cospi8sqrt2minus1_m, sinpi8_sqrt2_m; \
\
const_cospi8sqrt2minus1_m = __msa_fill_w(cospi8sqrt2minus1); \
sinpi8_sqrt2_m = __msa_fill_w(sinpi8sqrt2); \
a1_m = in0 + in2; \
b1_m = in0 - in2; \
c_tmp1_m = (in1 * sinpi8_sqrt2_m) >> 16; \
c_tmp2_m = in3 + ((in3 * const_cospi8sqrt2minus1_m) >> 16); \
c1_m = c_tmp1_m - c_tmp2_m; \
d_tmp1_m = in1 + ((in1 * const_cospi8sqrt2minus1_m) >> 16); \
d_tmp2_m = (in3 * sinpi8_sqrt2_m) >> 16; \
d1_m = d_tmp1_m + d_tmp2_m; \
BUTTERFLY_4(a1_m, b1_m, c1_m, d1_m, out0, out1, out2, out3); \
}
#define VP8_IDCT_1D_W(in0, in1, in2, in3, out0, out1, out2, out3) \
{ \
v4i32 a1_m, b1_m, c1_m, d1_m; \
v4i32 c_tmp1_m, c_tmp2_m, d_tmp1_m, d_tmp2_m; \
v4i32 const_cospi8sqrt2minus1_m, sinpi8_sqrt2_m; \
\
const_cospi8sqrt2minus1_m = __msa_fill_w(cospi8sqrt2minus1); \
sinpi8_sqrt2_m = __msa_fill_w(sinpi8sqrt2); \
a1_m = in0 + in2; \
b1_m = in0 - in2; \
c_tmp1_m = (in1 * sinpi8_sqrt2_m) >> 16; \
c_tmp2_m = in3 + ((in3 * const_cospi8sqrt2minus1_m) >> 16); \
c1_m = c_tmp1_m - c_tmp2_m; \
d_tmp1_m = in1 + ((in1 * const_cospi8sqrt2minus1_m) >> 16); \
d_tmp2_m = (in3 * sinpi8_sqrt2_m) >> 16; \
d1_m = d_tmp1_m + d_tmp2_m; \
BUTTERFLY_4(a1_m, b1_m, c1_m, d1_m, out0, out1, out2, out3); \
}
static void idct4x4_addblk_msa(int16_t *input, uint8_t *pred,
int32_t pred_stride,
uint8_t *dest, int32_t dest_stride)
{
v8i16 input0, input1;
v4i32 in0, in1, in2, in3, hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
v4i32 res0, res1, res2, res3;
v16i8 zero = { 0 };
v16i8 pred0, pred1, pred2, pred3, dest0, dest1, dest2, dest3;
v16i8 mask = { 0, 4, 8, 12, 20, 21, 22, 23, 24,
25, 26, 27, 28, 29, 30, 31 };
int32_t pred_stride, uint8_t *dest,
int32_t dest_stride) {
v8i16 input0, input1;
v4i32 in0, in1, in2, in3, hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
v4i32 res0, res1, res2, res3;
v16i8 zero = { 0 };
v16i8 pred0, pred1, pred2, pred3, dest0, dest1, dest2, dest3;
v16i8 mask = { 0, 4, 8, 12, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 };
LD_SH2(input, 8, input0, input1);
UNPCK_SH_SW(input0, in0, in1);
UNPCK_SH_SW(input1, in2, in3);
VP8_IDCT_1D_W(in0, in1, in2, in3, hz0, hz1, hz2, hz3);
TRANSPOSE4x4_SW_SW(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
VP8_IDCT_1D_W(hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3);
SRARI_W4_SW(vt0, vt1, vt2, vt3, 3);
TRANSPOSE4x4_SW_SW(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_SB4(pred, pred_stride, pred0, pred1, pred2, pred3);
ILVR_B4_SW(zero, pred0, zero, pred1, zero, pred2, zero, pred3, res0, res1,
res2, res3);
ILVR_H4_SW(zero, res0, zero, res1, zero, res2, zero, res3, res0, res1,
res2, res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
res0 = CLIP_SW_0_255(res0);
res1 = CLIP_SW_0_255(res1);
res2 = CLIP_SW_0_255(res2);
res3 = CLIP_SW_0_255(res3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
LD_SH2(input, 8, input0, input1);
UNPCK_SH_SW(input0, in0, in1);
UNPCK_SH_SW(input1, in2, in3);
VP8_IDCT_1D_W(in0, in1, in2, in3, hz0, hz1, hz2, hz3);
TRANSPOSE4x4_SW_SW(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
VP8_IDCT_1D_W(hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3);
SRARI_W4_SW(vt0, vt1, vt2, vt3, 3);
TRANSPOSE4x4_SW_SW(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_SB4(pred, pred_stride, pred0, pred1, pred2, pred3);
ILVR_B4_SW(zero, pred0, zero, pred1, zero, pred2, zero, pred3, res0, res1,
res2, res3);
ILVR_H4_SW(zero, res0, zero, res1, zero, res2, zero, res3, res0, res1, res2,
res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
res0 = CLIP_SW_0_255(res0);
res1 = CLIP_SW_0_255(res1);
res2 = CLIP_SW_0_255(res2);
res3 = CLIP_SW_0_255(res3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
}
static void idct4x4_addconst_msa(int16_t in_dc, uint8_t *pred,
int32_t pred_stride,
uint8_t *dest, int32_t dest_stride)
{
v8i16 vec;
v8i16 res0, res1, res2, res3;
v16i8 zero = { 0 };
v16i8 pred0, pred1, pred2, pred3, dest0, dest1, dest2, dest3;
v16i8 mask = { 0, 2, 4, 6, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 };
int32_t pred_stride, uint8_t *dest,
int32_t dest_stride) {
v8i16 vec;
v8i16 res0, res1, res2, res3;
v16i8 zero = { 0 };
v16i8 pred0, pred1, pred2, pred3, dest0, dest1, dest2, dest3;
v16i8 mask = { 0, 2, 4, 6, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 };
vec = __msa_fill_h(in_dc);
vec = __msa_srari_h(vec, 3);
LD_SB4(pred, pred_stride, pred0, pred1, pred2, pred3);
ILVR_B4_SH(zero, pred0, zero, pred1, zero, pred2, zero, pred3, res0, res1,
res2, res3);
ADD4(res0, vec, res1, vec, res2, vec, res3, vec, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
vec = __msa_fill_h(in_dc);
vec = __msa_srari_h(vec, 3);
LD_SB4(pred, pred_stride, pred0, pred1, pred2, pred3);
ILVR_B4_SH(zero, pred0, zero, pred1, zero, pred2, zero, pred3, res0, res1,
res2, res3);
ADD4(res0, vec, res1, vec, res2, vec, res3, vec, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
}
void vp8_short_inv_walsh4x4_msa(int16_t *input, int16_t *mb_dq_coeff)
{
v8i16 input0, input1;
v4i32 in0, in1, in2, in3, a1, b1, c1, d1;
v4i32 hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
void vp8_short_inv_walsh4x4_msa(int16_t *input, int16_t *mb_dq_coeff) {
v8i16 input0, input1;
v4i32 in0, in1, in2, in3, a1, b1, c1, d1;
v4i32 hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
LD_SH2(input, 8, input0, input1);
UNPCK_SH_SW(input0, in0, in1);
UNPCK_SH_SW(input1, in2, in3);
BUTTERFLY_4(in0, in1, in2, in3, a1, b1, c1, d1);
BUTTERFLY_4(a1, d1, c1, b1, hz0, hz1, hz3, hz2);
TRANSPOSE4x4_SW_SW(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
BUTTERFLY_4(hz0, hz1, hz2, hz3, a1, b1, c1, d1);
BUTTERFLY_4(a1, d1, c1, b1, vt0, vt1, vt3, vt2);
ADD4(vt0, 3, vt1, 3, vt2, 3, vt3, 3, vt0, vt1, vt2, vt3);
SRA_4V(vt0, vt1, vt2, vt3, 3);
mb_dq_coeff[0] = __msa_copy_s_h((v8i16)vt0, 0);
mb_dq_coeff[16] = __msa_copy_s_h((v8i16)vt1, 0);
mb_dq_coeff[32] = __msa_copy_s_h((v8i16)vt2, 0);
mb_dq_coeff[48] = __msa_copy_s_h((v8i16)vt3, 0);
mb_dq_coeff[64] = __msa_copy_s_h((v8i16)vt0, 2);
mb_dq_coeff[80] = __msa_copy_s_h((v8i16)vt1, 2);
mb_dq_coeff[96] = __msa_copy_s_h((v8i16)vt2, 2);
mb_dq_coeff[112] = __msa_copy_s_h((v8i16)vt3, 2);
mb_dq_coeff[128] = __msa_copy_s_h((v8i16)vt0, 4);
mb_dq_coeff[144] = __msa_copy_s_h((v8i16)vt1, 4);
mb_dq_coeff[160] = __msa_copy_s_h((v8i16)vt2, 4);
mb_dq_coeff[176] = __msa_copy_s_h((v8i16)vt3, 4);
mb_dq_coeff[192] = __msa_copy_s_h((v8i16)vt0, 6);
mb_dq_coeff[208] = __msa_copy_s_h((v8i16)vt1, 6);
mb_dq_coeff[224] = __msa_copy_s_h((v8i16)vt2, 6);
mb_dq_coeff[240] = __msa_copy_s_h((v8i16)vt3, 6);
LD_SH2(input, 8, input0, input1);
UNPCK_SH_SW(input0, in0, in1);
UNPCK_SH_SW(input1, in2, in3);
BUTTERFLY_4(in0, in1, in2, in3, a1, b1, c1, d1);
BUTTERFLY_4(a1, d1, c1, b1, hz0, hz1, hz3, hz2);
TRANSPOSE4x4_SW_SW(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
BUTTERFLY_4(hz0, hz1, hz2, hz3, a1, b1, c1, d1);
BUTTERFLY_4(a1, d1, c1, b1, vt0, vt1, vt3, vt2);
ADD4(vt0, 3, vt1, 3, vt2, 3, vt3, 3, vt0, vt1, vt2, vt3);
SRA_4V(vt0, vt1, vt2, vt3, 3);
mb_dq_coeff[0] = __msa_copy_s_h((v8i16)vt0, 0);
mb_dq_coeff[16] = __msa_copy_s_h((v8i16)vt1, 0);
mb_dq_coeff[32] = __msa_copy_s_h((v8i16)vt2, 0);
mb_dq_coeff[48] = __msa_copy_s_h((v8i16)vt3, 0);
mb_dq_coeff[64] = __msa_copy_s_h((v8i16)vt0, 2);
mb_dq_coeff[80] = __msa_copy_s_h((v8i16)vt1, 2);
mb_dq_coeff[96] = __msa_copy_s_h((v8i16)vt2, 2);
mb_dq_coeff[112] = __msa_copy_s_h((v8i16)vt3, 2);
mb_dq_coeff[128] = __msa_copy_s_h((v8i16)vt0, 4);
mb_dq_coeff[144] = __msa_copy_s_h((v8i16)vt1, 4);
mb_dq_coeff[160] = __msa_copy_s_h((v8i16)vt2, 4);
mb_dq_coeff[176] = __msa_copy_s_h((v8i16)vt3, 4);
mb_dq_coeff[192] = __msa_copy_s_h((v8i16)vt0, 6);
mb_dq_coeff[208] = __msa_copy_s_h((v8i16)vt1, 6);
mb_dq_coeff[224] = __msa_copy_s_h((v8i16)vt2, 6);
mb_dq_coeff[240] = __msa_copy_s_h((v8i16)vt3, 6);
}
static void dequant_idct4x4_addblk_msa(int16_t *input, int16_t *dequant_input,
uint8_t *dest, int32_t dest_stride)
{
v8i16 input0, input1, dequant_in0, dequant_in1, mul0, mul1;
v8i16 in0, in1, in2, in3;
v8i16 hz0_h, hz1_h, hz2_h, hz3_h;
v16i8 dest0, dest1, dest2, dest3;
v4i32 hz0_w, hz1_w, hz2_w, hz3_w;
v4i32 vt0, vt1, vt2, vt3, res0, res1, res2, res3;
v2i64 zero = { 0 };
v16i8 mask = { 0, 4, 8, 12, 20, 21, 22, 23, 24,
25, 26, 27, 28, 29, 30, 31 };
uint8_t *dest, int32_t dest_stride) {
v8i16 input0, input1, dequant_in0, dequant_in1, mul0, mul1;
v8i16 in0, in1, in2, in3;
v8i16 hz0_h, hz1_h, hz2_h, hz3_h;
v16i8 dest0, dest1, dest2, dest3;
v4i32 hz0_w, hz1_w, hz2_w, hz3_w;
v4i32 vt0, vt1, vt2, vt3, res0, res1, res2, res3;
v2i64 zero = { 0 };
v16i8 mask = { 0, 4, 8, 12, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 };
LD_SH2(input, 8, input0, input1);
LD_SH2(dequant_input, 8, dequant_in0, dequant_in1);
MUL2(input0, dequant_in0, input1, dequant_in1, mul0, mul1);
PCKEV_D2_SH(zero, mul0, zero, mul1, in0, in2);
PCKOD_D2_SH(zero, mul0, zero, mul1, in1, in3);
VP8_IDCT_1D_H(in0, in1, in2, in3, hz0_h, hz1_h, hz2_h, hz3_h);
PCKEV_D2_SH(hz1_h, hz0_h, hz3_h, hz2_h, mul0, mul1);
UNPCK_SH_SW(mul0, hz0_w, hz1_w);
UNPCK_SH_SW(mul1, hz2_w, hz3_w);
TRANSPOSE4x4_SW_SW(hz0_w, hz1_w, hz2_w, hz3_w, hz0_w, hz1_w, hz2_w, hz3_w);
VP8_IDCT_1D_W(hz0_w, hz1_w, hz2_w, hz3_w, vt0, vt1, vt2, vt3);
SRARI_W4_SW(vt0, vt1, vt2, vt3, 3);
TRANSPOSE4x4_SW_SW(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SW(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0, res1,
res2, res3);
ILVR_H4_SW(zero, res0, zero, res1, zero, res2, zero, res3, res0, res1,
res2, res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
res0 = CLIP_SW_0_255(res0);
res1 = CLIP_SW_0_255(res1);
res2 = CLIP_SW_0_255(res2);
res3 = CLIP_SW_0_255(res3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
LD_SH2(input, 8, input0, input1);
LD_SH2(dequant_input, 8, dequant_in0, dequant_in1);
MUL2(input0, dequant_in0, input1, dequant_in1, mul0, mul1);
PCKEV_D2_SH(zero, mul0, zero, mul1, in0, in2);
PCKOD_D2_SH(zero, mul0, zero, mul1, in1, in3);
VP8_IDCT_1D_H(in0, in1, in2, in3, hz0_h, hz1_h, hz2_h, hz3_h);
PCKEV_D2_SH(hz1_h, hz0_h, hz3_h, hz2_h, mul0, mul1);
UNPCK_SH_SW(mul0, hz0_w, hz1_w);
UNPCK_SH_SW(mul1, hz2_w, hz3_w);
TRANSPOSE4x4_SW_SW(hz0_w, hz1_w, hz2_w, hz3_w, hz0_w, hz1_w, hz2_w, hz3_w);
VP8_IDCT_1D_W(hz0_w, hz1_w, hz2_w, hz3_w, vt0, vt1, vt2, vt3);
SRARI_W4_SW(vt0, vt1, vt2, vt3, 3);
TRANSPOSE4x4_SW_SW(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_SB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SW(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0, res1,
res2, res3);
ILVR_H4_SW(zero, res0, zero, res1, zero, res2, zero, res3, res0, res1, res2,
res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
res0 = CLIP_SW_0_255(res0);
res1 = CLIP_SW_0_255(res1);
res2 = CLIP_SW_0_255(res2);
res3 = CLIP_SW_0_255(res3);
VSHF_B2_SB(res0, dest0, res1, dest1, mask, mask, dest0, dest1);
VSHF_B2_SB(res2, dest2, res3, dest3, mask, mask, dest2, dest3);
ST_SB4(dest0, dest1, dest2, dest3, dest, dest_stride);
}
static void dequant_idct4x4_addblk_2x_msa(int16_t *input,
int16_t *dequant_input,
uint8_t *dest, int32_t dest_stride)
{
v16u8 dest0, dest1, dest2, dest3;
v8i16 in0, in1, in2, in3;
v8i16 mul0, mul1, mul2, mul3, dequant_in0, dequant_in1;
v8i16 hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
v8i16 res0, res1, res2, res3;
v4i32 hz0l, hz1l, hz2l, hz3l, hz0r, hz1r, hz2r, hz3r;
v4i32 vt0l, vt1l, vt2l, vt3l, vt0r, vt1r, vt2r, vt3r;
v16i8 zero = { 0 };
int16_t *dequant_input, uint8_t *dest,
int32_t dest_stride) {
v16u8 dest0, dest1, dest2, dest3;
v8i16 in0, in1, in2, in3;
v8i16 mul0, mul1, mul2, mul3, dequant_in0, dequant_in1;
v8i16 hz0, hz1, hz2, hz3, vt0, vt1, vt2, vt3;
v8i16 res0, res1, res2, res3;
v4i32 hz0l, hz1l, hz2l, hz3l, hz0r, hz1r, hz2r, hz3r;
v4i32 vt0l, vt1l, vt2l, vt3l, vt0r, vt1r, vt2r, vt3r;
v16i8 zero = { 0 };
LD_SH4(input, 8, in0, in1, in2, in3);
LD_SH2(dequant_input, 8, dequant_in0, dequant_in1);
MUL4(in0, dequant_in0, in1, dequant_in1, in2, dequant_in0, in3, dequant_in1,
mul0, mul1, mul2, mul3);
PCKEV_D2_SH(mul2, mul0, mul3, mul1, in0, in2);
PCKOD_D2_SH(mul2, mul0, mul3, mul1, in1, in3);
VP8_IDCT_1D_H(in0, in1, in2, in3, hz0, hz1, hz2, hz3);
TRANSPOSE_TWO_4x4_H(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
UNPCK_SH_SW(hz0, hz0r, hz0l);
UNPCK_SH_SW(hz1, hz1r, hz1l);
UNPCK_SH_SW(hz2, hz2r, hz2l);
UNPCK_SH_SW(hz3, hz3r, hz3l);
VP8_IDCT_1D_W(hz0l, hz1l, hz2l, hz3l, vt0l, vt1l, vt2l, vt3l);
SRARI_W4_SW(vt0l, vt1l, vt2l, vt3l, 3);
VP8_IDCT_1D_W(hz0r, hz1r, hz2r, hz3r, vt0r, vt1r, vt2r, vt3r);
SRARI_W4_SW(vt0r, vt1r, vt2r, vt3r, 3);
PCKEV_H4_SH(vt0l, vt0r, vt1l, vt1r, vt2l, vt2r, vt3l, vt3r, vt0, vt1, vt2,
vt3);
TRANSPOSE_TWO_4x4_H(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_UB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SH(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0, res1,
res2, res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
PCKEV_B4_SH(res0, res0, res1, res1, res2, res2, res3, res3, res0, res1,
res2, res3);
PCKOD_D2_UB(dest0, res0, dest1, res1, dest0, dest1);
PCKOD_D2_UB(dest2, res2, dest3, res3, dest2, dest3);
ST_UB4(dest0, dest1, dest2, dest3, dest, dest_stride);
LD_SH4(input, 8, in0, in1, in2, in3);
LD_SH2(dequant_input, 8, dequant_in0, dequant_in1);
MUL4(in0, dequant_in0, in1, dequant_in1, in2, dequant_in0, in3, dequant_in1,
mul0, mul1, mul2, mul3);
PCKEV_D2_SH(mul2, mul0, mul3, mul1, in0, in2);
PCKOD_D2_SH(mul2, mul0, mul3, mul1, in1, in3);
VP8_IDCT_1D_H(in0, in1, in2, in3, hz0, hz1, hz2, hz3);
TRANSPOSE_TWO_4x4_H(hz0, hz1, hz2, hz3, hz0, hz1, hz2, hz3);
UNPCK_SH_SW(hz0, hz0r, hz0l);
UNPCK_SH_SW(hz1, hz1r, hz1l);
UNPCK_SH_SW(hz2, hz2r, hz2l);
UNPCK_SH_SW(hz3, hz3r, hz3l);
VP8_IDCT_1D_W(hz0l, hz1l, hz2l, hz3l, vt0l, vt1l, vt2l, vt3l);
SRARI_W4_SW(vt0l, vt1l, vt2l, vt3l, 3);
VP8_IDCT_1D_W(hz0r, hz1r, hz2r, hz3r, vt0r, vt1r, vt2r, vt3r);
SRARI_W4_SW(vt0r, vt1r, vt2r, vt3r, 3);
PCKEV_H4_SH(vt0l, vt0r, vt1l, vt1r, vt2l, vt2r, vt3l, vt3r, vt0, vt1, vt2,
vt3);
TRANSPOSE_TWO_4x4_H(vt0, vt1, vt2, vt3, vt0, vt1, vt2, vt3);
LD_UB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SH(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0, res1,
res2, res3);
ADD4(res0, vt0, res1, vt1, res2, vt2, res3, vt3, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
PCKEV_B4_SH(res0, res0, res1, res1, res2, res2, res3, res3, res0, res1, res2,
res3);
PCKOD_D2_UB(dest0, res0, dest1, res1, dest0, dest1);
PCKOD_D2_UB(dest2, res2, dest3, res3, dest2, dest3);
ST_UB4(dest0, dest1, dest2, dest3, dest, dest_stride);
__asm__ __volatile__(
"sw $zero, 0(%[input]) \n\t"
"sw $zero, 4(%[input]) \n\t"
"sw $zero, 8(%[input]) \n\t"
"sw $zero, 12(%[input]) \n\t"
"sw $zero, 16(%[input]) \n\t"
"sw $zero, 20(%[input]) \n\t"
"sw $zero, 24(%[input]) \n\t"
"sw $zero, 28(%[input]) \n\t"
"sw $zero, 32(%[input]) \n\t"
"sw $zero, 36(%[input]) \n\t"
"sw $zero, 40(%[input]) \n\t"
"sw $zero, 44(%[input]) \n\t"
"sw $zero, 48(%[input]) \n\t"
"sw $zero, 52(%[input]) \n\t"
"sw $zero, 56(%[input]) \n\t"
"sw $zero, 60(%[input]) \n\t"::
__asm__ __volatile__(
"sw $zero, 0(%[input]) \n\t"
"sw $zero, 4(%[input]) \n\t"
"sw $zero, 8(%[input]) \n\t"
"sw $zero, 12(%[input]) \n\t"
"sw $zero, 16(%[input]) \n\t"
"sw $zero, 20(%[input]) \n\t"
"sw $zero, 24(%[input]) \n\t"
"sw $zero, 28(%[input]) \n\t"
"sw $zero, 32(%[input]) \n\t"
"sw $zero, 36(%[input]) \n\t"
"sw $zero, 40(%[input]) \n\t"
"sw $zero, 44(%[input]) \n\t"
"sw $zero, 48(%[input]) \n\t"
"sw $zero, 52(%[input]) \n\t"
"sw $zero, 56(%[input]) \n\t"
"sw $zero, 60(%[input]) \n\t" ::
[input] "r"(input)
);
[input] "r"(input));
}
static void dequant_idct_addconst_2x_msa(int16_t *input, int16_t *dequant_input,
uint8_t *dest, int32_t dest_stride)
{
v8i16 input_dc0, input_dc1, vec;
v16u8 dest0, dest1, dest2, dest3;
v16i8 zero = { 0 };
v8i16 res0, res1, res2, res3;
uint8_t *dest, int32_t dest_stride) {
v8i16 input_dc0, input_dc1, vec;
v16u8 dest0, dest1, dest2, dest3;
v16i8 zero = { 0 };
v8i16 res0, res1, res2, res3;
input_dc0 = __msa_fill_h(input[0] * dequant_input[0]);
input_dc1 = __msa_fill_h(input[16] * dequant_input[0]);
SRARI_H2_SH(input_dc0, input_dc1, 3);
vec = (v8i16)__msa_pckev_d((v2i64)input_dc1, (v2i64)input_dc0);
input[0] = 0;
input[16] = 0;
LD_UB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SH(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0,
res1, res2, res3);
ADD4(res0, vec, res1, vec, res2, vec, res3, vec, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
PCKEV_B4_SH(res0, res0, res1, res1, res2, res2, res3, res3, res0, res1,
res2, res3);
PCKOD_D2_UB(dest0, res0, dest1, res1, dest0, dest1);
PCKOD_D2_UB(dest2, res2, dest3, res3, dest2, dest3);
ST_UB4(dest0, dest1, dest2, dest3, dest, dest_stride);
input_dc0 = __msa_fill_h(input[0] * dequant_input[0]);
input_dc1 = __msa_fill_h(input[16] * dequant_input[0]);
SRARI_H2_SH(input_dc0, input_dc1, 3);
vec = (v8i16)__msa_pckev_d((v2i64)input_dc1, (v2i64)input_dc0);
input[0] = 0;
input[16] = 0;
LD_UB4(dest, dest_stride, dest0, dest1, dest2, dest3);
ILVR_B4_SH(zero, dest0, zero, dest1, zero, dest2, zero, dest3, res0, res1,
res2, res3);
ADD4(res0, vec, res1, vec, res2, vec, res3, vec, res0, res1, res2, res3);
CLIP_SH4_0_255(res0, res1, res2, res3);
PCKEV_B4_SH(res0, res0, res1, res1, res2, res2, res3, res3, res0, res1, res2,
res3);
PCKOD_D2_UB(dest0, res0, dest1, res1, dest0, dest1);
PCKOD_D2_UB(dest2, res2, dest3, res3, dest2, dest3);
ST_UB4(dest0, dest1, dest2, dest3, dest, dest_stride);
}
void vp8_short_idct4x4llm_msa(int16_t *input, uint8_t *pred_ptr,
int32_t pred_stride, uint8_t *dst_ptr,
int32_t dst_stride)
{
idct4x4_addblk_msa(input, pred_ptr, pred_stride, dst_ptr, dst_stride);
int32_t dst_stride) {
idct4x4_addblk_msa(input, pred_ptr, pred_stride, dst_ptr, dst_stride);
}
void vp8_dc_only_idct_add_msa(int16_t input_dc, uint8_t *pred_ptr,
int32_t pred_stride, uint8_t *dst_ptr,
int32_t dst_stride)
{
idct4x4_addconst_msa(input_dc, pred_ptr, pred_stride, dst_ptr, dst_stride);
int32_t dst_stride) {
idct4x4_addconst_msa(input_dc, pred_ptr, pred_stride, dst_ptr, dst_stride);
}
void vp8_dequantize_b_msa(BLOCKD *d, int16_t *DQC)
{
v8i16 dqc0, dqc1, q0, q1, dq0, dq1;
void vp8_dequantize_b_msa(BLOCKD *d, int16_t *DQC) {
v8i16 dqc0, dqc1, q0, q1, dq0, dq1;
LD_SH2(DQC, 8, dqc0, dqc1);
LD_SH2(d->qcoeff, 8, q0, q1);
MUL2(dqc0, q0, dqc1, q1, dq0, dq1);
ST_SH2(dq0, dq1, d->dqcoeff, 8);
LD_SH2(DQC, 8, dqc0, dqc1);
LD_SH2(d->qcoeff, 8, q0, q1);
MUL2(dqc0, q0, dqc1, q1, dq0, dq1);
ST_SH2(dq0, dq1, d->dqcoeff, 8);
}
void vp8_dequant_idct_add_msa(int16_t *input, int16_t *dq,
uint8_t *dest, int32_t stride)
{
dequant_idct4x4_addblk_msa(input, dq, dest, stride);
void vp8_dequant_idct_add_msa(int16_t *input, int16_t *dq, uint8_t *dest,
int32_t stride) {
dequant_idct4x4_addblk_msa(input, dq, dest, stride);
__asm__ __volatile__ (
"sw $zero, 0(%[input]) \n\t"
"sw $zero, 4(%[input]) \n\t"
"sw $zero, 8(%[input]) \n\t"
"sw $zero, 12(%[input]) \n\t"
"sw $zero, 16(%[input]) \n\t"
"sw $zero, 20(%[input]) \n\t"
"sw $zero, 24(%[input]) \n\t"
"sw $zero, 28(%[input]) \n\t"
__asm__ __volatile__(
"sw $zero, 0(%[input]) \n\t"
"sw $zero, 4(%[input]) \n\t"
"sw $zero, 8(%[input]) \n\t"
"sw $zero, 12(%[input]) \n\t"
"sw $zero, 16(%[input]) \n\t"
"sw $zero, 20(%[input]) \n\t"
"sw $zero, 24(%[input]) \n\t"
"sw $zero, 28(%[input]) \n\t"
:
: [input] "r" (input)
);
:
: [input] "r"(input));
}
void vp8_dequant_idct_add_y_block_msa(int16_t *q, int16_t *dq,
uint8_t *dst, int32_t stride,
char *eobs)
{
int16_t *eobs_h = (int16_t *)eobs;
uint8_t i;
void vp8_dequant_idct_add_y_block_msa(int16_t *q, int16_t *dq, uint8_t *dst,
int32_t stride, char *eobs) {
int16_t *eobs_h = (int16_t *)eobs;
uint8_t i;
for (i = 4; i--;)
{
if (eobs_h[0])
{
if (eobs_h[0] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dst, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dst, stride);
}
}
q += 32;
if (eobs_h[1])
{
if (eobs_h[1] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dst + 8, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dst + 8, stride);
}
}
q += 32;
dst += (4 * stride);
eobs_h += 2;
}
}
void vp8_dequant_idct_add_uv_block_msa(int16_t *q, int16_t *dq,
uint8_t *dstu, uint8_t *dstv,
int32_t stride, char *eobs)
{
int16_t *eobs_h = (int16_t *)eobs;
if (eobs_h[0])
{
if (eobs_h[0] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dstu, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dstu, stride);
}
}
q += 32;
dstu += (stride * 4);
if (eobs_h[1])
{
if (eobs_h[1] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dstu, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dstu, stride);
}
for (i = 4; i--;) {
if (eobs_h[0]) {
if (eobs_h[0] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dst, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dst, stride);
}
}
q += 32;
if (eobs_h[2])
{
if (eobs_h[2] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dstv, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dstv, stride);
}
if (eobs_h[1]) {
if (eobs_h[1] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dst + 8, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dst + 8, stride);
}
}
q += 32;
dstv += (stride * 4);
if (eobs_h[3])
{
if (eobs_h[3] & 0xfefe)
{
dequant_idct4x4_addblk_2x_msa(q, dq, dstv, stride);
}
else
{
dequant_idct_addconst_2x_msa(q, dq, dstv, stride);
}
}
dst += (4 * stride);
eobs_h += 2;
}
}
void vp8_dequant_idct_add_uv_block_msa(int16_t *q, int16_t *dq, uint8_t *dstu,
uint8_t *dstv, int32_t stride,
char *eobs) {
int16_t *eobs_h = (int16_t *)eobs;
if (eobs_h[0]) {
if (eobs_h[0] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dstu, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dstu, stride);
}
}
q += 32;
dstu += (stride * 4);
if (eobs_h[1]) {
if (eobs_h[1] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dstu, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dstu, stride);
}
}
q += 32;
if (eobs_h[2]) {
if (eobs_h[2] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dstv, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dstv, stride);
}
}
q += 32;
dstv += (stride * 4);
if (eobs_h[3]) {
if (eobs_h[3] & 0xfefe) {
dequant_idct4x4_addblk_2x_msa(q, dq, dstv, stride);
} else {
dequant_idct_addconst_2x_msa(q, dq, dstv, stride);
}
}
}

