generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
f299d628f3
vpx/vp8/encoder/variance.h

430 lines
12 KiB
C
Raw Normal View History

Initial WebM release
2010-05-18 17:58:33 +02:00
/*
Use WebM in copyright notice for consistency Changes 'The VP8 project' to 'The WebM project', for consistency with other webmproject.org repositories. Fixes issue #97. Change-Id: I37c13ed5fbdb9d334ceef71c6350e9febed9bbba
2010-09-09 14:16:39 +02:00
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
Initial WebM release
2010-05-18 17:58:33 +02:00
*
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
* Use of this source code is governed by a BSD-style license
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
* in the file PATENTS. All contributing project authors may
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
* be found in the AUTHORS file in the root of the source tree.
Initial WebM release
2010-05-18 17:58:33 +02:00
*/
#ifndef VARIANCE_H
#define VARIANCE_H
#define prototype_sad(sym)\
unsigned int (sym)\
(\
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *ref_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int ref_stride, \
int max_sad\
)
#define prototype_sad_multi_same_address(sym)\
void (sym)\
(\
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *ref_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int ref_stride, \
unsigned int *sad_array\
)
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
#define prototype_sad_multi_same_address_1(sym)\
void (sym)\
(\
const unsigned char *src_ptr, \
int source_stride, \
const unsigned char *ref_ptr, \
int ref_stride, \
unsigned short *sad_array\
)
Initial WebM release
2010-05-18 17:58:33 +02:00
#define prototype_sad_multi_dif_address(sym)\
void (sym)\
(\
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
unsigned char *ref_ptr[4], \
int ref_stride, \
unsigned int *sad_array\
)
#define prototype_variance(sym) \
unsigned int (sym) \
(\
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *ref_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int ref_stride, \
unsigned int *sse\
)
#define prototype_variance2(sym) \
unsigned int (sym) \
(\
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *ref_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int ref_stride, \
unsigned int *sse,\
int *sum\
)
#define prototype_subpixvariance(sym) \
unsigned int (sym) \
( \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *src_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int source_stride, \
int xoffset, \
int yoffset, \
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
const unsigned char *ref_ptr, \
Initial WebM release
2010-05-18 17:58:33 +02:00
int Refstride, \
unsigned int *sse \
);
vp8cx- alternate ssim function with optimizations Change-Id: I91921b0a90dbaddc7010380b038955be347964b3
2011-03-08 15:05:18 +01:00
#define prototype_ssimpf(sym) \
void (sym) \
( \
unsigned char *s, \
int sp, \
unsigned char *r, \
int rp, \
unsigned long *sum_s, \
unsigned long *sum_r, \
unsigned long *sum_sq_s, \
unsigned long *sum_sq_r, \
unsigned long *sum_sxr \
);
Initial WebM release
2010-05-18 17:58:33 +02:00
Add const qualifiers to variance/SAD functions. These functions should never change their input, and there's no reason not to declare that. This allows them to be passed static const data. Change-Id: Ia49fe4b01e80e9afcb24b4844817694d4da5995c
2010-10-11 23:01:23 +02:00
#define prototype_getmbss(sym) unsigned int (sym)(const short *)
Initial WebM release
2010-05-18 17:58:33 +02:00
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
#define prototype_get16x16prederror(sym)\
unsigned int (sym)\
(\
const unsigned char *src_ptr, \
int source_stride, \
const unsigned char *ref_ptr, \
int ref_stride \
)
Initial WebM release
2010-05-18 17:58:33 +02:00
#if ARCH_X86 || ARCH_X86_64
#include "x86/variance_x86.h"
#endif
#if ARCH_ARM
#include "arm/variance_arm.h"
#endif
#ifndef vp8_variance_sad4x4