1350
vp8/common/mips/msa/loopfilter_filters_msa.c
View File

File diff suppressed because it is too large Load Diff

205
vp8/common/mips/msa/mfqe_msa.c
View File

@ -14,133 +14,126 @@
static void filter_by_weight8x8_msa(uint8_t *src_ptr, int32_t src_stride,
uint8_t *dst_ptr, int32_t dst_stride,
int32_t src_weight)
{
int32_t dst_weight = (1 << MFQE_PRECISION) - src_weight;
int32_t row;
uint64_t src0_d, src1_d, dst0_d, dst1_d;
v16i8 src0 = { 0 };
v16i8 src1 = { 0 };
v16i8 dst0 = { 0 };
v16i8 dst1 = { 0 };
v8i16 src_wt, dst_wt, res_h_r, res_h_l, src_r, src_l, dst_r, dst_l;
int32_t src_weight) {
int32_t dst_weight = (1 << MFQE_PRECISION) - src_weight;
int32_t row;
uint64_t src0_d, src1_d, dst0_d, dst1_d;
v16i8 src0 = { 0 };
v16i8 src1 = { 0 };
v16i8 dst0 = { 0 };
v16i8 dst1 = { 0 };
v8i16 src_wt, dst_wt, res_h_r, res_h_l, src_r, src_l, dst_r, dst_l;
src_wt = __msa_fill_h(src_weight);
dst_wt = __msa_fill_h(dst_weight);
src_wt = __msa_fill_h(src_weight);
dst_wt = __msa_fill_h(dst_weight);
for (row = 2; row--;)
{
LD2(src_ptr, src_stride, src0_d, src1_d);
src_ptr += (2 * src_stride);
LD2(dst_ptr, dst_stride, dst0_d, dst1_d);
INSERT_D2_SB(src0_d, src1_d, src0);
INSERT_D2_SB(dst0_d, dst1_d, dst0);
for (row = 2; row--;) {
LD2(src_ptr, src_stride, src0_d, src1_d);
src_ptr += (2 * src_stride);
LD2(dst_ptr, dst_stride, dst0_d, dst1_d);
INSERT_D2_SB(src0_d, src1_d, src0);
INSERT_D2_SB(dst0_d, dst1_d, dst0);
LD2(src_ptr, src_stride, src0_d, src1_d);
src_ptr += (2 * src_stride);
LD2((dst_ptr + 2 * dst_stride), dst_stride, dst0_d, dst1_d);
INSERT_D2_SB(src0_d, src1_d, src1);
INSERT_D2_SB(dst0_d, dst1_d, dst1);
LD2(src_ptr, src_stride, src0_d, src1_d);
src_ptr += (2 * src_stride);
LD2((dst_ptr + 2 * dst_stride), dst_stride, dst0_d, dst1_d);
INSERT_D2_SB(src0_d, src1_d, src1);
INSERT_D2_SB(dst0_d, dst1_d, dst1);
UNPCK_UB_SH(src0, src_r, src_l);
UNPCK_UB_SH(dst0, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
dst0 = (v16i8)__msa_pckev_b((v16i8)res_h_l, (v16i8)res_h_r);
ST8x2_UB(dst0, dst_ptr, dst_stride);
dst_ptr += (2 * dst_stride);
UNPCK_UB_SH(src0, src_r, src_l);
UNPCK_UB_SH(dst0, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
dst0 = (v16i8)__msa_pckev_b((v16i8)res_h_l, (v16i8)res_h_r);
ST8x2_UB(dst0, dst_ptr, dst_stride);
dst_ptr += (2 * dst_stride);
UNPCK_UB_SH(src1, src_r, src_l);
UNPCK_UB_SH(dst1, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
dst1 = (v16i8)__msa_pckev_b((v16i8)res_h_l, (v16i8)res_h_r);
ST8x2_UB(dst1, dst_ptr, dst_stride);
dst_ptr += (2 * dst_stride);
}
UNPCK_UB_SH(src1, src_r, src_l);
UNPCK_UB_SH(dst1, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
dst1 = (v16i8)__msa_pckev_b((v16i8)res_h_l, (v16i8)res_h_r);
ST8x2_UB(dst1, dst_ptr, dst_stride);
dst_ptr += (2 * dst_stride);
}
}
static void filter_by_weight16x16_msa(uint8_t *src_ptr, int32_t src_stride,
uint8_t *dst_ptr, int32_t dst_stride,
int32_t src_weight)
{
int32_t dst_weight = (1 << MFQE_PRECISION) - src_weight;
int32_t row;
v16i8 src0, src1, src2, src3;
v16i8 dst0, dst1, dst2, dst3;
v8i16 src_wt, dst_wt;
v8i16 res_h_r, res_h_l;
v8i16 src_r, src_l, dst_r, dst_l;
int32_t src_weight) {
int32_t dst_weight = (1 << MFQE_PRECISION) - src_weight;
int32_t row;
v16i8 src0, src1, src2, src3;
v16i8 dst0, dst1, dst2, dst3;
v8i16 src_wt, dst_wt;
v8i16 res_h_r, res_h_l;
v8i16 src_r, src_l, dst_r, dst_l;
src_wt = __msa_fill_h(src_weight);
dst_wt = __msa_fill_h(dst_weight);
src_wt = __msa_fill_h(src_weight);
dst_wt = __msa_fill_h(dst_weight);
for (row = 4; row--;)
{
LD_SB4(src_ptr, src_stride, src0, src1, src2, src3);
src_ptr += (4 * src_stride);
LD_SB4(dst_ptr, dst_stride, dst0, dst1, dst2, dst3);
for (row = 4; row--;) {
LD_SB4(src_ptr, src_stride, src0, src1, src2, src3);
src_ptr += (4 * src_stride);
LD_SB4(dst_ptr, dst_stride, dst0, dst1, dst2, dst3);
UNPCK_UB_SH(src0, src_r, src_l);
UNPCK_UB_SH(dst0, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src0, src_r, src_l);
UNPCK_UB_SH(dst0, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src1, src_r, src_l);
UNPCK_UB_SH(dst1, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src1, src_r, src_l);
UNPCK_UB_SH(dst1, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src2, src_r, src_l);
UNPCK_UB_SH(dst2, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src2, src_r, src_l);
UNPCK_UB_SH(dst2, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
UNPCK_UB_SH(src3, src_r, src_l);
UNPCK_UB_SH(dst3, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
}
UNPCK_UB_SH(src3, src_r, src_l);
UNPCK_UB_SH(dst3, dst_r, dst_l);
res_h_r = (src_r * src_wt);
res_h_r += (dst_r * dst_wt);
res_h_l = (src_l * src_wt);
res_h_l += (dst_l * dst_wt);
SRARI_H2_SH(res_h_r, res_h_l, MFQE_PRECISION);
PCKEV_ST_SB(res_h_r, res_h_l, dst_ptr);
dst_ptr += dst_stride;
}
}
void vp8_filter_by_weight16x16_msa(uint8_t *src_ptr, int32_t src_stride,
uint8_t *dst_ptr, int32_t dst_stride,
int32_t src_weight)
{
filter_by_weight16x16_msa(src_ptr, src_stride, dst_ptr, dst_stride,
src_weight);
int32_t src_weight) {
filter_by_weight16x16_msa(src_ptr, src_stride, dst_ptr, dst_stride,
src_weight);
}
void vp8_filter_by_weight8x8_msa(uint8_t *src_ptr, int32_t src_stride,
uint8_t *dst_ptr, int32_t dst_stride,
int32_t src_weight)
{
filter_by_weight8x8_msa(src_ptr, src_stride, dst_ptr, dst_stride,
src_weight);
int32_t src_weight) {
filter_by_weight8x8_msa(src_ptr, src_stride, dst_ptr, dst_stride, src_weight);
}

2771
vp8/common/mips/msa/sixtap_filter_msa.c
View File

File diff suppressed because it is too large Load Diff

1847
vp8/common/mips/msa/vp8_macros_msa.h
View File

File diff suppressed because it is too large Load Diff

52
vp8/common/modecont.c
View File

@ -8,33 +8,31 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "entropy.h"
const int vp8_mode_contexts[6][4] =
{
{
/* 0 */
7, 1, 1, 143,
},
{
/* 1 */
14, 18, 14, 107,
},
{
/* 2 */
135, 64, 57, 68,
},
{
/* 3 */
60, 56, 128, 65,
},
{
/* 4 */
159, 134, 128, 34,
},
{
/* 5 */
234, 188, 128, 28,
},
const int vp8_mode_contexts[6][4] = {
{
/* 0 */
7, 1, 1, 143,
},
{
/* 1 */
14, 18, 14, 107,
},
{
/* 2 */
135, 64, 57, 68,
},
{
/* 3 */
60, 56, 128, 65,
},
{
/* 4 */
159, 134, 128, 34,
},
{
/* 5 */
234, 188, 128, 28,
},
};

1
vp8/common/modecont.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_MODECONT_H_
#define VP8_COMMON_MODECONT_H_

17
vp8/common/mv.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_MV_H_
#define VP8_COMMON_MV_H_
#include "vpx/vpx_integer.h"
@ -17,17 +16,15 @@
extern "C" {
#endif
typedef struct
{
short row;
short col;
typedef struct {
short row;
short col;
} MV;
typedef union int_mv
{
uint32_t as_int;
MV as_mv;
} int_mv; /* facilitates faster equality tests and copies */
typedef union int_mv {
uint32_t as_int;
MV as_mv;
} int_mv; /* facilitates faster equality tests and copies */
#ifdef __cplusplus
} // extern "C"

422
vp8/common/onyx.h
View File

@ -8,13 +8,11 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ONYX_H_
#define VP8_COMMON_ONYX_H_
#ifdef __cplusplus
extern "C"
{
extern "C" {
#endif
#include "vpx_config.h"
@ -24,256 +22,258 @@ extern "C"
#include "vpx_scale/yv12config.h"
#include "ppflags.h"
struct VP8_COMP;
struct VP8_COMP;
/* Create/destroy static data structures. */
/* Create/destroy static data structures. */
typedef enum
{
NORMAL = 0,
FOURFIVE = 1,
THREEFIVE = 2,
ONETWO = 3
typedef enum {
NORMAL = 0,
FOURFIVE = 1,
THREEFIVE = 2,
ONETWO = 3
} VPX_SCALING;
} VPX_SCALING;
typedef enum {
USAGE_LOCAL_FILE_PLAYBACK = 0x0,
USAGE_STREAM_FROM_SERVER = 0x1,
USAGE_CONSTRAINED_QUALITY = 0x2,
USAGE_CONSTANT_QUALITY = 0x3
} END_USAGE;
typedef enum
{
USAGE_LOCAL_FILE_PLAYBACK = 0x0,
USAGE_STREAM_FROM_SERVER = 0x1,
USAGE_CONSTRAINED_QUALITY = 0x2,
USAGE_CONSTANT_QUALITY = 0x3
} END_USAGE;
typedef enum
{
MODE_REALTIME = 0x0,
MODE_GOODQUALITY = 0x1,
MODE_BESTQUALITY = 0x2,
MODE_FIRSTPASS = 0x3,
MODE_SECONDPASS = 0x4,
MODE_SECONDPASS_BEST = 0x5
} MODE;
typedef enum
{
FRAMEFLAGS_KEY = 1,
FRAMEFLAGS_GOLDEN = 2,
FRAMEFLAGS_ALTREF = 4
} FRAMETYPE_FLAGS;
typedef enum {
MODE_REALTIME = 0x0,
MODE_GOODQUALITY = 0x1,
MODE_BESTQUALITY = 0x2,
MODE_FIRSTPASS = 0x3,
MODE_SECONDPASS = 0x4,
MODE_SECONDPASS_BEST = 0x5
} MODE;
typedef enum {
FRAMEFLAGS_KEY = 1,
FRAMEFLAGS_GOLDEN = 2,
FRAMEFLAGS_ALTREF = 4
} FRAMETYPE_FLAGS;
#include <assert.h>
static INLINE void Scale2Ratio(int mode, int *hr, int *hs)
{
switch (mode)
{
case NORMAL:
*hr = 1;
*hs = 1;
break;
case FOURFIVE:
*hr = 4;
*hs = 5;
break;
case THREEFIVE:
*hr = 3;
*hs = 5;
break;
case ONETWO:
*hr = 1;
*hs = 2;
break;
default:
*hr = 1;
*hs = 1;
assert(0);
break;
}
}
static INLINE void Scale2Ratio(int mode, int *hr, int *hs) {
switch (mode) {
case NORMAL:
*hr = 1;
*hs = 1;
break;
case FOURFIVE:
*hr = 4;
*hs = 5;
break;
case THREEFIVE:
*hr = 3;
*hs = 5;
break;
case ONETWO:
*hr = 1;
*hs = 2;
break;
default:
*hr = 1;
*hs = 1;
assert(0);
break;
}
}
typedef struct
{
/* 4 versions of bitstream defined:
* 0 best quality/slowest decode, 3 lowest quality/fastest decode
*/
int Version;
int Width;
int Height;
struct vpx_rational timebase;
unsigned int target_bandwidth; /* kilobits per second */
typedef struct {
/* 4 versions of bitstream defined:
* 0 best quality/slowest decode, 3 lowest quality/fastest decode
*/
int Version;
int Width;
int Height;
struct vpx_rational timebase;
unsigned int target_bandwidth; /* kilobits per second */
/* Parameter used for applying denoiser.
* For temporal denoiser: noise_sensitivity = 0 means off,
* noise_sensitivity = 1 means temporal denoiser on for Y channel only,
* noise_sensitivity = 2 means temporal denoiser on for all channels.
* noise_sensitivity = 3 means aggressive denoising mode.
* noise_sensitivity >= 4 means adaptive denoising mode.
* Temporal denoiser is enabled via the configuration option:
* CONFIG_TEMPORAL_DENOISING.
* For spatial denoiser: noise_sensitivity controls the amount of
* pre-processing blur: noise_sensitivity = 0 means off.
* Spatial denoiser invoked under !CONFIG_TEMPORAL_DENOISING.
*/
int noise_sensitivity;
/* Parameter used for applying denoiser.
* For temporal denoiser: noise_sensitivity = 0 means off,
* noise_sensitivity = 1 means temporal denoiser on for Y channel only,
* noise_sensitivity = 2 means temporal denoiser on for all channels.
* noise_sensitivity = 3 means aggressive denoising mode.
* noise_sensitivity >= 4 means adaptive denoising mode.
* Temporal denoiser is enabled via the configuration option:
* CONFIG_TEMPORAL_DENOISING.
* For spatial denoiser: noise_sensitivity controls the amount of
* pre-processing blur: noise_sensitivity = 0 means off.
* Spatial denoiser invoked under !CONFIG_TEMPORAL_DENOISING.
*/
int noise_sensitivity;
/* parameter used for sharpening output: recommendation 0: */
int Sharpness;
int cpu_used;
unsigned int rc_max_intra_bitrate_pct;
unsigned int screen_content_mode;
/* parameter used for sharpening output: recommendation 0: */
int Sharpness;
int cpu_used;
unsigned int rc_max_intra_bitrate_pct;
unsigned int screen_content_mode;
/* mode ->
*(0)=Realtime/Live Encoding. This mode is optimized for realtim
* encoding (for example, capturing a television signal or feed
* from a live camera). ( speed setting controls how fast )
*(1)=Good Quality Fast Encoding. The encoder balances quality with
* the amount of time it takes to encode the output. ( speed
* setting controls how fast )
*(2)=One Pass - Best Quality. The encoder places priority on the
* quality of the output over encoding speed. The output is
* compressed at the highest possible quality. This option takes
* the longest amount of time to encode. ( speed setting ignored
* )
*(3)=Two Pass - First Pass. The encoder generates a file of
* statistics for use in the second encoding pass. ( speed
* setting controls how fast )
*(4)=Two Pass - Second Pass. The encoder uses the statistics that
* were generated in the first encoding pass to create the
* compressed output. ( speed setting controls how fast )
*(5)=Two Pass - Second Pass Best. The encoder uses the statistics
* that were generated in the first encoding pass to create the
* compressed output using the highest possible quality, and
* taking a longer amount of time to encode.. ( speed setting
* ignored )
*/
int Mode;
/* mode ->
*(0)=Realtime/Live Encoding. This mode is optimized for realtim
* encoding (for example, capturing a television signal or feed
* from a live camera). ( speed setting controls how fast )
*(1)=Good Quality Fast Encoding. The encoder balances quality with
* the amount of time it takes to encode the output. ( speed
* setting controls how fast )
*(2)=One Pass - Best Quality. The encoder places priority on the
* quality of the output over encoding speed. The output is
* compressed at the highest possible quality. This option takes
* the longest amount of time to encode. ( speed setting ignored
* )
*(3)=Two Pass - First Pass. The encoder generates a file of
* statistics for use in the second encoding pass. ( speed
* setting controls how fast )
*(4)=Two Pass - Second Pass. The encoder uses the statistics that
* were generated in the first encoding pass to create the
* compressed output. ( speed setting controls how fast )
*(5)=Two Pass - Second Pass Best. The encoder uses the statistics
* that were generated in the first encoding pass to create the
* compressed output using the highest possible quality, and
* taking a longer amount of time to encode.. ( speed setting
* ignored )
*/
int Mode;
/* Key Framing Operations */
int auto_key; /* automatically detect cut scenes */
int key_freq; /* maximum distance to key frame. */
/* Key Framing Operations */
int auto_key; /* automatically detect cut scenes */
int key_freq; /* maximum distance to key frame. */
/* lagged compression (if allow_lag == 0 lag_in_frames is ignored) */
int allow_lag;
int lag_in_frames; /* how many frames lag before we start encoding */
/* lagged compression (if allow_lag == 0 lag_in_frames is ignored) */
int allow_lag;
int lag_in_frames; /* how many frames lag before we start encoding */
/*
* DATARATE CONTROL OPTIONS
*/
/*
* DATARATE CONTROL OPTIONS
*/
int end_usage; /* vbr or cbr */
int end_usage; /* vbr or cbr */
/* buffer targeting aggressiveness */
int under_shoot_pct;
int over_shoot_pct;
/* buffer targeting aggressiveness */
int under_shoot_pct;
int over_shoot_pct;
/* buffering parameters */
int64_t starting_buffer_level;
int64_t optimal_buffer_level;
int64_t maximum_buffer_size;
/* buffering parameters */
int64_t starting_buffer_level;
int64_t optimal_buffer_level;
int64_t maximum_buffer_size;
int64_t starting_buffer_level_in_ms;
int64_t optimal_buffer_level_in_ms;
int64_t maximum_buffer_size_in_ms;
int64_t starting_buffer_level_in_ms;
int64_t optimal_buffer_level_in_ms;
int64_t maximum_buffer_size_in_ms;
/* controlling quality */
int fixed_q;
int worst_allowed_q;
int best_allowed_q;
int cq_level;
/* controlling quality */
int fixed_q;
int worst_allowed_q;
int best_allowed_q;
int cq_level;
/* allow internal resizing */
int allow_spatial_resampling;
int resample_down_water_mark;
int resample_up_water_mark;
/* allow internal resizing */
int allow_spatial_resampling;
int resample_down_water_mark;
int resample_up_water_mark;
/* allow internal frame rate alterations */
int allow_df;
int drop_frames_water_mark;
/* allow internal frame rate alterations */
int allow_df;
int drop_frames_water_mark;
/* two pass datarate control */
int two_pass_vbrbias;
int two_pass_vbrmin_section;
int two_pass_vbrmax_section;
/* two pass datarate control */
int two_pass_vbrbias;
int two_pass_vbrmin_section;
int two_pass_vbrmax_section;
/*
* END DATARATE CONTROL OPTIONS
*/
/*
* END DATARATE CONTROL OPTIONS
*/
/* these parameters aren't to be used in final build don't use!!! */
int play_alternate;
int alt_freq;
int alt_q;
int key_q;
int gold_q;
/* these parameters aren't to be used in final build don't use!!! */
int play_alternate;
int alt_freq;
int alt_q;
int key_q;
int gold_q;
int multi_threaded; /* how many threads to run the encoder on */
int token_partitions; /* how many token partitions to create */
int multi_threaded; /* how many threads to run the encoder on */
int token_partitions; /* how many token partitions to create */
/* early breakout threshold: for video conf recommend 800 */
int encode_breakout;
/* early breakout threshold: for video conf recommend 800 */
int encode_breakout;
/* Bitfield defining the error resiliency features to enable.
* Can provide decodable frames after losses in previous
* frames and decodable partitions after losses in the same frame.
*/
unsigned int error_resilient_mode;
/* Bitfield defining the error resiliency features to enable.
* Can provide decodable frames after losses in previous
* frames and decodable partitions after losses in the same frame.
*/
unsigned int error_resilient_mode;
int arnr_max_frames;
int arnr_strength;
int arnr_type;
int arnr_max_frames;
int arnr_strength;
int arnr_type;
vpx_fixed_buf_t two_pass_stats_in;
struct vpx_codec_pkt_list *output_pkt_list;
vpx_fixed_buf_t two_pass_stats_in;
struct vpx_codec_pkt_list *output_pkt_list;
vp8e_tuning tuning;
vp8e_tuning tuning;
/* Temporal scaling parameters */
unsigned int number_of_layers;
unsigned int target_bitrate[VPX_TS_MAX_PERIODICITY];
unsigned int rate_decimator[VPX_TS_MAX_PERIODICITY];
unsigned int periodicity;
unsigned int layer_id[VPX_TS_MAX_PERIODICITY];
/* Temporal scaling parameters */
unsigned int number_of_layers;
unsigned int target_bitrate[VPX_TS_MAX_PERIODICITY];
unsigned int rate_decimator[VPX_TS_MAX_PERIODICITY];
unsigned int periodicity;
unsigned int layer_id[VPX_TS_MAX_PERIODICITY];
#if CONFIG_MULTI_RES_ENCODING
/* Number of total resolutions encoded */
unsigned int mr_total_resolutions;
/* Number of total resolutions encoded */
unsigned int mr_total_resolutions;
/* Current encoder ID */
unsigned int mr_encoder_id;
/* Current encoder ID */
unsigned int mr_encoder_id;
/* Down-sampling factor */
vpx_rational_t mr_down_sampling_factor;
/* Down-sampling factor */
vpx_rational_t mr_down_sampling_factor;
/* Memory location to store low-resolution encoder's mode info */
void* mr_low_res_mode_info;
/* Memory location to store low-resolution encoder's mode info */
void *mr_low_res_mode_info;
#endif
} VP8_CONFIG;
} VP8_CONFIG;
void vp8_initialize();
void vp8_initialize();
struct VP8_COMP *vp8_create_compressor(VP8_CONFIG *oxcf);
void vp8_remove_compressor(struct VP8_COMP **comp);
struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf);
void vp8_remove_compressor(struct VP8_COMP* *comp);
void vp8_init_config(struct VP8_COMP *onyx, VP8_CONFIG *oxcf);
void vp8_change_config(struct VP8_COMP *onyx, VP8_CONFIG *oxcf);
void vp8_init_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
void vp8_change_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
int vp8_receive_raw_frame(struct VP8_COMP *comp, unsigned int frame_flags,
YV12_BUFFER_CONFIG *sd, int64_t time_stamp,
int64_t end_time_stamp);
int vp8_get_compressed_data(struct VP8_COMP *comp, unsigned int *frame_flags,
unsigned long *size, unsigned char *dest,
unsigned char *dest_end, int64_t *time_stamp,
int64_t *time_end, int flush);
int vp8_get_preview_raw_frame(struct VP8_COMP *comp, YV12_BUFFER_CONFIG *dest,
vp8_ppflags_t *flags);
int vp8_receive_raw_frame(struct VP8_COMP* comp, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time_stamp);
int vp8_get_compressed_data(struct VP8_COMP* comp, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush);
int vp8_get_preview_raw_frame(struct VP8_COMP* comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags);
int vp8_use_as_reference(struct VP8_COMP* comp, int ref_frame_flags);
int vp8_update_reference(struct VP8_COMP* comp, int ref_frame_flags);
int vp8_get_reference(struct VP8_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
int vp8_set_reference(struct VP8_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
int vp8_update_entropy(struct VP8_COMP* comp, int update);
int vp8_set_roimap(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4]);
int vp8_set_active_map(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols);
int vp8_set_internal_size(struct VP8_COMP* comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
int vp8_get_quantizer(struct VP8_COMP* c);
int vp8_use_as_reference(struct VP8_COMP *comp, int ref_frame_flags);
int vp8_update_reference(struct VP8_COMP *comp, int ref_frame_flags);
int vp8_get_reference(struct VP8_COMP *comp,
enum vpx_ref_frame_type ref_frame_flag,
YV12_BUFFER_CONFIG *sd);
int vp8_set_reference(struct VP8_COMP *comp,
enum vpx_ref_frame_type ref_frame_flag,
YV12_BUFFER_CONFIG *sd);
int vp8_update_entropy(struct VP8_COMP *comp, int update);
int vp8_set_roimap(struct VP8_COMP *comp, unsigned char *map, unsigned int rows,
unsigned int cols, int delta_q[4], int delta_lf[4],
unsigned int threshold[4]);
int vp8_set_active_map(struct VP8_COMP *comp, unsigned char *map,
unsigned int rows, unsigned int cols);
int vp8_set_internal_size(struct VP8_COMP *comp, VPX_SCALING horiz_mode,
VPX_SCALING vert_mode);
int vp8_get_quantizer(struct VP8_COMP *c);
#ifdef __cplusplus
}

180
vp8/common/onyxc_int.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ONYXC_INT_H_
#define VP8_COMMON_ONYXC_INT_H_
@ -38,144 +37,143 @@ extern "C" {
#define MAX_PARTITIONS 9
typedef struct frame_contexts
{
vp8_prob bmode_prob [VP8_BINTRAMODES-1];
vp8_prob ymode_prob [VP8_YMODES-1]; /* interframe intra mode probs */
vp8_prob uv_mode_prob [VP8_UV_MODES-1];
vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1];
vp8_prob coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES];
MV_CONTEXT mvc[2];
typedef struct frame_contexts {
vp8_prob bmode_prob[VP8_BINTRAMODES - 1];
vp8_prob ymode_prob[VP8_YMODES - 1]; /* interframe intra mode probs */
vp8_prob uv_mode_prob[VP8_UV_MODES - 1];
vp8_prob sub_mv_ref_prob[VP8_SUBMVREFS - 1];
vp8_prob
coef_probs[BLOCK_TYPES][COEF_BANDS][PREV_COEF_CONTEXTS][ENTROPY_NODES];
MV_CONTEXT mvc[2];
} FRAME_CONTEXT;
typedef enum
{
ONE_PARTITION = 0,
TWO_PARTITION = 1,
FOUR_PARTITION = 2,
EIGHT_PARTITION = 3
typedef enum {
ONE_PARTITION = 0,
TWO_PARTITION = 1,
FOUR_PARTITION = 2,
EIGHT_PARTITION = 3
} TOKEN_PARTITION;
typedef enum
{
RECON_CLAMP_REQUIRED = 0,
RECON_CLAMP_NOTREQUIRED = 1
typedef enum {
RECON_CLAMP_REQUIRED = 0,
RECON_CLAMP_NOTREQUIRED = 1
} CLAMP_TYPE;
typedef struct VP8Common
{
struct vpx_internal_error_info error;
{
struct vpx_internal_error_info error;
DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][2]);
DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][2]);
DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][2]);
DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][2]);
DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][2]);
DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][2]);
int Width;
int Height;
int horiz_scale;
int vert_scale;
int Width;
int Height;
int horiz_scale;
int vert_scale;
CLAMP_TYPE clamp_type;
CLAMP_TYPE clamp_type;
YV12_BUFFER_CONFIG *frame_to_show;
YV12_BUFFER_CONFIG *frame_to_show;
YV12_BUFFER_CONFIG yv12_fb[NUM_YV12_BUFFERS];
int fb_idx_ref_cnt[NUM_YV12_BUFFERS];
int new_fb_idx, lst_fb_idx, gld_fb_idx, alt_fb_idx;
YV12_BUFFER_CONFIG yv12_fb[NUM_YV12_BUFFERS];
int fb_idx_ref_cnt[NUM_YV12_BUFFERS];
int new_fb_idx, lst_fb_idx, gld_fb_idx, alt_fb_idx;
YV12_BUFFER_CONFIG temp_scale_frame;
YV12_BUFFER_CONFIG temp_scale_frame;
#if CONFIG_POSTPROC
YV12_BUFFER_CONFIG post_proc_buffer;
YV12_BUFFER_CONFIG post_proc_buffer_int;
int post_proc_buffer_int_used;
unsigned char *pp_limits_buffer; /* post-processing filter coefficients */
YV12_BUFFER_CONFIG post_proc_buffer;
YV12_BUFFER_CONFIG post_proc_buffer_int;
int post_proc_buffer_int_used;
unsigned char *pp_limits_buffer; /* post-processing filter coefficients */
#endif
FRAME_TYPE last_frame_type; /* Save last frame's frame type for motion search. */
FRAME_TYPE frame_type;
FRAME_TYPE
last_frame_type; /* Save last frame's frame type for motion search. */
FRAME_TYPE frame_type;
int show_frame;
int show_frame;
int frame_flags;
int MBs;
int mb_rows;
int mb_cols;
int mode_info_stride;
int frame_flags;
int MBs;
int mb_rows;
int mb_cols;
int mode_info_stride;
/* profile settings */
int mb_no_coeff_skip;
int no_lpf;
int use_bilinear_mc_filter;
int full_pixel;
/* profile settings */
int mb_no_coeff_skip;
int no_lpf;
int use_bilinear_mc_filter;
int full_pixel;
int base_qindex;
int base_qindex;
int y1dc_delta_q;
int y2dc_delta_q;
int y2ac_delta_q;
int uvdc_delta_q;
int uvac_delta_q;
int y1dc_delta_q;
int y2dc_delta_q;
int y2ac_delta_q;
int uvdc_delta_q;
int uvac_delta_q;
/* We allocate a MODE_INFO struct for each macroblock, together with
an extra row on top and column on the left to simplify prediction. */
/* We allocate a MODE_INFO struct for each macroblock, together with
an extra row on top and column on the left to simplify prediction. */
MODE_INFO *mip; /* Base of allocated array */
MODE_INFO *mi; /* Corresponds to upper left visible macroblock */
MODE_INFO *mip; /* Base of allocated array */
MODE_INFO *mi; /* Corresponds to upper left visible macroblock */
#if CONFIG_ERROR_CONCEALMENT
MODE_INFO *prev_mip; /* MODE_INFO array 'mip' from last decoded frame */
MODE_INFO *prev_mi; /* 'mi' from last frame (points into prev_mip) */
MODE_INFO *prev_mip; /* MODE_INFO array 'mip' from last decoded frame */
MODE_INFO *prev_mi; /* 'mi' from last frame (points into prev_mip) */
#endif
MODE_INFO *show_frame_mi; /* MODE_INFO for the last decoded frame
to show */
LOOPFILTERTYPE filter_type;
/* MODE_INFO for the last decoded frame to show */
MODE_INFO *show_frame_mi;
LOOPFILTERTYPE filter_type;
loop_filter_info_n lf_info;
loop_filter_info_n lf_info;
int filter_level;
int last_sharpness_level;
int sharpness_level;
int filter_level;
int last_sharpness_level;
int sharpness_level;
int refresh_last_frame; /* Two state 0 = NO, 1 = YES */
int refresh_golden_frame; /* Two state 0 = NO, 1 = YES */
int refresh_alt_ref_frame; /* Two state 0 = NO, 1 = YES */
int refresh_last_frame; /* Two state 0 = NO, 1 = YES */
int refresh_golden_frame; /* Two state 0 = NO, 1 = YES */
int refresh_alt_ref_frame; /* Two state 0 = NO, 1 = YES */
int copy_buffer_to_gf; /* 0 none, 1 Last to GF, 2 ARF to GF */
int copy_buffer_to_arf; /* 0 none, 1 Last to ARF, 2 GF to ARF */
int copy_buffer_to_gf; /* 0 none, 1 Last to GF, 2 ARF to GF */
int copy_buffer_to_arf; /* 0 none, 1 Last to ARF, 2 GF to ARF */
int refresh_entropy_probs; /* Two state 0 = NO, 1 = YES */
int refresh_entropy_probs; /* Two state 0 = NO, 1 = YES */
int ref_frame_sign_bias[MAX_REF_FRAMES]; /* Two state 0, 1 */
int ref_frame_sign_bias[MAX_REF_FRAMES]; /* Two state 0, 1 */
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context; /* row of context for each plane */
ENTROPY_CONTEXT_PLANES left_context; /* (up to) 4 contexts "" */
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context; /* row of context for each plane */
ENTROPY_CONTEXT_PLANES left_context; /* (up to) 4 contexts "" */
FRAME_CONTEXT lfc; /* last frame entropy */
FRAME_CONTEXT fc; /* this frame entropy */
FRAME_CONTEXT lfc; /* last frame entropy */
FRAME_CONTEXT fc; /* this frame entropy */
unsigned int current_video_frame;
unsigned int current_video_frame;
int version;
int version;
TOKEN_PARTITION multi_token_partition;
TOKEN_PARTITION multi_token_partition;
#ifdef PACKET_TESTING
VP8_HEADER oh;
VP8_HEADER oh;
#endif
#if CONFIG_POSTPROC_VISUALIZER
double bitrate;
double framerate;
double bitrate;
double framerate;
#endif
#if CONFIG_MULTITHREAD
int processor_core_count;
int processor_core_count;
#endif
#if CONFIG_POSTPROC
struct postproc_state postproc_state;
struct postproc_state postproc_state;
#endif
int cpu_caps;
int cpu_caps;
} VP8_COMMON;
#ifdef __cplusplus

53
vp8/common/onyxd.h
View File

@ -8,15 +8,12 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_ONYXD_H_
#define VP8_COMMON_ONYXD_H_
/* Create/destroy static data structures. */
#ifdef __cplusplus
extern "C"
{
extern "C" {
#endif
#include "vpx_scale/yv12config.h"
#include "ppflags.h"
@ -24,40 +21,40 @@ extern "C"
#include "vpx/vpx_codec.h"
#include "vpx/vp8.h"
struct VP8D_COMP;
struct VP8D_COMP;
typedef struct
{
int Width;
int Height;
int Version;
int postprocess;
int max_threads;
int error_concealment;
} VP8D_CONFIG;
typedef struct {
int Width;
int Height;
int Version;
int postprocess;
int max_threads;
int error_concealment;
} VP8D_CONFIG;
typedef enum
{
VP8D_OK = 0
} VP8D_SETTING;
typedef enum { VP8D_OK = 0 } VP8D_SETTING;
void vp8dx_initialize(void);
void vp8dx_initialize(void);
void vp8dx_set_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst, int x);
void vp8dx_set_setting(struct VP8D_COMP *comp, VP8D_SETTING oxst, int x);
int vp8dx_get_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst);
int vp8dx_get_setting(struct VP8D_COMP *comp, VP8D_SETTING oxst);
int vp8dx_receive_compressed_data(struct VP8D_COMP* comp,
size_t size, const uint8_t *dest,
int64_t time_stamp);
int vp8dx_get_raw_frame(struct VP8D_COMP* comp, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags);
int vp8dx_receive_compressed_data(struct VP8D_COMP *comp, size_t size,
const uint8_t *dest, int64_t time_stamp);
int vp8dx_get_raw_frame(struct VP8D_COMP *comp, YV12_BUFFER_CONFIG *sd,
int64_t *time_stamp, int64_t *time_end_stamp,
vp8_ppflags_t *flags);
vpx_codec_err_t vp8dx_get_reference(struct VP8D_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
vpx_codec_err_t vp8dx_set_reference(struct VP8D_COMP* comp, enum vpx_ref_frame_type ref_frame_flag, YV12_BUFFER_CONFIG *sd);
vpx_codec_err_t vp8dx_get_reference(struct VP8D_COMP *comp,
enum vpx_ref_frame_type ref_frame_flag,
YV12_BUFFER_CONFIG *sd);
vpx_codec_err_t vp8dx_set_reference(struct VP8D_COMP *comp,
enum vpx_ref_frame_type ref_frame_flag,
YV12_BUFFER_CONFIG *sd);
#ifdef __cplusplus
}
#endif
#endif // VP8_COMMON_ONYXD_H_

1454
vp8/common/postproc.c
View File

File diff suppressed because it is too large Load Diff

35
vp8/common/postproc.h
View File

@ -8,19 +8,17 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_POSTPROC_H_
#define VP8_COMMON_POSTPROC_H_
#include "vpx_ports/mem.h"
struct postproc_state
{
int last_q;
int last_noise;
int last_base_qindex;
int last_frame_valid;
int clamp;
int8_t *generated_noise;
struct postproc_state {
int last_q;
int last_noise;
int last_base_qindex;
int last_frame_valid;
int clamp;
int8_t *generated_noise;
};
#include "onyxc_int.h"
#include "ppflags.h"
@ -31,21 +29,12 @@ extern "C" {
int vp8_post_proc_frame(struct VP8Common *oci, YV12_BUFFER_CONFIG *dest,
vp8_ppflags_t *flags);
void vp8_de_noise(struct VP8Common *oci, YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *post, int q, int low_var_thresh, int flag,
int uvfilter);
void vp8_de_noise(struct VP8Common *oci,
YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *post,
int q,
int low_var_thresh,
int flag,
int uvfilter);
void vp8_deblock(struct VP8Common *oci,
YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *post,
int q,
int low_var_thresh,
int flag);
void vp8_deblock(struct VP8Common *oci, YV12_BUFFER_CONFIG *source,
YV12_BUFFER_CONFIG *post, int q, int low_var_thresh, int flag);
#define MFQE_PRECISION 4