#define vp8_variance_sad4x4 vp8_sad4x4_c
#endif
extern prototype_sad(vp8_variance_sad4x4);
#ifndef vp8_variance_sad8x8
#define vp8_variance_sad8x8 vp8_sad8x8_c
#endif
extern prototype_sad(vp8_variance_sad8x8);
#ifndef vp8_variance_sad8x16
#define vp8_variance_sad8x16 vp8_sad8x16_c
#endif
extern prototype_sad(vp8_variance_sad8x16);
#ifndef vp8_variance_sad16x8
#define vp8_variance_sad16x8 vp8_sad16x8_c
#endif
extern prototype_sad(vp8_variance_sad16x8);
#ifndef vp8_variance_sad16x16
#define vp8_variance_sad16x16 vp8_sad16x16_c
#endif
extern prototype_sad(vp8_variance_sad16x16);
//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
#ifndef vp8_variance_sad16x16x3
#define vp8_variance_sad16x16x3 vp8_sad16x16x3_c
#endif
extern prototype_sad_multi_same_address(vp8_variance_sad16x16x3);
#ifndef vp8_variance_sad16x8x3
#define vp8_variance_sad16x8x3 vp8_sad16x8x3_c
#endif
extern prototype_sad_multi_same_address(vp8_variance_sad16x8x3);
#ifndef vp8_variance_sad8x8x3
#define vp8_variance_sad8x8x3 vp8_sad8x8x3_c
#endif
extern prototype_sad_multi_same_address(vp8_variance_sad8x8x3);
#ifndef vp8_variance_sad8x16x3
#define vp8_variance_sad8x16x3 vp8_sad8x16x3_c
#endif
extern prototype_sad_multi_same_address(vp8_variance_sad8x16x3);
#ifndef vp8_variance_sad4x4x3
#define vp8_variance_sad4x4x3 vp8_sad4x4x3_c
#endif
extern prototype_sad_multi_same_address(vp8_variance_sad4x4x3);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
#ifndef vp8_variance_sad16x16x8
#define vp8_variance_sad16x16x8 vp8_sad16x16x8_c
#endif
extern prototype_sad_multi_same_address_1(vp8_variance_sad16x16x8);
#ifndef vp8_variance_sad16x8x8
#define vp8_variance_sad16x8x8 vp8_sad16x8x8_c
#endif
extern prototype_sad_multi_same_address_1(vp8_variance_sad16x8x8);
#ifndef vp8_variance_sad8x8x8
#define vp8_variance_sad8x8x8 vp8_sad8x8x8_c
#endif
extern prototype_sad_multi_same_address_1(vp8_variance_sad8x8x8);
#ifndef vp8_variance_sad8x16x8
#define vp8_variance_sad8x16x8 vp8_sad8x16x8_c
#endif
extern prototype_sad_multi_same_address_1(vp8_variance_sad8x16x8);
#ifndef vp8_variance_sad4x4x8
#define vp8_variance_sad4x4x8 vp8_sad4x4x8_c
#endif
extern prototype_sad_multi_same_address_1(vp8_variance_sad4x4x8);
Initial WebM release
2010-05-18 17:58:33 +02:00
//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
#ifndef vp8_variance_sad16x16x4d
#define vp8_variance_sad16x16x4d vp8_sad16x16x4d_c
#endif
extern prototype_sad_multi_dif_address(vp8_variance_sad16x16x4d);
#ifndef vp8_variance_sad16x8x4d
#define vp8_variance_sad16x8x4d vp8_sad16x8x4d_c
#endif
extern prototype_sad_multi_dif_address(vp8_variance_sad16x8x4d);
#ifndef vp8_variance_sad8x8x4d
#define vp8_variance_sad8x8x4d vp8_sad8x8x4d_c
#endif
extern prototype_sad_multi_dif_address(vp8_variance_sad8x8x4d);
#ifndef vp8_variance_sad8x16x4d
#define vp8_variance_sad8x16x4d vp8_sad8x16x4d_c
#endif
extern prototype_sad_multi_dif_address(vp8_variance_sad8x16x4d);
#ifndef vp8_variance_sad4x4x4d
#define vp8_variance_sad4x4x4d vp8_sad4x4x4d_c
#endif
extern prototype_sad_multi_dif_address(vp8_variance_sad4x4x4d);
//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
#ifndef vp8_variance_var4x4
#define vp8_variance_var4x4 vp8_variance4x4_c
#endif
extern prototype_variance(vp8_variance_var4x4);
#ifndef vp8_variance_var8x8
#define vp8_variance_var8x8 vp8_variance8x8_c
#endif
extern prototype_variance(vp8_variance_var8x8);
#ifndef vp8_variance_var8x16
#define vp8_variance_var8x16 vp8_variance8x16_c
#endif
extern prototype_variance(vp8_variance_var8x16);
#ifndef vp8_variance_var16x8
#define vp8_variance_var16x8 vp8_variance16x8_c
#endif
extern prototype_variance(vp8_variance_var16x8);
#ifndef vp8_variance_var16x16
#define vp8_variance_var16x16 vp8_variance16x16_c
#endif
extern prototype_variance(vp8_variance_var16x16);
//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
#ifndef vp8_variance_subpixvar4x4
#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixvar4x4);