45
vp8/common/ppflags.h
View File

@ -8,38 +8,35 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_PPFLAGS_H_
#define VP8_COMMON_PPFLAGS_H_
#ifdef __cplusplus
extern "C" {
#endif
enum
{
VP8D_NOFILTERING = 0,
VP8D_DEBLOCK = 1<<0,
VP8D_DEMACROBLOCK = 1<<1,
VP8D_ADDNOISE = 1<<2,
VP8D_DEBUG_TXT_FRAME_INFO = 1<<3,
VP8D_DEBUG_TXT_MBLK_MODES = 1<<4,
VP8D_DEBUG_TXT_DC_DIFF = 1<<5,
VP8D_DEBUG_TXT_RATE_INFO = 1<<6,
VP8D_DEBUG_DRAW_MV = 1<<7,
VP8D_DEBUG_CLR_BLK_MODES = 1<<8,
VP8D_DEBUG_CLR_FRM_REF_BLKS = 1<<9,
VP8D_MFQE = 1<<10
enum {
VP8D_NOFILTERING = 0,
VP8D_DEBLOCK = 1 << 0,
VP8D_DEMACROBLOCK = 1 << 1,
VP8D_ADDNOISE = 1 << 2,
VP8D_DEBUG_TXT_FRAME_INFO = 1 << 3,
VP8D_DEBUG_TXT_MBLK_MODES = 1 << 4,
VP8D_DEBUG_TXT_DC_DIFF = 1 << 5,
VP8D_DEBUG_TXT_RATE_INFO = 1 << 6,
VP8D_DEBUG_DRAW_MV = 1 << 7,
VP8D_DEBUG_CLR_BLK_MODES = 1 << 8,
VP8D_DEBUG_CLR_FRM_REF_BLKS = 1 << 9,
VP8D_MFQE = 1 << 10
};
typedef struct
{
int post_proc_flag;
int deblocking_level;
int noise_level;
int display_ref_frame_flag;
int display_mb_modes_flag;
int display_b_modes_flag;
int display_mv_flag;
typedef struct {
int post_proc_flag;
int deblocking_level;
int noise_level;
int display_ref_frame_flag;
int display_mb_modes_flag;
int display_b_modes_flag;
int display_mv_flag;
} vp8_ppflags_t;
#ifdef __cplusplus

167
vp8/common/quant_common.c
View File

@ -8,128 +8,117 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "quant_common.h"
static const int dc_qlookup[QINDEX_RANGE] =
{
4, 5, 6, 7, 8, 9, 10, 10, 11, 12, 13, 14, 15, 16, 17, 17,
18, 19, 20, 20, 21, 21, 22, 22, 23, 23, 24, 25, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43,
44, 45, 46, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58,
59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74,
75, 76, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89,
91, 93, 95, 96, 98, 100, 101, 102, 104, 106, 108, 110, 112, 114, 116, 118,
122, 124, 126, 128, 130, 132, 134, 136, 138, 140, 143, 145, 148, 151, 154, 157,
static const int dc_qlookup[QINDEX_RANGE] = {
4, 5, 6, 7, 8, 9, 10, 10, 11, 12, 13, 14, 15, 16, 17,
17, 18, 19, 20, 20, 21, 21, 22, 22, 23, 23, 24, 25, 25, 26,
27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40,
41, 42, 43, 44, 45, 46, 46, 47, 48, 49, 50, 51, 52, 53, 54,
55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69,
70, 71, 72, 73, 74, 75, 76, 76, 77, 78, 79, 80, 81, 82, 83,
84, 85, 86, 87, 88, 89, 91, 93, 95, 96, 98, 100, 101, 102, 104,
106, 108, 110, 112, 114, 116, 118, 122, 124, 126, 128, 130, 132, 134, 136,
138, 140, 143, 145, 148, 151, 154, 157,
};
static const int ac_qlookup[QINDEX_RANGE] =
{
4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51,
52, 53, 54, 55, 56, 57, 58, 60, 62, 64, 66, 68, 70, 72, 74, 76,
78, 80, 82, 84, 86, 88, 90, 92, 94, 96, 98, 100, 102, 104, 106, 108,
110, 112, 114, 116, 119, 122, 125, 128, 131, 134, 137, 140, 143, 146, 149, 152,
155, 158, 161, 164, 167, 170, 173, 177, 181, 185, 189, 193, 197, 201, 205, 209,
213, 217, 221, 225, 229, 234, 239, 245, 249, 254, 259, 264, 269, 274, 279, 284,
static const int ac_qlookup[QINDEX_RANGE] = {
4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33,
34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48,
49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 60, 62, 64, 66, 68,
70, 72, 74, 76, 78, 80, 82, 84, 86, 88, 90, 92, 94, 96, 98,
100, 102, 104, 106, 108, 110, 112, 114, 116, 119, 122, 125, 128, 131, 134,
137, 140, 143, 146, 149, 152, 155, 158, 161, 164, 167, 170, 173, 177, 181,
185, 189, 193, 197, 201, 205, 209, 213, 217, 221, 225, 229, 234, 239, 245,
249, 254, 259, 264, 269, 274, 279, 284,
};
int vp8_dc_quant(int QIndex, int Delta) {
int retval;
int vp8_dc_quant(int QIndex, int Delta)
{
int retval;
QIndex = QIndex + Delta;
QIndex = QIndex + Delta;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = dc_qlookup[ QIndex ];
return retval;
retval = dc_qlookup[QIndex];
return retval;
}
int vp8_dc2quant(int QIndex, int Delta)
{
int retval;
int vp8_dc2quant(int QIndex, int Delta) {
int retval;
QIndex = QIndex + Delta;
QIndex = QIndex + Delta;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = dc_qlookup[ QIndex ] * 2;
return retval;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = dc_qlookup[QIndex] * 2;
return retval;
}
int vp8_dc_uv_quant(int QIndex, int Delta)
{
int retval;
int vp8_dc_uv_quant(int QIndex, int Delta) {
int retval;
QIndex = QIndex + Delta;
QIndex = QIndex + Delta;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = dc_qlookup[ QIndex ];
retval = dc_qlookup[QIndex];
if (retval > 132)
retval = 132;
if (retval > 132) retval = 132;
return retval;
return retval;
}
int vp8_ac_yquant(int QIndex)
{
int retval;
int vp8_ac_yquant(int QIndex) {
int retval;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = ac_qlookup[ QIndex ];
return retval;
retval = ac_qlookup[QIndex];
return retval;
}
int vp8_ac2quant(int QIndex, int Delta)
{
int retval;
int vp8_ac2quant(int QIndex, int Delta) {
int retval;
QIndex = QIndex + Delta;
QIndex = QIndex + Delta;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
/* For all x in [0..284], x*155/100 is bitwise equal to (x*101581) >> 16.
* The smallest precision for that is '(x*6349) >> 12' but 16 is a good
* word size. */
retval = (ac_qlookup[ QIndex ] * 101581) >> 16;
/* For all x in [0..284], x*155/100 is bitwise equal to (x*101581) >> 16.
* The smallest precision for that is '(x*6349) >> 12' but 16 is a good
* word size. */
retval = (ac_qlookup[QIndex] * 101581) >> 16;
if (retval < 8)
retval = 8;
if (retval < 8) retval = 8;
return retval;
return retval;
}
int vp8_ac_uv_quant(int QIndex, int Delta)
{
int retval;
int vp8_ac_uv_quant(int QIndex, int Delta) {
int retval;
QIndex = QIndex + Delta;
QIndex = QIndex + Delta;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
if (QIndex > 127)
QIndex = 127;
else if (QIndex < 0)
QIndex = 0;
retval = ac_qlookup[ QIndex ];
return retval;
retval = ac_qlookup[QIndex];
return retval;
}

1
vp8/common/quant_common.h
View File

@ -11,7 +11,6 @@
#ifndef VP8_COMMON_QUANT_COMMON_H_
#define VP8_COMMON_QUANT_COMMON_H_
#include "string.h"
#include "blockd.h"
#include "onyxc_int.h"

829
vp8/common/reconinter.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include <limits.h>
#include <string.h>
@ -21,524 +20,474 @@
#include "onyxc_int.h"
#endif
void vp8_copy_mem16x16_c(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride)
{
void vp8_copy_mem16x16_c(unsigned char *src, int src_stride, unsigned char *dst,
int dst_stride) {
int r;
int r;
for (r = 0; r < 16; r++)
{
memcpy(dst, src, 16);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 16; r++) {
memcpy(dst, src, 16);
src += src_stride;
dst += dst_stride;
}
}
void vp8_copy_mem8x8_c(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride)
{
int r;
void vp8_copy_mem8x8_c(unsigned char *src, int src_stride, unsigned char *dst,
int dst_stride) {
int r;
for (r = 0; r < 8; r++)
{
memcpy(dst, src, 8);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 8; r++) {
memcpy(dst, src, 8);
src += src_stride;
dst += dst_stride;
}
}
void vp8_copy_mem8x4_c(
unsigned char *src,
int src_stride,
unsigned char *dst,
int dst_stride)
{
int r;
void vp8_copy_mem8x4_c(unsigned char *src, int src_stride, unsigned char *dst,
int dst_stride) {
int r;
for (r = 0; r < 4; r++)
{
memcpy(dst, src, 8);
src += src_stride;
dst += dst_stride;
}
for (r = 0; r < 4; r++) {
memcpy(dst, src, 8);
src += src_stride;
dst += dst_stride;
}
}
void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, unsigned char *base_pre,
int pre_stride, vp8_subpix_fn_t sppf) {
int r;
unsigned char *pred_ptr = d->predictor;
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride +
(d->bmi.mv.as_mv.col >> 3);
void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, unsigned char *base_pre, int pre_stride, vp8_subpix_fn_t sppf)
{
int r;
unsigned char *pred_ptr = d->predictor;
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
{
sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, pred_ptr, pitch);
}
else
{
for (r = 0; r < 4; r++)
{
pred_ptr[0] = ptr[0];
pred_ptr[1] = ptr[1];
pred_ptr[2] = ptr[2];
pred_ptr[3] = ptr[3];
pred_ptr += pitch;
ptr += pre_stride;
}
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7) {
sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7,
pred_ptr, pitch);
} else {
for (r = 0; r < 4; r++) {
pred_ptr[0] = ptr[0];
pred_ptr[1] = ptr[1];
pred_ptr[2] = ptr[2];
pred_ptr[3] = ptr[3];
pred_ptr += pitch;
ptr += pre_stride;
}
}
}
static void build_inter_predictors4b(MACROBLOCKD *x, BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride)
{
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
static void build_inter_predictors4b(MACROBLOCKD *x, BLOCKD *d,
unsigned char *dst, int dst_stride,
unsigned char *base_pre, int pre_stride) {
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride +
(d->bmi.mv.as_mv.col >> 3);
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
{
x->subpixel_predict8x8(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
}
else
{
vp8_copy_mem8x8(ptr, pre_stride, dst, dst_stride);
}
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7) {
x->subpixel_predict8x8(ptr, pre_stride, d->bmi.mv.as_mv.col & 7,
d->bmi.mv.as_mv.row & 7, dst, dst_stride);
} else {
vp8_copy_mem8x8(ptr, pre_stride, dst, dst_stride);
}
}
static void build_inter_predictors2b(MACROBLOCKD *x, BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride)
{
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
static void build_inter_predictors2b(MACROBLOCKD *x, BLOCKD *d,
unsigned char *dst, int dst_stride,
unsigned char *base_pre, int pre_stride) {
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride +
(d->bmi.mv.as_mv.col >> 3);
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
{
x->subpixel_predict8x4(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
}
else
{
vp8_copy_mem8x4(ptr, pre_stride, dst, dst_stride);
}
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7) {
x->subpixel_predict8x4(ptr, pre_stride, d->bmi.mv.as_mv.col & 7,
d->bmi.mv.as_mv.row & 7, dst, dst_stride);
} else {
vp8_copy_mem8x4(ptr, pre_stride, dst, dst_stride);
}
}
static void build_inter_predictors_b(BLOCKD *d, unsigned char *dst, int dst_stride, unsigned char *base_pre, int pre_stride, vp8_subpix_fn_t sppf)
{
int r;
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride + (d->bmi.mv.as_mv.col >> 3);
static void build_inter_predictors_b(BLOCKD *d, unsigned char *dst,
int dst_stride, unsigned char *base_pre,
int pre_stride, vp8_subpix_fn_t sppf) {
int r;
unsigned char *ptr;
ptr = base_pre + d->offset + (d->bmi.mv.as_mv.row >> 3) * pre_stride +
(d->bmi.mv.as_mv.col >> 3);
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
{
sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst, dst_stride);
}
else
{
for (r = 0; r < 4; r++)
{
dst[0] = ptr[0];
dst[1] = ptr[1];
dst[2] = ptr[2];
dst[3] = ptr[3];
dst += dst_stride;
ptr += pre_stride;
}
}
}
/*encoder only*/
void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x)
{
unsigned char *uptr, *vptr;
unsigned char *upred_ptr = &x->predictor[256];
unsigned char *vpred_ptr = &x->predictor[320];
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int offset;
int pre_stride = x->pre.uv_stride;
/* calc uv motion vectors */
mv_row += 1 | (mv_row >> (sizeof(int) * CHAR_BIT - 1));
mv_col += 1 | (mv_col >> (sizeof(int) * CHAR_BIT - 1));
mv_row /= 2;
mv_col /= 2;
mv_row &= x->fullpixel_mask;
mv_col &= x->fullpixel_mask;
offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
uptr = x->pre.u_buffer + offset;
vptr = x->pre.v_buffer + offset;
if ((mv_row | mv_col) & 7)
{
x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, 8);
x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, 8);
}
else
{
vp8_copy_mem8x8(uptr, pre_stride, upred_ptr, 8);
vp8_copy_mem8x8(vptr, pre_stride, vpred_ptr, 8);
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7) {
sppf(ptr, pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst,
dst_stride);
} else {
for (r = 0; r < 4; r++) {
dst[0] = ptr[0];
dst[1] = ptr[1];
dst[2] = ptr[2];
dst[3] = ptr[3];
dst += dst_stride;
ptr += pre_stride;
}
}
}
/*encoder only*/
void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x)
{
int i, j;
int pre_stride = x->pre.uv_stride;
unsigned char *base_pre;
void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x) {
unsigned char *uptr, *vptr;
unsigned char *upred_ptr = &x->predictor[256];
unsigned char *vpred_ptr = &x->predictor[320];
/* build uv mvs */
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
int yoffset = i * 8 + j * 2;
int uoffset = 16 + i * 2 + j;
int voffset = 20 + i * 2 + j;
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int offset;
int pre_stride = x->pre.uv_stride;
int temp;
/* calc uv motion vectors */
mv_row += 1 | (mv_row >> (sizeof(int) * CHAR_BIT - 1));
mv_col += 1 | (mv_col >> (sizeof(int) * CHAR_BIT - 1));
mv_row /= 2;
mv_col /= 2;
mv_row &= x->fullpixel_mask;
mv_col &= x->fullpixel_mask;
temp = x->block[yoffset ].bmi.mv.as_mv.row
+ x->block[yoffset+1].bmi.mv.as_mv.row
+ x->block[yoffset+4].bmi.mv.as_mv.row
+ x->block[yoffset+5].bmi.mv.as_mv.row;
offset = (mv_row >> 3) * pre_stride + (mv_col >> 3);
uptr = x->pre.u_buffer + offset;
vptr = x->pre.v_buffer + offset;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
temp = x->block[yoffset ].bmi.mv.as_mv.col
+ x->block[yoffset+1].bmi.mv.as_mv.col
+ x->block[yoffset+4].bmi.mv.as_mv.col
+ x->block[yoffset+5].bmi.mv.as_mv.col;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
}
}
base_pre = x->pre.u_buffer;
for (i = 16; i < 20; i += 2)
{
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i+1];
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
else
{
vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride, x->subpixel_predict);
vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride, x->subpixel_predict);
}
}
base_pre = x->pre.v_buffer;
for (i = 20; i < 24; i += 2)
{
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i+1];
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
else
{
vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride, x->subpixel_predict);
vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride, x->subpixel_predict);
}
}
if ((mv_row | mv_col) & 7) {
x->subpixel_predict8x8(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr,
8);
x->subpixel_predict8x8(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr,
8);
} else {
vp8_copy_mem8x8(uptr, pre_stride, upred_ptr, 8);
vp8_copy_mem8x8(vptr, pre_stride, vpred_ptr, 8);
}
}
/*encoder only*/
void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
unsigned char *dst_y,
int dst_ystride)
{
unsigned char *ptr_base;
unsigned char *ptr;
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int pre_stride = x->pre.y_stride;
void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x) {
int i, j;
int pre_stride = x->pre.uv_stride;
unsigned char *base_pre;
ptr_base = x->pre.y_buffer;
ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
/* build uv mvs */
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
int yoffset = i * 8 + j * 2;
int uoffset = 16 + i * 2 + j;
int voffset = 20 + i * 2 + j;
if ((mv_row | mv_col) & 7)
{
x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7,
dst_y, dst_ystride);
int temp;
temp = x->block[yoffset].bmi.mv.as_mv.row +
x->block[yoffset + 1].bmi.mv.as_mv.row +
x->block[yoffset + 4].bmi.mv.as_mv.row +
x->block[yoffset + 5].bmi.mv.as_mv.row;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
temp = x->block[yoffset].bmi.mv.as_mv.col +
x->block[yoffset + 1].bmi.mv.as_mv.col +
x->block[yoffset + 4].bmi.mv.as_mv.col +
x->block[yoffset + 5].bmi.mv.as_mv.col;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
}
else
{
vp8_copy_mem16x16(ptr, pre_stride, dst_y,
dst_ystride);
}
base_pre = x->pre.u_buffer;
for (i = 16; i < 20; i += 2) {
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i + 1];
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
else {
vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride,
x->subpixel_predict);
vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride,
x->subpixel_predict);
}
}
base_pre = x->pre.v_buffer;
for (i = 20; i < 24; i += 2) {
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i + 1];
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, d0->predictor, 8, base_pre, pre_stride);
else {
vp8_build_inter_predictors_b(d0, 8, base_pre, pre_stride,
x->subpixel_predict);
vp8_build_inter_predictors_b(d1, 8, base_pre, pre_stride,
x->subpixel_predict);
}
}
}
static void clamp_mv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
{
/* If the MV points so far into the UMV border that no visible pixels
* are used for reconstruction, the subpel part of the MV can be
* discarded and the MV limited to 16 pixels with equivalent results.
*
* This limit kicks in at 19 pixels for the top and left edges, for
* the 16 pixels plus 3 taps right of the central pixel when subpel
* filtering. The bottom and right edges use 16 pixels plus 2 pixels
* left of the central pixel when filtering.
*/
if (mv->col < (xd->mb_to_left_edge - (19 << 3)))
mv->col = xd->mb_to_left_edge - (16 << 3);
else if (mv->col > xd->mb_to_right_edge + (18 << 3))
mv->col = xd->mb_to_right_edge + (16 << 3);
/*encoder only*/
void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x, unsigned char *dst_y,
int dst_ystride) {
unsigned char *ptr_base;
unsigned char *ptr;
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int pre_stride = x->pre.y_stride;
if (mv->row < (xd->mb_to_top_edge - (19 << 3)))
mv->row = xd->mb_to_top_edge - (16 << 3);
else if (mv->row > xd->mb_to_bottom_edge + (18 << 3))
mv->row = xd->mb_to_bottom_edge + (16 << 3);
ptr_base = x->pre.y_buffer;
ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
if ((mv_row | mv_col) & 7) {
x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, dst_y,
dst_ystride);
} else {
vp8_copy_mem16x16(ptr, pre_stride, dst_y, dst_ystride);
}
}
static void clamp_mv_to_umv_border(MV *mv, const MACROBLOCKD *xd) {
/* If the MV points so far into the UMV border that no visible pixels
* are used for reconstruction, the subpel part of the MV can be
* discarded and the MV limited to 16 pixels with equivalent results.
*
* This limit kicks in at 19 pixels for the top and left edges, for
* the 16 pixels plus 3 taps right of the central pixel when subpel
* filtering. The bottom and right edges use 16 pixels plus 2 pixels
* left of the central pixel when filtering.
*/
if (mv->col < (xd->mb_to_left_edge - (19 << 3)))
mv->col = xd->mb_to_left_edge - (16 << 3);
else if (mv->col > xd->mb_to_right_edge + (18 << 3))
mv->col = xd->mb_to_right_edge + (16 << 3);
if (mv->row < (xd->mb_to_top_edge - (19 << 3)))
mv->row = xd->mb_to_top_edge - (16 << 3);
else if (mv->row > xd->mb_to_bottom_edge + (18 << 3))
mv->row = xd->mb_to_bottom_edge + (16 << 3);
}
/* A version of the above function for chroma block MVs.*/
static void clamp_uvmv_to_umv_border(MV *mv, const MACROBLOCKD *xd)
{
mv->col = (2*mv->col < (xd->mb_to_left_edge - (19 << 3))) ?
(xd->mb_to_left_edge - (16 << 3)) >> 1 : mv->col;
mv->col = (2*mv->col > xd->mb_to_right_edge + (18 << 3)) ?
(xd->mb_to_right_edge + (16 << 3)) >> 1 : mv->col;
static void clamp_uvmv_to_umv_border(MV *mv, const MACROBLOCKD *xd) {
mv->col = (2 * mv->col < (xd->mb_to_left_edge - (19 << 3)))
? (xd->mb_to_left_edge - (16 << 3)) >> 1
: mv->col;
mv->col = (2 * mv->col > xd->mb_to_right_edge + (18 << 3))
? (xd->mb_to_right_edge + (16 << 3)) >> 1
: mv->col;
mv->row = (2*mv->row < (xd->mb_to_top_edge - (19 << 3))) ?
(xd->mb_to_top_edge - (16 << 3)) >> 1 : mv->row;
mv->row = (2*mv->row > xd->mb_to_bottom_edge + (18 << 3)) ?
(xd->mb_to_bottom_edge + (16 << 3)) >> 1 : mv->row;
mv->row = (2 * mv->row < (xd->mb_to_top_edge - (19 << 3)))
? (xd->mb_to_top_edge - (16 << 3)) >> 1
: mv->row;
mv->row = (2 * mv->row > xd->mb_to_bottom_edge + (18 << 3))
? (xd->mb_to_bottom_edge + (16 << 3)) >> 1
: mv->row;
}
void vp8_build_inter16x16_predictors_mb(MACROBLOCKD *x,
unsigned char *dst_y,
void vp8_build_inter16x16_predictors_mb(MACROBLOCKD *x, unsigned char *dst_y,
unsigned char *dst_u,
unsigned char *dst_v,
int dst_ystride,
int dst_uvstride)
{
int offset;
unsigned char *ptr;
unsigned char *uptr, *vptr;
unsigned char *dst_v, int dst_ystride,
int dst_uvstride) {
int offset;
unsigned char *ptr;
unsigned char *uptr, *vptr;
int_mv _16x16mv;
int_mv _16x16mv;
unsigned char *ptr_base = x->pre.y_buffer;
int pre_stride = x->pre.y_stride;
unsigned char *ptr_base = x->pre.y_buffer;
int pre_stride = x->pre.y_stride;
_16x16mv.as_int = x->mode_info_context->mbmi.mv.as_int;
_16x16mv.as_int = x->mode_info_context->mbmi.mv.as_int;
if (x->mode_info_context->mbmi.need_to_clamp_mvs)
{
clamp_mv_to_umv_border(&_16x16mv.as_mv, x);
}
if (x->mode_info_context->mbmi.need_to_clamp_mvs) {
clamp_mv_to_umv_border(&_16x16mv.as_mv, x);
}
ptr = ptr_base + ( _16x16mv.as_mv.row >> 3) * pre_stride + (_16x16mv.as_mv.col >> 3);
ptr = ptr_base + (_16x16mv.as_mv.row >> 3) * pre_stride +
(_16x16mv.as_mv.col >> 3);
if ( _16x16mv.as_int & 0x00070007)
{
x->subpixel_predict16x16(ptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_y, dst_ystride);
}
else
{
vp8_copy_mem16x16(ptr, pre_stride, dst_y, dst_ystride);
}
if (_16x16mv.as_int & 0x00070007) {
x->subpixel_predict16x16(ptr, pre_stride, _16x16mv.as_mv.col & 7,
_16x16mv.as_mv.row & 7, dst_y, dst_ystride);
} else {
vp8_copy_mem16x16(ptr, pre_stride, dst_y, dst_ystride);
}
/* calc uv motion vectors */
_16x16mv.as_mv.row += 1 | (_16x16mv.as_mv.row >> (sizeof(int) * CHAR_BIT - 1));
_16x16mv.as_mv.col += 1 | (_16x16mv.as_mv.col >> (sizeof(int) * CHAR_BIT - 1));
_16x16mv.as_mv.row /= 2;
_16x16mv.as_mv.col /= 2;
_16x16mv.as_mv.row &= x->fullpixel_mask;
_16x16mv.as_mv.col &= x->fullpixel_mask;
/* calc uv motion vectors */
_16x16mv.as_mv.row +=
1 | (_16x16mv.as_mv.row >> (sizeof(int) * CHAR_BIT - 1));
_16x16mv.as_mv.col +=
1 | (_16x16mv.as_mv.col >> (sizeof(int) * CHAR_BIT - 1));
_16x16mv.as_mv.row /= 2;
_16x16mv.as_mv.col /= 2;
_16x16mv.as_mv.row &= x->fullpixel_mask;
_16x16mv.as_mv.col &= x->fullpixel_mask;
pre_stride >>= 1;
offset = ( _16x16mv.as_mv.row >> 3) * pre_stride + (_16x16mv.as_mv.col >> 3);
uptr = x->pre.u_buffer + offset;
vptr = x->pre.v_buffer + offset;
pre_stride >>= 1;
offset = (_16x16mv.as_mv.row >> 3) * pre_stride + (_16x16mv.as_mv.col >> 3);
uptr = x->pre.u_buffer + offset;
vptr = x->pre.v_buffer + offset;
if ( _16x16mv.as_int & 0x00070007)
{
x->subpixel_predict8x8(uptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_u, dst_uvstride);
x->subpixel_predict8x8(vptr, pre_stride, _16x16mv.as_mv.col & 7, _16x16mv.as_mv.row & 7, dst_v, dst_uvstride);
}
else
{
vp8_copy_mem8x8(uptr, pre_stride, dst_u, dst_uvstride);
vp8_copy_mem8x8(vptr, pre_stride, dst_v, dst_uvstride);
}
if (_16x16mv.as_int & 0x00070007) {
x->subpixel_predict8x8(uptr, pre_stride, _16x16mv.as_mv.col & 7,
_16x16mv.as_mv.row & 7, dst_u, dst_uvstride);
x->subpixel_predict8x8(vptr, pre_stride, _16x16mv.as_mv.col & 7,
_16x16mv.as_mv.row & 7, dst_v, dst_uvstride);
} else {
vp8_copy_mem8x8(uptr, pre_stride, dst_u, dst_uvstride);
vp8_copy_mem8x8(vptr, pre_stride, dst_v, dst_uvstride);
}
}
static void build_inter4x4_predictors_mb(MACROBLOCKD *x)
{
int i;
unsigned char *base_dst = x->dst.y_buffer;
unsigned char *base_pre = x->pre.y_buffer;
static void build_inter4x4_predictors_mb(MACROBLOCKD *x) {
int i;
unsigned char *base_dst = x->dst.y_buffer;
unsigned char *base_pre = x->pre.y_buffer;
if (x->mode_info_context->mbmi.partitioning < 3)
{
BLOCKD *b;
int dst_stride = x->dst.y_stride;
if (x->mode_info_context->mbmi.partitioning < 3) {
BLOCKD *b;
int dst_stride = x->dst.y_stride;
x->block[ 0].bmi = x->mode_info_context->bmi[ 0];
x->block[ 2].bmi = x->mode_info_context->bmi[ 2];
x->block[ 8].bmi = x->mode_info_context->bmi[ 8];
x->block[10].bmi = x->mode_info_context->bmi[10];
if (x->mode_info_context->mbmi.need_to_clamp_mvs)
{
clamp_mv_to_umv_border(&x->block[ 0].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[ 2].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[ 8].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[10].bmi.mv.as_mv, x);
}
b = &x->block[ 0];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
b = &x->block[ 2];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
b = &x->block[ 8];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
b = &x->block[10];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre, dst_stride);
}
else
{
for (i = 0; i < 16; i += 2)
{
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i+1];
int dst_stride = x->dst.y_stride;
x->block[i+0].bmi = x->mode_info_context->bmi[i+0];
x->block[i+1].bmi = x->mode_info_context->bmi[i+1];
if (x->mode_info_context->mbmi.need_to_clamp_mvs)
{
clamp_mv_to_umv_border(&x->block[i+0].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[i+1].bmi.mv.as_mv, x);
}
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
else
{
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
}
}
}
base_dst = x->dst.u_buffer;
base_pre = x->pre.u_buffer;
for (i = 16; i < 20; i += 2)
{
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i+1];
int dst_stride = x->dst.uv_stride;
/* Note: uv mvs already clamped in build_4x4uvmvs() */
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
else
{
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
}
x->block[0].bmi = x->mode_info_context->bmi[0];
x->block[2].bmi = x->mode_info_context->bmi[2];
x->block[8].bmi = x->mode_info_context->bmi[8];
x->block[10].bmi = x->mode_info_context->bmi[10];
if (x->mode_info_context->mbmi.need_to_clamp_mvs) {
clamp_mv_to_umv_border(&x->block[0].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[2].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[8].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[10].bmi.mv.as_mv, x);
}
base_dst = x->dst.v_buffer;
base_pre = x->pre.v_buffer;
for (i = 20; i < 24; i += 2)
{
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i+1];
int dst_stride = x->dst.uv_stride;
b = &x->block[0];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre,
dst_stride);
b = &x->block[2];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre,
dst_stride);
b = &x->block[8];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre,
dst_stride);
b = &x->block[10];
build_inter_predictors4b(x, b, base_dst + b->offset, dst_stride, base_pre,
dst_stride);
} else {
for (i = 0; i < 16; i += 2) {
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i + 1];
int dst_stride = x->dst.y_stride;
/* Note: uv mvs already clamped in build_4x4uvmvs() */
x->block[i + 0].bmi = x->mode_info_context->bmi[i + 0];
x->block[i + 1].bmi = x->mode_info_context->bmi[i + 1];
if (x->mode_info_context->mbmi.need_to_clamp_mvs) {
clamp_mv_to_umv_border(&x->block[i + 0].bmi.mv.as_mv, x);
clamp_mv_to_umv_border(&x->block[i + 1].bmi.mv.as_mv, x);
}
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride);
else
{
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre, dst_stride, x->subpixel_predict);
}
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride,
base_pre, dst_stride);
else {
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride,
base_pre, dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride,
base_pre, dst_stride, x->subpixel_predict);
}
}
}
base_dst = x->dst.u_buffer;
base_pre = x->pre.u_buffer;
for (i = 16; i < 20; i += 2) {
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i + 1];
int dst_stride = x->dst.uv_stride;
/* Note: uv mvs already clamped in build_4x4uvmvs() */
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride,
base_pre, dst_stride);
else {
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre,
dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre,
dst_stride, x->subpixel_predict);
}
}
base_dst = x->dst.v_buffer;
base_pre = x->pre.v_buffer;
for (i = 20; i < 24; i += 2) {
BLOCKD *d0 = &x->block[i];
BLOCKD *d1 = &x->block[i + 1];
int dst_stride = x->dst.uv_stride;
/* Note: uv mvs already clamped in build_4x4uvmvs() */
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
build_inter_predictors2b(x, d0, base_dst + d0->offset, dst_stride,
base_pre, dst_stride);
else {
build_inter_predictors_b(d0, base_dst + d0->offset, dst_stride, base_pre,
dst_stride, x->subpixel_predict);
build_inter_predictors_b(d1, base_dst + d1->offset, dst_stride, base_pre,
dst_stride, x->subpixel_predict);
}
}
}
static
void build_4x4uvmvs(MACROBLOCKD *x)
{
int i, j;
static void build_4x4uvmvs(MACROBLOCKD *x) {
int i, j;
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
int yoffset = i * 8 + j * 2;
int uoffset = 16 + i * 2 + j;
int voffset = 20 + i * 2 + j;
for (i = 0; i < 2; i++) {
for (j = 0; j < 2; j++) {
int yoffset = i * 8 + j * 2;
int uoffset = 16 + i * 2 + j;
int voffset = 20 + i * 2 + j;
int temp;
int temp;
temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.row
+ x->mode_info_context->bmi[yoffset + 1].mv.as_mv.row
+ x->mode_info_context->bmi[yoffset + 4].mv.as_mv.row
+ x->mode_info_context->bmi[yoffset + 5].mv.as_mv.row;
temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.row +
x->mode_info_context->bmi[yoffset + 1].mv.as_mv.row +
x->mode_info_context->bmi[yoffset + 4].mv.as_mv.row +
x->mode_info_context->bmi[yoffset + 5].mv.as_mv.row;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
x->block[uoffset].bmi.mv.as_mv.row = (temp / 8) & x->fullpixel_mask;
temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.col
+ x->mode_info_context->bmi[yoffset + 1].mv.as_mv.col
+ x->mode_info_context->bmi[yoffset + 4].mv.as_mv.col
+ x->mode_info_context->bmi[yoffset + 5].mv.as_mv.col;
temp = x->mode_info_context->bmi[yoffset + 0].mv.as_mv.col +
x->mode_info_context->bmi[yoffset + 1].mv.as_mv.col +
x->mode_info_context->bmi[yoffset + 4].mv.as_mv.col +
x->mode_info_context->bmi[yoffset + 5].mv.as_mv.col;
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
temp += 4 + ((temp >> (sizeof(temp) * CHAR_BIT - 1)) * 8);
x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
x->block[uoffset].bmi.mv.as_mv.col = (temp / 8) & x->fullpixel_mask;
if (x->mode_info_context->mbmi.need_to_clamp_mvs)
clamp_uvmv_to_umv_border(&x->block[uoffset].bmi.mv.as_mv, x);
if (x->mode_info_context->mbmi.need_to_clamp_mvs)
clamp_uvmv_to_umv_border(&x->block[uoffset].bmi.mv.as_mv, x);
x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
}
x->block[voffset].bmi.mv.as_int = x->block[uoffset].bmi.mv.as_int;
}
}
}
void vp8_build_inter_predictors_mb(MACROBLOCKD *xd)
{
if (xd->mode_info_context->mbmi.mode != SPLITMV)
{
vp8_build_inter16x16_predictors_mb(xd, xd->dst.y_buffer,
xd->dst.u_buffer, xd->dst.v_buffer,
xd->dst.y_stride, xd->dst.uv_stride);
}
else
{
build_4x4uvmvs(xd);
build_inter4x4_predictors_mb(xd);
}
void vp8_build_inter_predictors_mb(MACROBLOCKD *xd) {
if (xd->mode_info_context->mbmi.mode != SPLITMV) {
vp8_build_inter16x16_predictors_mb(xd, xd->dst.y_buffer, xd->dst.u_buffer,
xd->dst.v_buffer, xd->dst.y_stride,
xd->dst.uv_stride);
} else {
build_4x4uvmvs(xd);
build_inter4x4_predictors_mb(xd);
}
}

14
vp8/common/reconinter.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_RECONINTER_H_
#define VP8_COMMON_RECONINTER_H_
@ -17,21 +16,16 @@ extern "C" {
#endif
extern void vp8_build_inter_predictors_mb(MACROBLOCKD *x);
extern void vp8_build_inter16x16_predictors_mb(MACROBLOCKD *x,
unsigned char *dst_y,
unsigned char *dst_u,
unsigned char *dst_v,
int dst_ystride,
int dst_uvstride);
extern void vp8_build_inter16x16_predictors_mb(
MACROBLOCKD *x, unsigned char *dst_y, unsigned char *dst_u,
unsigned char *dst_v, int dst_ystride, int dst_uvstride);
extern void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
unsigned char *dst_y,
int dst_ystride);
extern void vp8_build_inter_predictors_b(BLOCKD *d, int pitch,
unsigned char *base_pre,
int pre_stride,
vp8_subpix_fn_t sppf);
int pre_stride, vp8_subpix_fn_t sppf);
extern void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x);
extern void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x);

129
vp8/common/reconintra.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "./vpx_config.h"
#include "./vpx_dsp_rtcd.h"
#include "./vp8_rtcd.h"
@ -19,9 +18,9 @@
#include "vp8/common/reconintra4x4.h"
enum {
SIZE_16,
SIZE_8,
NUM_SIZES,
SIZE_16,
SIZE_8,
NUM_SIZES,
};
typedef void (*intra_pred_fn)(uint8_t *dst, ptrdiff_t stride,
@ -30,88 +29,68 @@ typedef void (*intra_pred_fn)(uint8_t *dst, ptrdiff_t stride,
static intra_pred_fn pred[4][NUM_SIZES];
static intra_pred_fn dc_pred[2][2][NUM_SIZES];
static void vp8_init_intra_predictors_internal(void)
{
#define INIT_SIZE(sz) \
pred[V_PRED][SIZE_##sz] = vpx_v_predictor_##sz##x##sz; \
pred[H_PRED][SIZE_##sz] = vpx_h_predictor_##sz##x##sz; \
pred[TM_PRED][SIZE_##sz] = vpx_tm_predictor_##sz##x##sz; \
\
dc_pred[0][0][SIZE_##sz] = vpx_dc_128_predictor_##sz##x##sz; \
dc_pred[0][1][SIZE_##sz] = vpx_dc_top_predictor_##sz##x##sz; \
dc_pred[1][0][SIZE_##sz] = vpx_dc_left_predictor_##sz##x##sz; \
dc_pred[1][1][SIZE_##sz] = vpx_dc_predictor_##sz##x##sz
static void vp8_init_intra_predictors_internal(void) {
#define INIT_SIZE(sz) \
pred[V_PRED][SIZE_##sz] = vpx_v_predictor_##sz##x##sz; \
pred[H_PRED][SIZE_##sz] = vpx_h_predictor_##sz##x##sz; \
pred[TM_PRED][SIZE_##sz] = vpx_tm_predictor_##sz##x##sz; \
\
dc_pred[0][0][SIZE_##sz] = vpx_dc_128_predictor_##sz##x##sz; \
dc_pred[0][1][SIZE_##sz] = vpx_dc_top_predictor_##sz##x##sz; \
dc_pred[1][0][SIZE_##sz] = vpx_dc_left_predictor_##sz##x##sz; \
dc_pred[1][1][SIZE_##sz] = vpx_dc_predictor_##sz##x##sz
INIT_SIZE(16);
INIT_SIZE(8);
vp8_init_intra4x4_predictors_internal();
INIT_SIZE(16);
INIT_SIZE(8);
vp8_init_intra4x4_predictors_internal();
}
void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x,
unsigned char * yabove_row,
unsigned char * yleft,
int left_stride,
unsigned char * ypred_ptr,
int y_stride)
{
MB_PREDICTION_MODE mode = x->mode_info_context->mbmi.mode;
DECLARE_ALIGNED(16, uint8_t, yleft_col[16]);
int i;
intra_pred_fn fn;
void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x, unsigned char *yabove_row,
unsigned char *yleft, int left_stride,
unsigned char *ypred_ptr, int y_stride) {
MB_PREDICTION_MODE mode = x->mode_info_context->mbmi.mode;
DECLARE_ALIGNED(16, uint8_t, yleft_col[16]);
int i;
intra_pred_fn fn;
for (i = 0; i < 16; i++)
{
yleft_col[i] = yleft[i* left_stride];
}
for (i = 0; i < 16; i++) {
yleft_col[i] = yleft[i * left_stride];
}
if (mode == DC_PRED)
{
fn = dc_pred[x->left_available][x->up_available][SIZE_16];
}
else
{
fn = pred[mode][SIZE_16];
}
if (mode == DC_PRED) {
fn = dc_pred[x->left_available][x->up_available][SIZE_16];
} else {
fn = pred[mode][SIZE_16];
}
fn(ypred_ptr, y_stride, yabove_row, yleft_col);
fn(ypred_ptr, y_stride, yabove_row, yleft_col);
}
void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x,
unsigned char * uabove_row,
unsigned char * vabove_row,
unsigned char * uleft,
unsigned char * vleft,
int left_stride,
unsigned char * upred_ptr,
unsigned char * vpred_ptr,
int pred_stride)
{
MB_PREDICTION_MODE uvmode = x->mode_info_context->mbmi.uv_mode;
unsigned char uleft_col[8];
unsigned char vleft_col[8];
int i;
intra_pred_fn fn;
void vp8_build_intra_predictors_mbuv_s(
MACROBLOCKD *x, unsigned char *uabove_row, unsigned char *vabove_row,
unsigned char *uleft, unsigned char *vleft, int left_stride,
unsigned char *upred_ptr, unsigned char *vpred_ptr, int pred_stride) {
MB_PREDICTION_MODE uvmode = x->mode_info_context->mbmi.uv_mode;
unsigned char uleft_col[8];
unsigned char vleft_col[8];
int i;
intra_pred_fn fn;
for (i = 0; i < 8; i++)
{
uleft_col[i] = uleft[i * left_stride];
vleft_col[i] = vleft[i * left_stride];
}
for (i = 0; i < 8; i++) {
uleft_col[i] = uleft[i * left_stride];
vleft_col[i] = vleft[i * left_stride];
}
if (uvmode == DC_PRED)
{
fn = dc_pred[x->left_available][x->up_available][SIZE_8];
}
else
{
fn = pred[uvmode][SIZE_8];
}
if (uvmode == DC_PRED) {
fn = dc_pred[x->left_available][x->up_available][SIZE_8];
} else {
fn = pred[uvmode][SIZE_8];
}
fn(upred_ptr, pred_stride, uabove_row, uleft_col);
fn(vpred_ptr, pred_stride, vabove_row, vleft_col);
fn(upred_ptr, pred_stride, uabove_row, uleft_col);
fn(vpred_ptr, pred_stride, vabove_row, vleft_col);
}
void vp8_init_intra_predictors(void)
{
once(vp8_init_intra_predictors_internal);
void vp8_init_intra_predictors(void) {
once(vp8_init_intra_predictors_internal);
}

23
vp8/common/reconintra.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_RECONINTRA_H_
#define VP8_COMMON_RECONINTRA_H_
@ -18,22 +17,14 @@
extern "C" {
#endif
void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x,
unsigned char *yabove_row,
unsigned char *yleft,
int left_stride,
unsigned char *ypred_ptr,
int y_stride);
void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x, unsigned char *yabove_row,
unsigned char *yleft, int left_stride,
unsigned char *ypred_ptr, int y_stride);
void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x,
unsigned char * uabove_row,
unsigned char * vabove_row,
unsigned char * uleft,
unsigned char * vleft,
int left_stride,
unsigned char * upred_ptr,
unsigned char * vpred_ptr,
int pred_stride);
void vp8_build_intra_predictors_mbuv_s(
MACROBLOCKD *x, unsigned char *uabove_row, unsigned char *vabove_row,
unsigned char *uleft, unsigned char *vleft, int left_stride,
unsigned char *upred_ptr, unsigned char *vpred_ptr, int pred_stride);
void vp8_init_intra_predictors(void);

49
vp8/common/reconintra4x4.c
View File

@ -21,35 +21,32 @@ typedef void (*intra_pred_fn)(uint8_t *dst, ptrdiff_t stride,
static intra_pred_fn pred[10];
void vp8_init_intra4x4_predictors_internal(void)
{
pred[B_DC_PRED] = vpx_dc_predictor_4x4;
pred[B_TM_PRED] = vpx_tm_predictor_4x4;
pred[B_VE_PRED] = vpx_ve_predictor_4x4;
pred[B_HE_PRED] = vpx_he_predictor_4x4;
pred[B_LD_PRED] = vpx_d45e_predictor_4x4;
pred[B_RD_PRED] = vpx_d135_predictor_4x4;
pred[B_VR_PRED] = vpx_d117_predictor_4x4;
pred[B_VL_PRED] = vpx_d63f_predictor_4x4;
pred[B_HD_PRED] = vpx_d153_predictor_4x4;
pred[B_HU_PRED] = vpx_d207_predictor_4x4;
void vp8_init_intra4x4_predictors_internal(void) {
pred[B_DC_PRED] = vpx_dc_predictor_4x4;
pred[B_TM_PRED] = vpx_tm_predictor_4x4;
pred[B_VE_PRED] = vpx_ve_predictor_4x4;
pred[B_HE_PRED] = vpx_he_predictor_4x4;
pred[B_LD_PRED] = vpx_d45e_predictor_4x4;
pred[B_RD_PRED] = vpx_d135_predictor_4x4;
pred[B_VR_PRED] = vpx_d117_predictor_4x4;
pred[B_VL_PRED] = vpx_d63f_predictor_4x4;
pred[B_HD_PRED] = vpx_d153_predictor_4x4;
pred[B_HU_PRED] = vpx_d207_predictor_4x4;
}
void vp8_intra4x4_predict(unsigned char *above,
unsigned char *yleft, int left_stride,
B_PREDICTION_MODE b_mode,
void vp8_intra4x4_predict(unsigned char *above, unsigned char *yleft,
int left_stride, B_PREDICTION_MODE b_mode,
unsigned char *dst, int dst_stride,
unsigned char top_left)
{
unsigned char Left[4];
unsigned char Aboveb[12], *Above = Aboveb + 4;
unsigned char top_left) {
unsigned char Left[4];
unsigned char Aboveb[12], *Above = Aboveb + 4;
Left[0] = yleft[0];
Left[1] = yleft[left_stride];
Left[2] = yleft[2 * left_stride];
Left[3] = yleft[3 * left_stride];
memcpy(Above, above, 8);
Above[-1] = top_left;
Left[0] = yleft[0];
Left[1] = yleft[left_stride];
Left[2] = yleft[2 * left_stride];
Left[3] = yleft[3 * left_stride];
memcpy(Above, above, 8);
Above[-1] = top_left;
pred[b_mode](dst, dst_stride, Above, Left);
pred[b_mode](dst, dst_stride, Above, Left);
}

27
vp8/common/reconintra4x4.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_RECONINTRA4X4_H_
#define VP8_COMMON_RECONINTRA4X4_H_
#include "vp8/common/blockd.h"
@ -18,24 +17,22 @@ extern "C" {
#endif
static INLINE void intra_prediction_down_copy(MACROBLOCKD *xd,
unsigned char *above_right_src)
{
int dst_stride = xd->dst.y_stride;
unsigned char *above_right_dst = xd->dst.y_buffer - dst_stride + 16;
unsigned char *above_right_src) {
int dst_stride = xd->dst.y_stride;
unsigned char *above_right_dst = xd->dst.y_buffer - dst_stride + 16;
unsigned int *src_ptr = (unsigned int *)above_right_src;
unsigned int *dst_ptr0 = (unsigned int *)(above_right_dst + 4 * dst_stride);
unsigned int *dst_ptr1 = (unsigned int *)(above_right_dst + 8 * dst_stride);
unsigned int *dst_ptr2 = (unsigned int *)(above_right_dst + 12 * dst_stride);
unsigned int *src_ptr = (unsigned int *)above_right_src;
unsigned int *dst_ptr0 = (unsigned int *)(above_right_dst + 4 * dst_stride);
unsigned int *dst_ptr1 = (unsigned int *)(above_right_dst + 8 * dst_stride);
unsigned int *dst_ptr2 = (unsigned int *)(above_right_dst + 12 * dst_stride);
*dst_ptr0 = *src_ptr;
*dst_ptr1 = *src_ptr;
*dst_ptr2 = *src_ptr;
*dst_ptr0 = *src_ptr;
*dst_ptr1 = *src_ptr;
*dst_ptr2 = *src_ptr;
}
void vp8_intra4x4_predict(unsigned char *Above,
unsigned char *yleft, int left_stride,
B_PREDICTION_MODE b_mode,
void vp8_intra4x4_predict(unsigned char *Above, unsigned char *yleft,
int left_stride, B_PREDICTION_MODE b_mode,
unsigned char *dst, int dst_stride,
unsigned char top_left);