#ifndef vp8_variance_subpixvar8x8
#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixvar8x8);
#ifndef vp8_variance_subpixvar8x16
#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixvar8x16);
#ifndef vp8_variance_subpixvar16x8
#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixvar16x8);
#ifndef vp8_variance_subpixvar16x16
#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixvar16x16);
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
#ifndef vp8_variance_halfpixvar16x16_h
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_c
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
#endif
extern prototype_variance(vp8_variance_halfpixvar16x16_h);
#ifndef vp8_variance_halfpixvar16x16_v
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_c
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
#endif
extern prototype_variance(vp8_variance_halfpixvar16x16_v);
#ifndef vp8_variance_halfpixvar16x16_hv
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_c
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
#endif
extern prototype_variance(vp8_variance_halfpixvar16x16_hv);
Initial WebM release
2010-05-18 17:58:33 +02:00
#ifndef vp8_variance_subpixmse16x16
#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_c
#endif
extern prototype_subpixvariance(vp8_variance_subpixmse16x16);
//-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
#ifndef vp8_variance_getmbss
#define vp8_variance_getmbss vp8_get_mb_ss_c
#endif
extern prototype_getmbss(vp8_variance_getmbss);
#ifndef vp8_variance_mse16x16
#define vp8_variance_mse16x16 vp8_mse16x16_c
#endif
extern prototype_variance(vp8_variance_mse16x16);
#ifndef vp8_variance_get16x16prederror
#define vp8_variance_get16x16prederror vp8_get16x16pred_error_c
#endif
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
extern prototype_get16x16prederror(vp8_variance_get16x16prederror);
Initial WebM release
2010-05-18 17:58:33 +02:00
#ifndef vp8_variance_get8x8var
#define vp8_variance_get8x8var vp8_get8x8var_c
#endif
extern prototype_variance2(vp8_variance_get8x8var);
#ifndef vp8_variance_get16x16var
#define vp8_variance_get16x16var vp8_get16x16var_c
#endif
extern prototype_variance2(vp8_variance_get16x16var);
#ifndef vp8_variance_get4x4sse_cs
#define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_c
#endif
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
extern prototype_get16x16prederror(vp8_variance_get4x4sse_cs);
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8cx- alternate ssim function with optimizations Change-Id: I91921b0a90dbaddc7010380b038955be347964b3
2011-03-08 15:05:18 +01:00
#ifndef vp8_ssimpf
#define vp8_ssimpf ssim_parms_c
#endif
extern prototype_ssimpf(vp8_ssimpf)
#ifndef vp8_ssimpf_8x8
#define vp8_ssimpf_8x8 ssim_parms_8x8_c
#endif
extern prototype_ssimpf(vp8_ssimpf_8x8)
Initial WebM release
2010-05-18 17:58:33 +02:00
typedef prototype_sad(*vp8_sad_fn_t);
typedef prototype_sad_multi_same_address(*vp8_sad_multi_fn_t);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
typedef prototype_sad_multi_same_address_1(*vp8_sad_multi1_fn_t);
Initial WebM release
2010-05-18 17:58:33 +02:00
typedef prototype_sad_multi_dif_address(*vp8_sad_multi_d_fn_t);
typedef prototype_variance(*vp8_variance_fn_t);
typedef prototype_variance2(*vp8_variance2_fn_t);
typedef prototype_subpixvariance(*vp8_subpixvariance_fn_t);
typedef prototype_getmbss(*vp8_getmbss_fn_t);
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
typedef prototype_ssimpf(*vp8_ssimpf_fn_t);
typedef prototype_get16x16prederror(*vp8_get16x16prederror_fn_t);
vp8cx- alternate ssim function with optimizations Change-Id: I91921b0a90dbaddc7010380b038955be347964b3
2011-03-08 15:05:18 +01:00
Initial WebM release
2010-05-18 17:58:33 +02:00
typedef struct
{
vp8_sad_fn_t sad4x4;
vp8_sad_fn_t sad8x8;
vp8_sad_fn_t sad8x16;
vp8_sad_fn_t sad16x8;
vp8_sad_fn_t sad16x16;
vp8_variance_fn_t var4x4;
vp8_variance_fn_t var8x8;