6
vp8/common/rtcd.c
View File

@ -12,8 +12,4 @@
#include "./vp8_rtcd.h"
#include "vpx_ports/vpx_once.h"
void vp8_rtcd()
{
once(setup_rtcd_internal);
}
void vp8_rtcd() { once(setup_rtcd_internal); }

36
vp8/common/setupintrarecon.c
View File

@ -8,32 +8,28 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "setupintrarecon.h"
#include "vpx_mem/vpx_mem.h"
void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf)
{
int i;
void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf) {
int i;
/* set up frame new frame for intra coded blocks */
memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
for (i = 0; i < ybf->y_height; i++)
ybf->y_buffer[ybf->y_stride *i - 1] = (unsigned char) 129;
/* set up frame new frame for intra coded blocks */
memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
for (i = 0; i < ybf->y_height; i++)
ybf->y_buffer[ybf->y_stride * i - 1] = (unsigned char)129;
memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
for (i = 0; i < ybf->uv_height; i++)
ybf->u_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
for (i = 0; i < ybf->uv_height; i++)
ybf->v_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129;
memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
for (i = 0; i < ybf->uv_height; i++)
ybf->u_buffer[ybf->uv_stride * i - 1] = (unsigned char)129;
memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
for (i = 0; i < ybf->uv_height; i++)
ybf->v_buffer[ybf->uv_stride * i - 1] = (unsigned char)129;
}
void vp8_setup_intra_recon_top_line(YV12_BUFFER_CONFIG *ybf)
{
memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
void vp8_setup_intra_recon_top_line(YV12_BUFFER_CONFIG *ybf) {
memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5);
memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5);
}

17
vp8/common/setupintrarecon.h
View File

@ -22,20 +22,15 @@ extern void vp8_setup_intra_recon_top_line(YV12_BUFFER_CONFIG *ybf);
static INLINE void setup_intra_recon_left(unsigned char *y_buffer,
unsigned char *u_buffer,
unsigned char *v_buffer,
int y_stride,
int uv_stride)
{
int i;
unsigned char *v_buffer, int y_stride,
int uv_stride) {
int i;
for (i = 0; i < 16; i++)
y_buffer[y_stride *i] = (unsigned char) 129;
for (i = 0; i < 16; i++) y_buffer[y_stride * i] = (unsigned char)129;
for (i = 0; i < 8; i++)
u_buffer[uv_stride *i] = (unsigned char) 129;
for (i = 0; i < 8; i++) u_buffer[uv_stride * i] = (unsigned char)129;
for (i = 0; i < 8; i++)
v_buffer[uv_stride *i] = (unsigned char) 129;
for (i = 0; i < 8; i++) v_buffer[uv_stride * i] = (unsigned char)129;
}
#ifdef __cplusplus

32
vp8/common/swapyv12buffer.c
View File

@ -8,27 +8,25 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "swapyv12buffer.h"
void vp8_swap_yv12_buffer(YV12_BUFFER_CONFIG *new_frame, YV12_BUFFER_CONFIG *last_frame)
{
unsigned char *temp;
void vp8_swap_yv12_buffer(YV12_BUFFER_CONFIG *new_frame,
YV12_BUFFER_CONFIG *last_frame) {
unsigned char *temp;
temp = last_frame->buffer_alloc;
last_frame->buffer_alloc = new_frame->buffer_alloc;
new_frame->buffer_alloc = temp;
temp = last_frame->buffer_alloc;
last_frame->buffer_alloc = new_frame->buffer_alloc;
new_frame->buffer_alloc = temp;
temp = last_frame->y_buffer;
last_frame->y_buffer = new_frame->y_buffer;
new_frame->y_buffer = temp;
temp = last_frame->y_buffer;
last_frame->y_buffer = new_frame->y_buffer;
new_frame->y_buffer = temp;
temp = last_frame->u_buffer;
last_frame->u_buffer = new_frame->u_buffer;
new_frame->u_buffer = temp;
temp = last_frame->v_buffer;
last_frame->v_buffer = new_frame->v_buffer;
new_frame->v_buffer = temp;
temp = last_frame->u_buffer;
last_frame->u_buffer = new_frame->u_buffer;
new_frame->u_buffer = temp;
temp = last_frame->v_buffer;
last_frame->v_buffer = new_frame->v_buffer;
new_frame->v_buffer = temp;
}

4
vp8/common/swapyv12buffer.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_SWAPYV12BUFFER_H_
#define VP8_COMMON_SWAPYV12BUFFER_H_
@ -18,7 +17,8 @@
extern "C" {
#endif
void vp8_swap_yv12_buffer(YV12_BUFFER_CONFIG *new_frame, YV12_BUFFER_CONFIG *last_frame);
void vp8_swap_yv12_buffer(YV12_BUFFER_CONFIG *new_frame,
YV12_BUFFER_CONFIG *last_frame);
#ifdef __cplusplus
} // extern "C"

194
vp8/common/textblit.c
View File

@ -10,121 +10,109 @@
#include <stdlib.h>
void vp8_blit_text(const char *msg, unsigned char *address, const int pitch) {
int letter_bitmap;
unsigned char *output_pos = address;
int colpos;
const int font[] = {
0x0, 0x5C00, 0x8020, 0xAFABEA, 0xD7EC0, 0x1111111, 0x1855740,
0x18000, 0x45C0, 0x74400, 0x51140, 0x23880, 0xC4000, 0x21080,
0x80000, 0x111110, 0xE9D72E, 0x87E40, 0x12AD732, 0xAAD62A, 0x4F94C4,
0x4D6B7, 0x456AA, 0x3E8423, 0xAAD6AA, 0xAAD6A2, 0x2800, 0x2A00,
0x8A880, 0x52940, 0x22A20, 0x15422, 0x6AD62E, 0x1E4A53E, 0xAAD6BF,
0x8C62E, 0xE8C63F, 0x118D6BF, 0x1094BF, 0xCAC62E, 0x1F2109F, 0x118FE31,
0xF8C628, 0x8A89F, 0x108421F, 0x1F1105F, 0x1F4105F, 0xE8C62E, 0x2294BF,
0x164C62E, 0x12694BF, 0x8AD6A2, 0x10FC21, 0x1F8421F, 0x744107, 0xF8220F,
0x1151151, 0x117041, 0x119D731, 0x47E0, 0x1041041, 0xFC400, 0x10440,
0x1084210, 0x820
};
colpos = 0;
void vp8_blit_text(const char *msg, unsigned char *address, const int pitch)
{
int letter_bitmap;
unsigned char *output_pos = address;
int colpos;
const int font[] =
{
0x0, 0x5C00, 0x8020, 0xAFABEA, 0xD7EC0, 0x1111111, 0x1855740, 0x18000,
0x45C0, 0x74400, 0x51140, 0x23880, 0xC4000, 0x21080, 0x80000, 0x111110,
0xE9D72E, 0x87E40, 0x12AD732, 0xAAD62A, 0x4F94C4, 0x4D6B7, 0x456AA,
0x3E8423, 0xAAD6AA, 0xAAD6A2, 0x2800, 0x2A00, 0x8A880, 0x52940, 0x22A20,
0x15422, 0x6AD62E, 0x1E4A53E, 0xAAD6BF, 0x8C62E, 0xE8C63F, 0x118D6BF,
0x1094BF, 0xCAC62E, 0x1F2109F, 0x118FE31, 0xF8C628, 0x8A89F, 0x108421F,
0x1F1105F, 0x1F4105F, 0xE8C62E, 0x2294BF, 0x164C62E, 0x12694BF, 0x8AD6A2,
0x10FC21, 0x1F8421F, 0x744107, 0xF8220F, 0x1151151, 0x117041, 0x119D731,
0x47E0, 0x1041041, 0xFC400, 0x10440, 0x1084210, 0x820
};
colpos = 0;
while (msg[colpos] != 0) {
char letter = msg[colpos];
int fontcol, fontrow;
while (msg[colpos] != 0)
{
char letter = msg[colpos];
int fontcol, fontrow;
if (letter <= 'Z' && letter >= ' ')
letter_bitmap = font[letter - ' '];
else if (letter <= 'z' && letter >= 'a')
letter_bitmap = font[letter - 'a' + 'A' - ' '];
else
letter_bitmap = font[0];
if (letter <= 'Z' && letter >= ' ')
letter_bitmap = font[letter-' '];
else if (letter <= 'z' && letter >= 'a')
letter_bitmap = font[letter-'a'+'A' - ' '];
else
letter_bitmap = font[0];
for (fontcol = 6; fontcol >= 0; fontcol--)
for (fontrow = 0; fontrow < 5; fontrow++)
output_pos[fontrow * pitch + fontcol] =
((letter_bitmap >> (fontcol * 5)) & (1 << fontrow) ? 255 : 0);
for (fontcol = 6; fontcol >= 0 ; fontcol--)
for (fontrow = 0; fontrow < 5; fontrow++)
output_pos[fontrow *pitch + fontcol] =
((letter_bitmap >> (fontcol * 5)) & (1 << fontrow) ? 255 : 0);
output_pos += 7;
colpos++;
}
output_pos += 7;
colpos++;
}
}
static void plot (const int x, const int y, unsigned char *image, const int pitch)
{
image [x+y*pitch] ^= 255;
static void plot(const int x, const int y, unsigned char *image,
const int pitch) {
image[x + y * pitch] ^= 255;
}
/* Bresenham line algorithm */
void vp8_blit_line(int x0, int x1, int y0, int y1, unsigned char *image, const int pitch)
{
int steep = abs(y1 - y0) > abs(x1 - x0);
int deltax, deltay;
int error, ystep, y, x;
void vp8_blit_line(int x0, int x1, int y0, int y1, unsigned char *image,
const int pitch) {
int steep = abs(y1 - y0) > abs(x1 - x0);
int deltax, deltay;
int error, ystep, y, x;
if (steep)
{
int t;
t = x0;
x0 = y0;
y0 = t;
if (steep) {
int t;
t = x0;
x0 = y0;
y0 = t;
t = x1;
x1 = y1;
y1 = t;
t = x1;
x1 = y1;
y1 = t;
}
if (x0 > x1) {
int t;
t = x0;
x0 = x1;
x1 = t;
t = y0;
y0 = y1;
y1 = t;
}
deltax = x1 - x0;
deltay = abs(y1 - y0);
error = deltax / 2;
y = y0;
if (y0 < y1)
ystep = 1;
else
ystep = -1;
if (steep) {
for (x = x0; x <= x1; x++) {
plot(y, x, image, pitch);
error = error - deltay;
if (error < 0) {
y = y + ystep;
error = error + deltax;
}
}
} else {
for (x = x0; x <= x1; x++) {
plot(x, y, image, pitch);
if (x0 > x1)
{
int t;
t = x0;
x0 = x1;
x1 = t;
t = y0;
y0 = y1;
y1 = t;
}
deltax = x1 - x0;
deltay = abs(y1 - y0);
error = deltax / 2;
y = y0;
if (y0 < y1)
ystep = 1;
else
ystep = -1;
if (steep)
{
for (x = x0; x <= x1; x++)
{
plot(y,x, image, pitch);
error = error - deltay;
if (error < 0)
{
y = y + ystep;
error = error + deltax;
}
}
}
else
{
for (x = x0; x <= x1; x++)
{
plot(x,y, image, pitch);
error = error - deltay;
if (error < 0)
{
y = y + ystep;
error = error + deltax;
}
}
error = error - deltay;
if (error < 0) {
y = y + ystep;
error = error + deltax;
}
}
}
}

166
vp8/common/threading.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_THREADING_H_
#define VP8_COMMON_THREADING_H_
@ -30,10 +29,12 @@ extern "C" {
#define THREAD_SPECIFIC_INDEX DWORD
#define pthread_t HANDLE
#define pthread_attr_t DWORD
#define pthread_detach(thread) if(thread!=NULL)CloseHandle(thread)
#define pthread_detach(thread) \
if (thread != NULL) CloseHandle(thread)
#define thread_sleep(nms) Sleep(nms)
#define pthread_cancel(thread) terminate_thread(thread,0)
#define ts_key_create(ts_key, destructor) {ts_key = TlsAlloc();};
#define pthread_cancel(thread) terminate_thread(thread, 0)
#define ts_key_create(ts_key, destructor) \
{ ts_key = TlsAlloc(); };
#define pthread_getspecific(ts_key) TlsGetValue(ts_key)
#define pthread_setspecific(ts_key, value) TlsSetValue(ts_key, (void *)value)
#define pthread_self() GetCurrentThreadId()
@ -53,9 +54,9 @@ extern "C" {
#define thread_sleep(nms) DosSleep(nms)
#define pthread_cancel(thread) DosKillThread(thread)
#define ts_key_create(ts_key, destructor) \
DosAllocThreadLocalMemory(1, &(ts_key));
DosAllocThreadLocalMemory(1, &(ts_key));
#define pthread_getspecific(ts_key) ((void *)(*(ts_key)))
#define pthread_setspecific(ts_key, value) (*(ts_key)=(ULONG)(value))
#define pthread_setspecific(ts_key, value) (*(ts_key) = (ULONG)(value))
#define pthread_self() _gettid()
#else
#ifdef __APPLE__
@ -75,85 +76,82 @@ extern "C" {
#define THREAD_FUNCTION void *
#define THREAD_FUNCTION_RETURN void *
#define THREAD_SPECIFIC_INDEX pthread_key_t
#define ts_key_create(ts_key, destructor) pthread_key_create (&(ts_key), destructor);
#define ts_key_create(ts_key, destructor) \
pthread_key_create(&(ts_key), destructor);
#endif
/* Synchronization macros: Win32 and Pthreads */
#if defined(_WIN32) && !HAVE_PTHREAD_H
#define sem_t HANDLE
#define pause(voidpara) __asm PAUSE
#define sem_init(sem, sem_attr1, sem_init_value) (int)((*sem = CreateSemaphore(NULL,0,32768,NULL))==NULL)
#define sem_wait(sem) (int)(WAIT_OBJECT_0 != WaitForSingleObject(*sem,INFINITE))
#define sem_post(sem) ReleaseSemaphore(*sem,1,NULL)
#define sem_destroy(sem) if(*sem)((int)(CloseHandle(*sem))==TRUE)
#define sem_init(sem, sem_attr1, sem_init_value) \
(int)((*sem = CreateSemaphore(NULL, 0, 32768, NULL)) == NULL)
#define sem_wait(sem) \
(int)(WAIT_OBJECT_0 != WaitForSingleObject(*sem, INFINITE))
#define sem_post(sem) ReleaseSemaphore(*sem, 1, NULL)
#define sem_destroy(sem) \
if (*sem) ((int)(CloseHandle(*sem)) == TRUE)
#define thread_sleep(nms) Sleep(nms)
#elif defined(__OS2__)
typedef struct
{
HEV event;
HMTX wait_mutex;
HMTX count_mutex;
int count;
typedef struct {
HEV event;
HMTX wait_mutex;
HMTX count_mutex;
int count;
} sem_t;
static inline int sem_init(sem_t *sem, int pshared, unsigned int value)
{
DosCreateEventSem(NULL, &sem->event, pshared ? DC_SEM_SHARED : 0,
value > 0 ? TRUE : FALSE);
DosCreateMutexSem(NULL, &sem->wait_mutex, 0, FALSE);
DosCreateMutexSem(NULL, &sem->count_mutex, 0, FALSE);
static inline int sem_init(sem_t *sem, int pshared, unsigned int value) {
DosCreateEventSem(NULL, &sem->event, pshared ? DC_SEM_SHARED : 0,
value > 0 ? TRUE : FALSE);
DosCreateMutexSem(NULL, &sem->wait_mutex, 0, FALSE);
DosCreateMutexSem(NULL, &sem->count_mutex, 0, FALSE);
sem->count = value;
sem->count = value;
return 0;
return 0;
}
static inline int sem_wait(sem_t * sem)
{
DosRequestMutexSem(sem->wait_mutex, -1);
static inline int sem_wait(sem_t *sem) {
DosRequestMutexSem(sem->wait_mutex, -1);
DosWaitEventSem(sem->event, -1);
DosWaitEventSem(sem->event, -1);
DosRequestMutexSem(sem->count_mutex, -1);
DosRequestMutexSem(sem->count_mutex, -1);
sem->count--;
if (sem->count == 0)
{
ULONG post_count;
sem->count--;
if (sem->count == 0) {
ULONG post_count;
DosResetEventSem(sem->event, &post_count);
}
DosResetEventSem(sem->event, &post_count);
}
DosReleaseMutexSem(sem->count_mutex);
DosReleaseMutexSem(sem->count_mutex);
DosReleaseMutexSem(sem->wait_mutex);
DosReleaseMutexSem(sem->wait_mutex);
return 0;
return 0;
}
static inline int sem_post(sem_t * sem)
{
DosRequestMutexSem(sem->count_mutex, -1);
static inline int sem_post(sem_t *sem) {
DosRequestMutexSem(sem->count_mutex, -1);
if (sem->count < 32768)
{
sem->count++;
DosPostEventSem(sem->event);
}
if (sem->count < 32768) {
sem->count++;
DosPostEventSem(sem->event);
}
DosReleaseMutexSem(sem->count_mutex);
DosReleaseMutexSem(sem->count_mutex);
return 0;
return 0;
}
static inline int sem_destroy(sem_t * sem)
{
DosCloseEventSem(sem->event);
DosCloseMutexSem(sem->wait_mutex);
DosCloseMutexSem(sem->count_mutex);
static inline int sem_destroy(sem_t *sem) {
DosCloseEventSem(sem->event);
DosCloseMutexSem(sem->wait_mutex);
DosCloseMutexSem(sem->count_mutex);
return 0;
return 0;
}
#define thread_sleep(nms) DosSleep(nms)
@ -162,15 +160,20 @@ static inline int sem_destroy(sem_t * sem)
#ifdef __APPLE__
#define sem_t semaphore_t
#define sem_init(X,Y,Z) semaphore_create(mach_task_self(), X, SYNC_POLICY_FIFO, Z)
#define sem_wait(sem) (semaphore_wait(*sem) )
#define sem_init(X, Y, Z) \
semaphore_create(mach_task_self(), X, SYNC_POLICY_FIFO, Z)
#define sem_wait(sem) (semaphore_wait(*sem))
#define sem_post(sem) semaphore_signal(*sem)
#define sem_destroy(sem) semaphore_destroy(mach_task_self(),*sem)
#define thread_sleep(nms) /* { struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
#define sem_destroy(sem) semaphore_destroy(mach_task_self(), *sem)
#define thread_sleep(nms)
/* { struct timespec ts;ts.tv_sec=0; ts.tv_nsec =
1000*nms;nanosleep(&ts, NULL);} */
#else
#include <unistd.h>
#include <sched.h>
#define thread_sleep(nms) sched_yield();/* {struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
#define thread_sleep(nms) sched_yield();
/* {struct timespec ts;ts.tv_sec=0;
ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} */
#endif
/* Not Windows. Assume pthreads */
@ -185,42 +188,41 @@ static inline int sem_destroy(sem_t * sem)
#include "vpx_util/vpx_thread.h"
static INLINE void mutex_lock(pthread_mutex_t *const mutex) {
const int kMaxTryLocks = 4000;
int locked = 0;
int i;
const int kMaxTryLocks = 4000;
int locked = 0;
int i;
for (i = 0; i < kMaxTryLocks; ++i) {
if (!pthread_mutex_trylock(mutex)) {
locked = 1;
break;
}
for (i = 0; i < kMaxTryLocks; ++i) {
if (!pthread_mutex_trylock(mutex)) {
locked = 1;
break;
}
}
if (!locked)
pthread_mutex_lock(mutex);
if (!locked) pthread_mutex_lock(mutex);
}
static INLINE int protected_read(pthread_mutex_t *const mutex, const int *p) {
int ret;
mutex_lock(mutex);
ret = *p;
pthread_mutex_unlock(mutex);
return ret;
int ret;
mutex_lock(mutex);
ret = *p;
pthread_mutex_unlock(mutex);
return ret;
}
static INLINE void sync_read(pthread_mutex_t *const mutex, int mb_col,
const int *last_row_current_mb_col,
const int nsync) {
while (mb_col > (protected_read(mutex, last_row_current_mb_col) - nsync)) {
x86_pause_hint();
thread_sleep(0);
}
while (mb_col > (protected_read(mutex, last_row_current_mb_col) - nsync)) {
x86_pause_hint();
thread_sleep(0);
}
}
static INLINE void protected_write(pthread_mutex_t *mutex, int *p, int v) {
mutex_lock(mutex);
*p = v;
pthread_mutex_unlock(mutex);
mutex_lock(mutex);
*p = v;
pthread_mutex_unlock(mutex);
}
#endif /* CONFIG_OS_SUPPORT && CONFIG_MULTITHREAD */

154
vp8/common/treecoder.c
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#if CONFIG_DEBUG
#include <assert.h>
#endif
@ -16,128 +15,95 @@
#include "treecoder.h"
static void tree2tok(
struct vp8_token_struct *const p,
vp8_tree t,
int i,
int v,
int L
)
{
v += v;
++L;
static void tree2tok(struct vp8_token_struct *const p, vp8_tree t, int i, int v,
int L) {
v += v;
++L;
do
{
const vp8_tree_index j = t[i++];
do {
const vp8_tree_index j = t[i++];
if (j <= 0)
{
p[-j].value = v;
p[-j].Len = L;
}
else
tree2tok(p, t, j, v, L);
}
while (++v & 1);
if (j <= 0) {
p[-j].value = v;
p[-j].Len = L;
} else
tree2tok(p, t, j, v, L);
} while (++v & 1);
}
void vp8_tokens_from_tree(struct vp8_token_struct *p, vp8_tree t)
{
tree2tok(p, t, 0, 0, 0);
void vp8_tokens_from_tree(struct vp8_token_struct *p, vp8_tree t) {
tree2tok(p, t, 0, 0, 0);
}
void vp8_tokens_from_tree_offset(struct vp8_token_struct *p, vp8_tree t,
int offset)
{
tree2tok(p - offset, t, 0, 0, 0);
int offset) {
tree2tok(p - offset, t, 0, 0, 0);
}
static void branch_counts(
int n, /* n = size of alphabet */
vp8_token tok [ /* n */ ],
vp8_tree tree,
unsigned int branch_ct [ /* n-1 */ ] [2],
const unsigned int num_events[ /* n */ ]
)
{
const int tree_len = n - 1;
int t = 0;
static void branch_counts(int n, /* n = size of alphabet */
vp8_token tok[/* n */], vp8_tree tree,
unsigned int branch_ct[/* n-1 */][2],
const unsigned int num_events[/* n */]) {
const int tree_len = n - 1;
int t = 0;
#if CONFIG_DEBUG
assert(tree_len);
assert(tree_len);
#endif
do
{
branch_ct[t][0] = branch_ct[t][1] = 0;
}
while (++t < tree_len);
do {
branch_ct[t][0] = branch_ct[t][1] = 0;
} while (++t < tree_len);
t = 0;
t = 0;
do
{
int L = tok[t].Len;
const int enc = tok[t].value;
const unsigned int ct = num_events[t];
do {
int L = tok[t].Len;
const int enc = tok[t].value;
const unsigned int ct = num_events[t];
vp8_tree_index i = 0;
vp8_tree_index i = 0;
do
{
const int b = (enc >> --L) & 1;
const int j = i >> 1;
do {
const int b = (enc >> --L) & 1;
const int j = i >> 1;
#if CONFIG_DEBUG
assert(j < tree_len && 0 <= L);
assert(j < tree_len && 0 <= L);
#endif
branch_ct [j] [b] += ct;
i = tree[ i + b];
}
while (i > 0);
branch_ct[j][b] += ct;
i = tree[i + b];
} while (i > 0);
#if CONFIG_DEBUG
assert(!L);
assert(!L);
#endif
}
while (++t < n);
} while (++t < n);
}
void vp8_tree_probs_from_distribution(int n, /* n = size of alphabet */
vp8_token tok[/* n */], vp8_tree tree,
vp8_prob probs[/* n-1 */],
unsigned int branch_ct[/* n-1 */][2],
const unsigned int num_events[/* n */],
unsigned int Pfac, int rd) {
const int tree_len = n - 1;
int t = 0;
void vp8_tree_probs_from_distribution(
int n, /* n = size of alphabet */
vp8_token tok [ /* n */ ],
vp8_tree tree,
vp8_prob probs [ /* n-1 */ ],
unsigned int branch_ct [ /* n-1 */ ] [2],
const unsigned int num_events[ /* n */ ],
unsigned int Pfac,
int rd
)
{
const int tree_len = n - 1;
int t = 0;
branch_counts(n, tok, tree, branch_ct, num_events);
branch_counts(n, tok, tree, branch_ct, num_events);
do
{
const unsigned int *const c = branch_ct[t];
const unsigned int tot = c[0] + c[1];
do {
const unsigned int *const c = branch_ct[t];
const unsigned int tot = c[0] + c[1];
#if CONFIG_DEBUG
assert(tot < (1 << 24)); /* no overflow below */
assert(tot < (1 << 24)); /* no overflow below */
#endif
if (tot)
{
const unsigned int p = ((c[0] * Pfac) + (rd ? tot >> 1 : 0)) / tot;
probs[t] = p < 256 ? (p ? p : 1) : 255; /* agree w/old version for now */
}
else
probs[t] = vp8_prob_half;
}
while (++t < tree_len);
if (tot) {
const unsigned int p = ((c[0] * Pfac) + (rd ? tot >> 1 : 0)) / tot;
probs[t] = p < 256 ? (p ? p : 1) : 255; /* agree w/old version for now */
} else
probs[t] = vp8_prob_half;
} while (++t < tree_len);
}

50
vp8/common/treecoder.h
View File

@ -8,7 +8,6 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef VP8_COMMON_TREECODER_H_
#define VP8_COMMON_TREECODER_H_
@ -18,10 +17,9 @@ extern "C" {
typedef unsigned char vp8bc_index_t; /* probability index */
typedef unsigned char vp8_prob;
#define vp8_prob_half ( (vp8_prob) 128)
#define vp8_prob_half ((vp8_prob)128)
typedef signed char vp8_tree_index;
struct bool_coder_spec;
@ -34,10 +32,7 @@ typedef const bool_coder_spec c_bool_coder_spec;
typedef const bool_writer c_bool_writer;
typedef const bool_reader c_bool_reader;
# define vp8_complement( x) (255 - x)
#define vp8_complement(x) (255 - x)
/* We build coding trees compactly in arrays.
Each node of the tree is a pair of vp8_tree_indices.
@ -48,11 +43,9 @@ typedef const bool_reader c_bool_reader;
typedef const vp8_tree_index vp8_tree[], *vp8_tree_p;
typedef const struct vp8_token_struct
{
int value;
int Len;
typedef const struct vp8_token_struct {
int value;
int Len;
} vp8_token;
/* Construct encoding array from tree. */
@ -61,35 +54,26 @@ void vp8_tokens_from_tree(struct vp8_token_struct *, vp8_tree);
void vp8_tokens_from_tree_offset(struct vp8_token_struct *, vp8_tree,
int offset);
/* Convert array of token occurrence counts into a table of probabilities
for the associated binary encoding tree. Also writes count of branches
taken for each node on the tree; this facilitiates decisions as to
probability updates. */
void vp8_tree_probs_from_distribution(
int n, /* n = size of alphabet */
vp8_token tok [ /* n */ ],
vp8_tree tree,
vp8_prob probs [ /* n-1 */ ],
unsigned int branch_ct [ /* n-1 */ ] [2],
const unsigned int num_events[ /* n */ ],
unsigned int Pfactor,
int Round
);
void vp8_tree_probs_from_distribution(int n, /* n = size of alphabet */
vp8_token tok[/* n */], vp8_tree tree,
vp8_prob probs[/* n-1 */],
unsigned int branch_ct[/* n-1 */][2],
const unsigned int num_events[/* n */],
unsigned int Pfactor, int Round);
/* Variant of above using coder spec rather than hardwired 8-bit probs. */
void vp8bc_tree_probs_from_distribution(
int n, /* n = size of alphabet */
vp8_token tok [ /* n */ ],
vp8_tree tree,
vp8_prob probs [ /* n-1 */ ],
unsigned int branch_ct [ /* n-1 */ ] [2],
const unsigned int num_events[ /* n */ ],
c_bool_coder_spec *s
);
void vp8bc_tree_probs_from_distribution(int n, /* n = size of alphabet */
vp8_token tok[/* n */], vp8_tree tree,
vp8_prob probs[/* n-1 */],
unsigned int branch_ct[/* n-1 */][2],
const unsigned int num_events[/* n */],
c_bool_coder_spec *s);
#ifdef __cplusplus
} // extern "C"

334
vp8/common/vp8_entropymodedata.h
View File

@ -17,235 +17,153 @@ extern "C" {
/*Generated file, included by entropymode.c*/
const struct vp8_token_struct vp8_bmode_encodings[VP8_BINTRAMODES] =
{
{ 0, 1 },
{ 2, 2 },
{ 6, 3 },
{ 28, 5 },
{ 30, 5 },
{ 58, 6 },
{ 59, 6 },
{ 62, 6 },
{ 126, 7 },
{ 127, 7 }
const struct vp8_token_struct vp8_bmode_encodings[VP8_BINTRAMODES] = {
{ 0, 1 }, { 2, 2 }, { 6, 3 }, { 28, 5 }, { 30, 5 },
{ 58, 6 }, { 59, 6 }, { 62, 6 }, { 126, 7 }, { 127, 7 }
};
const struct vp8_token_struct vp8_ymode_encodings[VP8_YMODES] =
{
{ 0, 1 },
{ 4, 3 },
{ 5, 3 },
{ 6, 3 },
{ 7, 3 }
const struct vp8_token_struct vp8_ymode_encodings[VP8_YMODES] = {
{ 0, 1 }, { 4, 3 }, { 5, 3 }, { 6, 3 }, { 7, 3 }
};
const struct vp8_token_struct vp8_kf_ymode_encodings[VP8_YMODES] =
{
{ 4, 3 },
{ 5, 3 },
{ 6, 3 },
{ 7, 3 },
{ 0, 1 }
const struct vp8_token_struct vp8_kf_ymode_encodings[VP8_YMODES] = {
{ 4, 3 }, { 5, 3 }, { 6, 3 }, { 7, 3 }, { 0, 1 }
};
const struct vp8_token_struct vp8_uv_mode_encodings[VP8_UV_MODES] =
{
{ 0, 1 },
{ 2, 2 },
{ 6, 3 },
{ 7, 3 }
const struct vp8_token_struct vp8_uv_mode_encodings[VP8_UV_MODES] = {
{ 0, 1 }, { 2, 2 }, { 6, 3 }, { 7, 3 }
};
const struct vp8_token_struct vp8_mbsplit_encodings[VP8_NUMMBSPLITS] =
{
{ 6, 3 },
{ 7, 3 },
{ 2, 2 },
{ 0, 1 }
const struct vp8_token_struct vp8_mbsplit_encodings[VP8_NUMMBSPLITS] = {
{ 6, 3 }, { 7, 3 }, { 2, 2 }, { 0, 1 }
};
const struct vp8_token_struct vp8_mv_ref_encoding_array[VP8_MVREFS] =
{
{ 2, 2 },
{ 6, 3 },
{ 0, 1 },
{ 14, 4 },
{ 15, 4 }
const struct vp8_token_struct vp8_mv_ref_encoding_array[VP8_MVREFS] = {
{ 2, 2 }, { 6, 3 }, { 0, 1 }, { 14, 4 }, { 15, 4 }
};
const struct vp8_token_struct vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS] =
{
{ 0, 1 },
{ 2, 2 },
{ 6, 3 },
{ 7, 3 }
const struct vp8_token_struct vp8_sub_mv_ref_encoding_array[VP8_SUBMVREFS] = {
{ 0, 1 }, { 2, 2 }, { 6, 3 }, { 7, 3 }
};
const struct vp8_token_struct vp8_small_mvencodings[8] =
{
{ 0, 3 },
{ 1, 3 },
{ 2, 3 },
{ 3, 3 },
{ 4, 3 },
{ 5, 3 },
{ 6, 3 },
{ 7, 3 }
const struct vp8_token_struct vp8_small_mvencodings[8] = {
{ 0, 3 }, { 1, 3 }, { 2, 3 }, { 3, 3 }, { 4, 3 }, { 5, 3 }, { 6, 3 }, { 7, 3 }
};
const vp8_prob vp8_ymode_prob[VP8_YMODES-1] =
{
112, 86, 140, 37
};
const vp8_prob vp8_ymode_prob[VP8_YMODES - 1] = { 112, 86, 140, 37 };
const vp8_prob vp8_kf_ymode_prob[VP8_YMODES-1] =
{
145, 156, 163, 128
};
const vp8_prob vp8_kf_ymode_prob[VP8_YMODES - 1] = { 145, 156, 163, 128 };
const vp8_prob vp8_uv_mode_prob[VP8_UV_MODES-1] =
{
162, 101, 204
};
const vp8_prob vp8_uv_mode_prob[VP8_UV_MODES - 1] = { 162, 101, 204 };
const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES-1] =
{
142, 114, 183
};
const vp8_prob vp8_kf_uv_mode_prob[VP8_UV_MODES - 1] = { 142, 114, 183 };
const vp8_prob vp8_bmode_prob[VP8_BINTRAMODES-1] =
{
120, 90, 79, 133, 87, 85, 80, 111, 151
};
const vp8_prob vp8_bmode_prob[VP8_BINTRAMODES - 1] = { 120, 90, 79, 133, 87,
85, 80, 111, 151 };
const vp8_prob vp8_kf_bmode_prob
[VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1] =
{
{
{ 231, 120, 48, 89, 115, 113, 120, 152, 112 },
{ 152, 179, 64, 126, 170, 118, 46, 70, 95 },
{ 175, 69, 143, 80, 85, 82, 72, 155, 103 },
{ 56, 58, 10, 171, 218, 189, 17, 13, 152 },
{ 144, 71, 10, 38, 171, 213, 144, 34, 26 },
{ 114, 26, 17, 163, 44, 195, 21, 10, 173 },
{ 121, 24, 80, 195, 26, 62, 44, 64, 85 },
{ 170, 46, 55, 19, 136, 160, 33, 206, 71 },
{ 63, 20, 8, 114, 114, 208, 12, 9, 226 },
{ 81, 40, 11, 96, 182, 84, 29, 16, 36 }
},
{
{ 134, 183, 89, 137, 98, 101, 106, 165, 148 },
{ 72, 187, 100, 130, 157, 111, 32, 75, 80 },
{ 66, 102, 167, 99, 74, 62, 40, 234, 128 },
{ 41, 53, 9, 178, 241, 141, 26, 8, 107 },
{ 104, 79, 12, 27, 217, 255, 87, 17, 7 },
{ 74, 43, 26, 146, 73, 166, 49, 23, 157 },
{ 65, 38, 105, 160, 51, 52, 31, 115, 128 },
{ 87, 68, 71, 44, 114, 51, 15, 186, 23 },
{ 47, 41, 14, 110, 182, 183, 21, 17, 194 },
{ 66, 45, 25, 102, 197, 189, 23, 18, 22 }
},
{
{ 88, 88, 147, 150, 42, 46, 45, 196, 205 },
{ 43, 97, 183, 117, 85, 38, 35, 179, 61 },
{ 39, 53, 200, 87, 26, 21, 43, 232, 171 },
{ 56, 34, 51, 104, 114, 102, 29, 93, 77 },
{ 107, 54, 32, 26, 51, 1, 81, 43, 31 },
{ 39, 28, 85, 171, 58, 165, 90, 98, 64 },
{ 34, 22, 116, 206, 23, 34, 43, 166, 73 },
{ 68, 25, 106, 22, 64, 171, 36, 225, 114 },
{ 34, 19, 21, 102, 132, 188, 16, 76, 124 },
{ 62, 18, 78, 95, 85, 57, 50, 48, 51 }
},
{
{ 193, 101, 35, 159, 215, 111, 89, 46, 111 },
{ 60, 148, 31, 172, 219, 228, 21, 18, 111 },
{ 112, 113, 77, 85, 179, 255, 38, 120, 114 },
{ 40, 42, 1, 196, 245, 209, 10, 25, 109 },
{ 100, 80, 8, 43, 154, 1, 51, 26, 71 },
{ 88, 43, 29, 140, 166, 213, 37, 43, 154 },
{ 61, 63, 30, 155, 67, 45, 68, 1, 209 },
{ 142, 78, 78, 16, 255, 128, 34, 197, 171 },
{ 41, 40, 5, 102, 211, 183, 4, 1, 221 },
{ 51, 50, 17, 168, 209, 192, 23, 25, 82 }
},
{
{ 125, 98, 42, 88, 104, 85, 117, 175, 82 },
{ 95, 84, 53, 89, 128, 100, 113, 101, 45 },
{ 75, 79, 123, 47, 51, 128, 81, 171, 1 },
{ 57, 17, 5, 71, 102, 57, 53, 41, 49 },
{ 115, 21, 2, 10, 102, 255, 166, 23, 6 },
{ 38, 33, 13, 121, 57, 73, 26, 1, 85 },
{ 41, 10, 67, 138, 77, 110, 90, 47, 114 },
{ 101, 29, 16, 10, 85, 128, 101, 196, 26 },
{ 57, 18, 10, 102, 102, 213, 34, 20, 43 },
{ 117, 20, 15, 36, 163, 128, 68, 1, 26 }
},
{
{ 138, 31, 36, 171, 27, 166, 38, 44, 229 },
{ 67, 87, 58, 169, 82, 115, 26, 59, 179 },
{ 63, 59, 90, 180, 59, 166, 93, 73, 154 },
{ 40, 40, 21, 116, 143, 209, 34, 39, 175 },
{ 57, 46, 22, 24, 128, 1, 54, 17, 37 },
{ 47, 15, 16, 183, 34, 223, 49, 45, 183 },
{ 46, 17, 33, 183, 6, 98, 15, 32, 183 },
{ 65, 32, 73, 115, 28, 128, 23, 128, 205 },
{ 40, 3, 9, 115, 51, 192, 18, 6, 223 },
{ 87, 37, 9, 115, 59, 77, 64, 21, 47 }
},
{
{ 104, 55, 44, 218, 9, 54, 53, 130, 226 },
{ 64, 90, 70, 205, 40, 41, 23, 26, 57 },
{ 54, 57, 112, 184, 5, 41, 38, 166, 213 },
{ 30, 34, 26, 133, 152, 116, 10, 32, 134 },
{ 75, 32, 12, 51, 192, 255, 160, 43, 51 },
{ 39, 19, 53, 221, 26, 114, 32, 73, 255 },
{ 31, 9, 65, 234, 2, 15, 1, 118, 73 },
{ 88, 31, 35, 67, 102, 85, 55, 186, 85 },
{ 56, 21, 23, 111, 59, 205, 45, 37, 192 },
{ 55, 38, 70, 124, 73, 102, 1, 34, 98 }
},
{
{ 102, 61, 71, 37, 34, 53, 31, 243, 192 },
{ 69, 60, 71, 38, 73, 119, 28, 222, 37 },
{ 68, 45, 128, 34, 1, 47, 11, 245, 171 },
{ 62, 17, 19, 70, 146, 85, 55, 62, 70 },
{ 75, 15, 9, 9, 64, 255, 184, 119, 16 },
{ 37, 43, 37, 154, 100, 163, 85, 160, 1 },
{ 63, 9, 92, 136, 28, 64, 32, 201, 85 },
{ 86, 6, 28, 5, 64, 255, 25, 248, 1 },
{ 56, 8, 17, 132, 137, 255, 55, 116, 128 },
{ 58, 15, 20, 82, 135, 57, 26, 121, 40 }
},
{
{ 164, 50, 31, 137, 154, 133, 25, 35, 218 },
{ 51, 103, 44, 131, 131, 123, 31, 6, 158 },
{ 86, 40, 64, 135, 148, 224, 45, 183, 128 },
{ 22, 26, 17, 131, 240, 154, 14, 1, 209 },
{ 83, 12, 13, 54, 192, 255, 68, 47, 28 },
{ 45, 16, 21, 91, 64, 222, 7, 1, 197 },
{ 56, 21, 39, 155, 60, 138, 23, 102, 213 },
{ 85, 26, 85, 85, 128, 128, 32, 146, 171 },
{ 18, 11, 7, 63, 144, 171, 4, 4, 246 },
{ 35, 27, 10, 146, 174, 171, 12, 26, 128 }
},
{
{ 190, 80, 35, 99, 180, 80, 126, 54, 45 },
{ 85, 126, 47, 87, 176, 51, 41, 20, 32 },
{ 101, 75, 128, 139, 118, 146, 116, 128, 85 },
{ 56, 41, 15, 176, 236, 85, 37, 9, 62 },
{ 146, 36, 19, 30, 171, 255, 97, 27, 20 },
{ 71, 30, 17, 119, 118, 255, 17, 18, 138 },
{ 101, 38, 60, 138, 55, 70, 43, 26, 142 },
{ 138, 45, 61, 62, 219, 1, 81, 188, 64 },
{ 32, 41, 20, 117, 151, 142, 20, 21, 163 },
{ 112, 19, 12, 61, 195, 128, 48, 4, 24 }
}
};
const vp8_prob
vp8_kf_bmode_prob[VP8_BINTRAMODES][VP8_BINTRAMODES][VP8_BINTRAMODES - 1] = {
{ { 231, 120, 48, 89, 115, 113, 120, 152, 112 },
{ 152, 179, 64, 126, 170, 118, 46, 70, 95 },
{ 175, 69, 143, 80, 85, 82, 72, 155, 103 },
{ 56, 58, 10, 171, 218, 189, 17, 13, 152 },
{ 144, 71, 10, 38, 171, 213, 144, 34, 26 },
{ 114, 26, 17, 163, 44, 195, 21, 10, 173 },
{ 121, 24, 80, 195, 26, 62, 44, 64, 85 },
{ 170, 46, 55, 19, 136, 160, 33, 206, 71 },
{ 63, 20, 8, 114, 114, 208, 12, 9, 226 },
{ 81, 40, 11, 96, 182, 84, 29, 16, 36 } },
{ { 134, 183, 89, 137, 98, 101, 106, 165, 148 },
{ 72, 187, 100, 130, 157, 111, 32, 75, 80 },
{ 66, 102, 167, 99, 74, 62, 40, 234, 128 },
{ 41, 53, 9, 178, 241, 141, 26, 8, 107 },
{ 104, 79, 12, 27, 217, 255, 87, 17, 7 },
{ 74, 43, 26, 146, 73, 166, 49, 23, 157 },
{ 65, 38, 105, 160, 51, 52, 31, 115, 128 },
{ 87, 68, 71, 44, 114, 51, 15, 186, 23 },
{ 47, 41, 14, 110, 182, 183, 21, 17, 194 },
{ 66, 45, 25, 102, 197, 189, 23, 18, 22 } },
{ { 88, 88, 147, 150, 42, 46, 45, 196, 205 },
{ 43, 97, 183, 117, 85, 38, 35, 179, 61 },
{ 39, 53, 200, 87, 26, 21, 43, 232, 171 },
{ 56, 34, 51, 104, 114, 102, 29, 93, 77 },
{ 107, 54, 32, 26, 51, 1, 81, 43, 31 },
{ 39, 28, 85, 171, 58, 165, 90, 98, 64 },
{ 34, 22, 116, 206, 23, 34, 43, 166, 73 },
{ 68, 25, 106, 22, 64, 171, 36, 225, 114 },
{ 34, 19, 21, 102, 132, 188, 16, 76, 124 },
{ 62, 18, 78, 95, 85, 57, 50, 48, 51 } },
{ { 193, 101, 35, 159, 215, 111, 89, 46, 111 },
{ 60, 148, 31, 172, 219, 228, 21, 18, 111 },
{ 112, 113, 77, 85, 179, 255, 38, 120, 114 },
{ 40, 42, 1, 196, 245, 209, 10, 25, 109 },
{ 100, 80, 8, 43, 154, 1, 51, 26, 71 },
{ 88, 43, 29, 140, 166, 213, 37, 43, 154 },
{ 61, 63, 30, 155, 67, 45, 68, 1, 209 },
{ 142, 78, 78, 16, 255, 128, 34, 197, 171 },
{ 41, 40, 5, 102, 211, 183, 4, 1, 221 },
{ 51, 50, 17, 168, 209, 192, 23, 25, 82 } },
{ { 125, 98, 42, 88, 104, 85, 117, 175, 82 },
{ 95, 84, 53, 89, 128, 100, 113, 101, 45 },
{ 75, 79, 123, 47, 51, 128, 81, 171, 1 },
{ 57, 17, 5, 71, 102, 57, 53, 41, 49 },
{ 115, 21, 2, 10, 102, 255, 166, 23, 6 },
{ 38, 33, 13, 121, 57, 73, 26, 1, 85 },
{ 41, 10, 67, 138, 77, 110, 90, 47, 114 },
{ 101, 29, 16, 10, 85, 128, 101, 196, 26 },
{ 57, 18, 10, 102, 102, 213, 34, 20, 43 },
{ 117, 20, 15, 36, 163, 128, 68, 1, 26 } },
{ { 138, 31, 36, 171, 27, 166, 38, 44, 229 },
{ 67, 87, 58, 169, 82, 115, 26, 59, 179 },
{ 63, 59, 90, 180, 59, 166, 93, 73, 154 },
{ 40, 40, 21, 116, 143, 209, 34, 39, 175 },
{ 57, 46, 22, 24, 128, 1, 54, 17, 37 },
{ 47, 15, 16, 183, 34, 223, 49, 45, 183 },
{ 46, 17, 33, 183, 6, 98, 15, 32, 183 },
{ 65, 32, 73, 115, 28, 128, 23, 128, 205 },
{ 40, 3, 9, 115, 51, 192, 18, 6, 223 },
{ 87, 37, 9, 115, 59, 77, 64, 21, 47 } },
{ { 104, 55, 44, 218, 9, 54, 53, 130, 226 },
{ 64, 90, 70, 205, 40, 41, 23, 26, 57 },
{ 54, 57, 112, 184, 5, 41, 38, 166, 213 },
{ 30, 34, 26, 133, 152, 116, 10, 32, 134 },
{ 75, 32, 12, 51, 192, 255, 160, 43, 51 },
{ 39, 19, 53, 221, 26, 114, 32, 73, 255 },
{ 31, 9, 65, 234, 2, 15, 1, 118, 73 },
{ 88, 31, 35, 67, 102, 85, 55, 186, 85 },
{ 56, 21, 23, 111, 59, 205, 45, 37, 192 },
{ 55, 38, 70, 124, 73, 102, 1, 34, 98 } },
{ { 102, 61, 71, 37, 34, 53, 31, 243, 192 },
{ 69, 60, 71, 38, 73, 119, 28, 222, 37 },
{ 68, 45, 128, 34, 1, 47, 11, 245, 171 },
{ 62, 17, 19, 70, 146, 85, 55, 62, 70 },
{ 75, 15, 9, 9, 64, 255, 184, 119, 16 },
{ 37, 43, 37, 154, 100, 163, 85, 160, 1 },
{ 63, 9, 92, 136, 28, 64, 32, 201, 85 },
{ 86, 6, 28, 5, 64, 255, 25, 248, 1 },
{ 56, 8, 17, 132, 137, 255, 55, 116, 128 },
{ 58, 15, 20, 82, 135, 57, 26, 121, 40 } },
{ { 164, 50, 31, 137, 154, 133, 25, 35, 218 },
{ 51, 103, 44, 131, 131, 123, 31, 6, 158 },
{ 86, 40, 64, 135, 148, 224, 45, 183, 128 },
{ 22, 26, 17, 131, 240, 154, 14, 1, 209 },
{ 83, 12, 13, 54, 192, 255, 68, 47, 28 },
{ 45, 16, 21, 91, 64, 222, 7, 1, 197 },
{ 56, 21, 39, 155, 60, 138, 23, 102, 213 },
{ 85, 26, 85, 85, 128, 128, 32, 146, 171 },
{ 18, 11, 7, 63, 144, 171, 4, 4, 246 },
{ 35, 27, 10, 146, 174, 171, 12, 26, 128 } },
{ { 190, 80, 35, 99, 180, 80, 126, 54, 45 },
{ 85, 126, 47, 87, 176, 51, 41, 20, 32 },
{ 101, 75, 128, 139, 118, 146, 116, 128, 85 },
{ 56, 41, 15, 176, 236, 85, 37, 9, 62 },
{ 146, 36, 19, 30, 171, 255, 97, 27, 20 },
{ 71, 30, 17, 119, 118, 255, 17, 18, 138 },
{ 101, 38, 60, 138, 55, 70, 43, 26, 142 },
{ 138, 45, 61, 62, 219, 1, 81, 188, 64 },
{ 32, 41, 20, 117, 151, 142, 20, 21, 163 },
{ 112, 19, 12, 61, 195, 128, 48, 4, 24 } }
};
#ifdef __cplusplus
} // extern "C"