vp8_variance_fn_t var8x16;
vp8_variance_fn_t var16x8;
vp8_variance_fn_t var16x16;
vp8_subpixvariance_fn_t subpixvar4x4;
vp8_subpixvariance_fn_t subpixvar8x8;
vp8_subpixvariance_fn_t subpixvar8x16;
vp8_subpixvariance_fn_t subpixvar16x8;
vp8_subpixvariance_fn_t subpixvar16x16;
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
vp8_variance_fn_t halfpixvar16x16_h;
vp8_variance_fn_t halfpixvar16x16_v;
vp8_variance_fn_t halfpixvar16x16_hv;
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8_subpixvariance_fn_t subpixmse16x16;
vp8_getmbss_fn_t getmbss;
vp8_variance_fn_t mse16x16;
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
vp8_get16x16prederror_fn_t get16x16prederror;
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8_variance2_fn_t get8x8var;
vp8_variance2_fn_t get16x16var;
Return sse value in vp8_variance SSE2 functions Minor modification. Change-Id: I09511d38fd1451d5c4106a48acdb3f766ce59cb7
2011-05-25 15:26:29 +02:00
vp8_get16x16prederror_fn_t get4x4sse_cs;
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8_sad_multi_fn_t sad16x16x3;
vp8_sad_multi_fn_t sad16x8x3;
vp8_sad_multi_fn_t sad8x16x3;
vp8_sad_multi_fn_t sad8x8x3;
vp8_sad_multi_fn_t sad4x4x3;
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
vp8_sad_multi1_fn_t sad16x16x8;
vp8_sad_multi1_fn_t sad16x8x8;
vp8_sad_multi1_fn_t sad8x16x8;
vp8_sad_multi1_fn_t sad8x8x8;
vp8_sad_multi1_fn_t sad4x4x8;
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8_sad_multi_d_fn_t sad16x16x4d;
vp8_sad_multi_d_fn_t sad16x8x4d;
vp8_sad_multi_d_fn_t sad8x16x4d;
vp8_sad_multi_d_fn_t sad8x8x4d;
vp8_sad_multi_d_fn_t sad4x4x4d;
changed configure option name to reduce confusion Renamed configure option "enable-psnr" to "enable-internal-stats" to better reflect the purpose of the option and eliminate the confusion reported in http://code.google.com/p/webm/issues/detail?id=35 Change-Id: If72df6fdb9f1e33dab1329240ba4d8911d2f1f7a
2011-04-29 18:37:59 +02:00
#if CONFIG_INTERNAL_STATS
vp8cx- alternate ssim function with optimizations Change-Id: I91921b0a90dbaddc7010380b038955be347964b3
2011-03-08 15:05:18 +01:00
vp8_ssimpf_fn_t ssimpf_8x8;
vp8_ssimpf_fn_t ssimpf;
#endif
Initial WebM release
2010-05-18 17:58:33 +02:00
} vp8_variance_rtcd_vtable_t;
typedef struct
{
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
vp8_sad_fn_t sdf;
vp8_variance_fn_t vf;
Initial WebM release
2010-05-18 17:58:33 +02:00
vp8_subpixvariance_fn_t svf;
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
vp8_variance_fn_t svf_halfpix_h;
vp8_variance_fn_t svf_halfpix_v;
vp8_variance_fn_t svf_halfpix_hv;
vp8_sad_multi_fn_t sdx3f;
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
vp8_sad_multi1_fn_t sdx8f;
Add half-pixel variance RTCD functions NEON has optimized 16x16 half-pixel variance functions, but they were not part of the RTCD framework. Add these functions to RTCD, so that other platforms can make use of this optimization in the future and special-case ARM code can be removed. A number of functions were taking two variance functions as parameters. These functions were changed to take a single parameter, a pointer to a struct containing all the variance functions for that block size. This provides additional flexibility for calling additional variance functions (the half-pixel special case, for example) and by initializing the table for all block sizes, we don't have to construct this function pointer table for each macroblock. Change-Id: I78289ff36b2715f9a7aa04d5f6fbe3d23acdc29c
2010-10-26 21:34:16 +02:00
vp8_sad_multi_d_fn_t sdx4df;
vp8cx- alternate ssim function with optimizations Change-Id: I91921b0a90dbaddc7010380b038955be347964b3
2011-03-08 15:05:18 +01:00
Initial WebM release
2010-05-18 17:58:33 +02:00
} vp8_variance_fn_ptr_t;
#if CONFIG_RUNTIME_CPU_DETECT
#define VARIANCE_INVOKE(ctx,fn) (ctx)->fn
#else
#define VARIANCE_INVOKE(ctx,fn) vp8_variance_##fn
#endif
#endif
Reference in New Issue Copy Permalink