937
vp8/common/vp8_loopfilter.c
View File

File diff suppressed because it is too large Load Diff

34
vp8/common/x86/filter_x86.c
View File

@ -10,26 +10,20 @@
#include "vp8/common/x86/filter_x86.h"
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_4[8][8]) =
{
{ 128, 128, 128, 128, 0, 0, 0, 0 },
{ 112, 112, 112, 112, 16, 16, 16, 16 },
{ 96, 96, 96, 96, 32, 32, 32, 32 },
{ 80, 80, 80, 80, 48, 48, 48, 48 },
{ 64, 64, 64, 64, 64, 64, 64, 64 },
{ 48, 48, 48, 48, 80, 80, 80, 80 },
{ 32, 32, 32, 32, 96, 96, 96, 96 },
{ 16, 16, 16, 16, 112, 112, 112, 112 }
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_4[8][8]) = {
{ 128, 128, 128, 128, 0, 0, 0, 0 }, { 112, 112, 112, 112, 16, 16, 16, 16 },
{ 96, 96, 96, 96, 32, 32, 32, 32 }, { 80, 80, 80, 80, 48, 48, 48, 48 },
{ 64, 64, 64, 64, 64, 64, 64, 64 }, { 48, 48, 48, 48, 80, 80, 80, 80 },
{ 32, 32, 32, 32, 96, 96, 96, 96 }, { 16, 16, 16, 16, 112, 112, 112, 112 }
};
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_8[8][16]) =
{
{ 128, 128, 128, 128, 128, 128, 128, 128, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 112, 112, 112, 112, 112, 112, 112, 112, 16, 16, 16, 16, 16, 16, 16, 16 },
{ 96, 96, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32 },
{ 80, 80, 80, 80, 80, 80, 80, 80, 48, 48, 48, 48, 48, 48, 48, 48 },
{ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64 },
{ 48, 48, 48, 48, 48, 48, 48, 48, 80, 80, 80, 80, 80, 80, 80, 80 },
{ 32, 32, 32, 32, 32, 32, 32, 32, 96, 96, 96, 96, 96, 96, 96, 96 },
{ 16, 16, 16, 16, 16, 16, 16, 16, 112, 112, 112, 112, 112, 112, 112, 112 }
DECLARE_ALIGNED(16, const short, vp8_bilinear_filters_x86_8[8][16]) = {
{ 128, 128, 128, 128, 128, 128, 128, 128, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 112, 112, 112, 112, 112, 112, 112, 112, 16, 16, 16, 16, 16, 16, 16, 16 },
{ 96, 96, 96, 96, 96, 96, 96, 96, 32, 32, 32, 32, 32, 32, 32, 32 },
{ 80, 80, 80, 80, 80, 80, 80, 80, 48, 48, 48, 48, 48, 48, 48, 48 },
{ 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64 },
{ 48, 48, 48, 48, 48, 48, 48, 48, 80, 80, 80, 80, 80, 80, 80, 80 },
{ 32, 32, 32, 32, 32, 32, 32, 32, 96, 96, 96, 96, 96, 96, 96, 96 },
{ 16, 16, 16, 16, 16, 16, 16, 16, 112, 112, 112, 112, 112, 112, 112, 112 }
};

181
vp8/common/x86/idct_blk_mmx.c
View File

@ -15,114 +15,97 @@
extern void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q);
void vp8_dequantize_b_mmx(BLOCKD *d, short *DQC)
{
short *sq = (short *) d->qcoeff;
short *dq = (short *) d->dqcoeff;
void vp8_dequantize_b_mmx(BLOCKD *d, short *DQC) {
short *sq = (short *)d->qcoeff;
short *dq = (short *)d->dqcoeff;
vp8_dequantize_b_impl_mmx(sq, dq, DQC);
vp8_dequantize_b_impl_mmx(sq, dq, DQC);
}
void vp8_dequant_idct_add_y_block_mmx
(short *q, short *dq,
unsigned char *dst, int stride, char *eobs)
{
int i;
void vp8_dequant_idct_add_y_block_mmx(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i;
for (i = 0; i < 4; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx (q, dq, dst, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_mmx (q[0]*dq[0], dst, stride, dst, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx (q+16, dq, dst+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_mmx (q[16]*dq[0], dst+4, stride,
dst+4, stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
if (eobs[2] > 1)
vp8_dequant_idct_add_mmx (q+32, dq, dst+8, stride);
else if (eobs[2] == 1)
{
vp8_dc_only_idct_add_mmx (q[32]*dq[0], dst+8, stride,
dst+8, stride);
memset(q + 32, 0, 2 * sizeof(q[0]));
}
if (eobs[3] > 1)
vp8_dequant_idct_add_mmx (q+48, dq, dst+12, stride);
else if (eobs[3] == 1)
{
vp8_dc_only_idct_add_mmx (q[48]*dq[0], dst+12, stride,
dst+12, stride);
memset(q + 48, 0, 2 * sizeof(q[0]));
}
q += 64;
dst += 4*stride;
eobs += 4;
for (i = 0; i < 4; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx(q, dq, dst, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_mmx(q[0] * dq[0], dst, stride, dst, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx(q + 16, dq, dst + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_mmx(q[16] * dq[0], dst + 4, stride, dst + 4, stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
if (eobs[2] > 1)
vp8_dequant_idct_add_mmx(q + 32, dq, dst + 8, stride);
else if (eobs[2] == 1) {
vp8_dc_only_idct_add_mmx(q[32] * dq[0], dst + 8, stride, dst + 8, stride);
memset(q + 32, 0, 2 * sizeof(q[0]));
}
if (eobs[3] > 1)
vp8_dequant_idct_add_mmx(q + 48, dq, dst + 12, stride);
else if (eobs[3] == 1) {
vp8_dc_only_idct_add_mmx(q[48] * dq[0], dst + 12, stride, dst + 12,
stride);
memset(q + 48, 0, 2 * sizeof(q[0]));
}
q += 64;
dst += 4 * stride;
eobs += 4;
}
}
void vp8_dequant_idct_add_uv_block_mmx
(short *q, short *dq,
unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
{
int i;
void vp8_dequant_idct_add_uv_block_mmx(short *q, short *dq, unsigned char *dstu,
unsigned char *dstv, int stride,
char *eobs) {
int i;
for (i = 0; i < 2; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx (q, dq, dstu, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_mmx (q[0]*dq[0], dstu, stride, dstu, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx (q+16, dq, dstu+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_mmx (q[16]*dq[0], dstu+4, stride,
dstu+4, stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
q += 32;
dstu += 4*stride;
eobs += 2;
for (i = 0; i < 2; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx(q, dq, dstu, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_mmx(q[0] * dq[0], dstu, stride, dstu, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
for (i = 0; i < 2; i++)
{
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx (q, dq, dstv, stride);
else if (eobs[0] == 1)
{
vp8_dc_only_idct_add_mmx (q[0]*dq[0], dstv, stride, dstv, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx (q+16, dq, dstv+4, stride);
else if (eobs[1] == 1)
{
vp8_dc_only_idct_add_mmx (q[16]*dq[0], dstv+4, stride,
dstv+4, stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
q += 32;
dstv += 4*stride;
eobs += 2;
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx(q + 16, dq, dstu + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_mmx(q[16] * dq[0], dstu + 4, stride, dstu + 4,
stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
q += 32;
dstu += 4 * stride;
eobs += 2;
}
for (i = 0; i < 2; i++) {
if (eobs[0] > 1)
vp8_dequant_idct_add_mmx(q, dq, dstv, stride);
else if (eobs[0] == 1) {
vp8_dc_only_idct_add_mmx(q[0] * dq[0], dstv, stride, dstv, stride);
memset(q, 0, 2 * sizeof(q[0]));
}
if (eobs[1] > 1)
vp8_dequant_idct_add_mmx(q + 16, dq, dstv + 4, stride);
else if (eobs[1] == 1) {
vp8_dc_only_idct_add_mmx(q[16] * dq[0], dstv + 4, stride, dstv + 4,
stride);
memset(q + 16, 0, 2 * sizeof(q[0]));
}
q += 32;
dstv += 4 * stride;
eobs += 2;
}
}

131
vp8/common/x86/idct_blk_sse2.c
View File

@ -11,79 +11,68 @@
#include "vpx_config.h"
#include "vp8_rtcd.h"
void vp8_idct_dequant_0_2x_sse2
(short *q, short *dq ,
unsigned char *dst, int dst_stride);
void vp8_idct_dequant_full_2x_sse2
(short *q, short *dq ,
unsigned char *dst, int dst_stride);
void vp8_idct_dequant_0_2x_sse2(short *q, short *dq, unsigned char *dst,
int dst_stride);
void vp8_idct_dequant_full_2x_sse2(short *q, short *dq, unsigned char *dst,
int dst_stride);
void vp8_dequant_idct_add_y_block_sse2
(short *q, short *dq,
unsigned char *dst, int stride, char *eobs)
{
int i;
void vp8_dequant_idct_add_y_block_sse2(short *q, short *dq, unsigned char *dst,
int stride, char *eobs) {
int i;
for (i = 0; i < 4; i++)
{
if (((short *)(eobs))[0])
{
if (((short *)(eobs))[0] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q, dq, dst, stride);
else
vp8_idct_dequant_0_2x_sse2 (q, dq, dst, stride);
}
if (((short *)(eobs))[1])
{
if (((short *)(eobs))[1] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q+32, dq, dst+8, stride);
else
vp8_idct_dequant_0_2x_sse2 (q+32, dq, dst+8, stride);
}
q += 64;
dst += stride*4;
eobs += 4;
}
}
void vp8_dequant_idct_add_uv_block_sse2
(short *q, short *dq,
unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
{
if (((short *)(eobs))[0])
{
if (((short *)(eobs))[0] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q, dq, dstu, stride);
else
vp8_idct_dequant_0_2x_sse2 (q, dq, dstu, stride);
}
q += 32;
dstu += stride*4;
if (((short *)(eobs))[1])
{
if (((short *)(eobs))[1] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q, dq, dstu, stride);
else
vp8_idct_dequant_0_2x_sse2 (q, dq, dstu, stride);
}
q += 32;
if (((short *)(eobs))[2])
{
if (((short *)(eobs))[2] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q, dq, dstv, stride);
else
vp8_idct_dequant_0_2x_sse2 (q, dq, dstv, stride);
}
q += 32;
dstv += stride*4;
if (((short *)(eobs))[3])
{
if (((short *)(eobs))[3] & 0xfefe)
vp8_idct_dequant_full_2x_sse2 (q, dq, dstv, stride);
for (i = 0; i < 4; i++) {
if (((short *)(eobs))[0]) {
if (((short *)(eobs))[0] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q, dq, dst, stride);
else
vp8_idct_dequant_0_2x_sse2 (q, dq, dstv, stride);
vp8_idct_dequant_0_2x_sse2(q, dq, dst, stride);
}
if (((short *)(eobs))[1]) {
if (((short *)(eobs))[1] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q + 32, dq, dst + 8, stride);
else
vp8_idct_dequant_0_2x_sse2(q + 32, dq, dst + 8, stride);
}
q += 64;
dst += stride * 4;
eobs += 4;
}
}
void vp8_dequant_idct_add_uv_block_sse2(short *q, short *dq,
unsigned char *dstu,
unsigned char *dstv, int stride,
char *eobs) {
if (((short *)(eobs))[0]) {
if (((short *)(eobs))[0] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q, dq, dstu, stride);
else
vp8_idct_dequant_0_2x_sse2(q, dq, dstu, stride);
}
q += 32;
dstu += stride * 4;
if (((short *)(eobs))[1]) {
if (((short *)(eobs))[1] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q, dq, dstu, stride);
else
vp8_idct_dequant_0_2x_sse2(q, dq, dstu, stride);
}
q += 32;
if (((short *)(eobs))[2]) {
if (((short *)(eobs))[2] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q, dq, dstv, stride);
else
vp8_idct_dequant_0_2x_sse2(q, dq, dstv, stride);
}
q += 32;
dstv += stride * 4;
if (((short *)(eobs))[3]) {
if (((short *)(eobs))[3] & 0xfefe)
vp8_idct_dequant_full_2x_sse2(q, dq, dstv, stride);
else
vp8_idct_dequant_0_2x_sse2(q, dq, dstv, stride);
}
}

235
vp8/common/x86/loopfilter_x86.c
View File

@ -8,20 +8,19 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_config.h"
#include "vp8/common/loopfilter.h"
#define prototype_loopfilter(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
const unsigned char *limit, const unsigned char *thresh, int count)
#define prototype_loopfilter(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit, \
const unsigned char *limit, const unsigned char *thresh, int count)
#define prototype_loopfilter_nc(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit,\
const unsigned char *limit, const unsigned char *thresh)
#define prototype_loopfilter_nc(sym) \
void sym(unsigned char *src, int pitch, const unsigned char *blimit, \
const unsigned char *limit, const unsigned char *thresh)
#define prototype_simple_loopfilter(sym) \
void sym(unsigned char *y, int ystride, const unsigned char *blimit)
void sym(unsigned char *y, int ystride, const unsigned char *blimit)
prototype_loopfilter(vp8_mbloop_filter_vertical_edge_mmx);
prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_mmx);
@ -47,152 +46,178 @@ extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_sse2;
#if HAVE_MMX
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, 1);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_mbv_mmx(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->mblim, lfi->lim,
lfi->hev_thr, 1);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_bh_mmx(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, lfi->blim,
lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr, 1);
}
void vp8_loop_filter_bhs_mmx(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
{
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, blimit);
void vp8_loop_filter_bhs_mmx(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 4 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 8 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride,
blimit);
}
/* Vertical B Filtering */
void vp8_loop_filter_bv_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_loop_filter_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
void vp8_loop_filter_bv_mmx(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_loop_filter_vertical_edge_mmx(y_ptr + 4, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 8, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 2);
if (u_ptr)
vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (u_ptr)
vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, 1);
if (v_ptr)
vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->blim, lfi->lim,
lfi->hev_thr, 1);
}
void vp8_loop_filter_bvs_mmx(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
{
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 12, y_stride, blimit);
void vp8_loop_filter_bvs_mmx(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_mmx(y_ptr + 12, y_stride, blimit);
}
#endif
/* Horizontal MB filtering */
#if HAVE_SSE2
void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr);
void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, v_ptr);
if (u_ptr)
vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, v_ptr);
}
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim, lfi->hev_thr);
void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mblim, lfi->lim,
lfi->hev_thr);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim, lfi->lim, lfi->hev_thr, v_ptr);
if (u_ptr)
vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->mblim,
lfi->lim, lfi->hev_thr, v_ptr);
}
/* Horizontal B Filtering */
void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
#if ARCH_X86_64
vp8_loop_filter_bh_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_bh_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr,
2);
#else
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride,
lfi->blim, lfi->lim, lfi->hev_thr);
#endif
if (u_ptr)
vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, v_ptr + 4 * uv_stride);
if (u_ptr)
vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride,
lfi->blim, lfi->lim, lfi->hev_thr,
v_ptr + 4 * uv_stride);
}
void vp8_loop_filter_bhs_sse2(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
{
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride, blimit);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, blimit);
void vp8_loop_filter_bhs_sse2(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 4 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 8 * y_stride, y_stride,
blimit);
vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride,
blimit);
}
/* Vertical B Filtering */
void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi)
{
void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr,
unsigned char *v_ptr, int y_stride, int uv_stride,
loop_filter_info *lfi) {
#if ARCH_X86_64
vp8_loop_filter_bv_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr, 2);
vp8_loop_filter_bv_y_sse2(y_ptr, y_stride, lfi->blim, lfi->lim, lfi->hev_thr,
2);
#else
vp8_loop_filter_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->blim, lfi->lim, lfi->hev_thr);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 4, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 8, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr);
vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->blim, lfi->lim,
lfi->hev_thr);
#endif
if (u_ptr)
vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->blim, lfi->lim, lfi->hev_thr, v_ptr + 4);
if (u_ptr)
vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->blim,
lfi->lim, lfi->hev_thr, v_ptr + 4);
}
void vp8_loop_filter_bvs_sse2(unsigned char *y_ptr, int y_stride, const unsigned char *blimit)
{
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, blimit);
void vp8_loop_filter_bvs_sse2(unsigned char *y_ptr, int y_stride,
const unsigned char *blimit) {
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 4, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 8, y_stride, blimit);
vp8_loop_filter_simple_vertical_edge_sse2(y_ptr + 12, y_stride, blimit);
}
#endif

935
vp8/common/x86/vp8_asm_stubs.c
View File

File diff suppressed because it is too large Load Diff

99
vp8/decoder/dboolhuff.c
View File

@ -8,70 +8,61 @@
* be found in the AUTHORS file in the root of the source tree.
*/
#include "dboolhuff.h"
#include "vp8/common/common.h"
#include "vpx_dsp/vpx_dsp_common.h"
int vp8dx_start_decode(BOOL_DECODER *br,
const unsigned char *source,
unsigned int source_sz,
vpx_decrypt_cb decrypt_cb,
void *decrypt_state)
{
br->user_buffer_end = source+source_sz;
br->user_buffer = source;
br->value = 0;
br->count = -8;
br->range = 255;
br->decrypt_cb = decrypt_cb;
br->decrypt_state = decrypt_state;
int vp8dx_start_decode(BOOL_DECODER *br, const unsigned char *source,
unsigned int source_sz, vpx_decrypt_cb decrypt_cb,
void *decrypt_state) {
br->user_buffer_end = source + source_sz;
br->user_buffer = source;
br->value = 0;
br->count = -8;
br->range = 255;
br->decrypt_cb = decrypt_cb;
br->decrypt_state = decrypt_state;
if (source_sz && !source)
return 1;
if (source_sz && !source) return 1;
/* Populate the buffer */
vp8dx_bool_decoder_fill(br);
/* Populate the buffer */
vp8dx_bool_decoder_fill(br);
return 0;
return 0;
}
void vp8dx_bool_decoder_fill(BOOL_DECODER *br)
{
const unsigned char *bufptr = br->user_buffer;
VP8_BD_VALUE value = br->value;
int count = br->count;
int shift = VP8_BD_VALUE_SIZE - CHAR_BIT - (count + CHAR_BIT);
size_t bytes_left = br->user_buffer_end - bufptr;
size_t bits_left = bytes_left * CHAR_BIT;
int x = shift + CHAR_BIT - (int)bits_left;
int loop_end = 0;
unsigned char decrypted[sizeof(VP8_BD_VALUE) + 1];
void vp8dx_bool_decoder_fill(BOOL_DECODER *br) {
const unsigned char *bufptr = br->user_buffer;
VP8_BD_VALUE value = br->value;
int count = br->count;
int shift = VP8_BD_VALUE_SIZE - CHAR_BIT - (count + CHAR_BIT);
size_t bytes_left = br->user_buffer_end - bufptr;
size_t bits_left = bytes_left * CHAR_BIT;
int x = shift + CHAR_BIT - (int)bits_left;
int loop_end = 0;
unsigned char decrypted[sizeof(VP8_BD_VALUE) + 1];
if (br->decrypt_cb) {
size_t n = VPXMIN(sizeof(decrypted), bytes_left);
br->decrypt_cb(br->decrypt_state, bufptr, decrypted, (int)n);
bufptr = decrypted;
if (br->decrypt_cb) {
size_t n = VPXMIN(sizeof(decrypted), bytes_left);
br->decrypt_cb(br->decrypt_state, bufptr, decrypted, (int)n);
bufptr = decrypted;
}
if (x >= 0) {
count += VP8_LOTS_OF_BITS;
loop_end = x;
}
if (x < 0 || bits_left) {
while (shift >= loop_end) {
count += CHAR_BIT;
value |= (VP8_BD_VALUE)*bufptr << shift;
++bufptr;
++br->user_buffer;
shift -= CHAR_BIT;
}
}
if(x >= 0)
{
count += VP8_LOTS_OF_BITS;
loop_end = x;
}
if (x < 0 || bits_left)
{
while(shift >= loop_end)
{
count += CHAR_BIT;
value |= (VP8_BD_VALUE)*bufptr << shift;
++bufptr;
++br->user_buffer;
shift -= CHAR_BIT;
}
}
br->value = value;
br->count = count;
br->value = value;
br->count = count;
}

Some files were not shown because too many files have changed in this diff Show More