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

vp9: move avx2 convolve fns to intrinsics file

Browse Source 
+ synchronize filter function signatures

this makes any intrinsics filters available for inlining and has the
side-effect of making those filters static, quieting missing-prototype
warnings.

Change-Id: I1cd55c9d52547793ad65aa90c7620f0e426edaa2
This commit is contained in:
James Zern 2015-05-20 20:32:11 -07:00
parent ef2b3cce50
commit 2161e44025
2 changed files with 90 additions and 84 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
vp9/common/x86/vp9_asm_stubs.c
View File

@ -12,58 +12,6 @@
#include "./vpx_config.h"
#include "vp9/common/x86/convolve.h"
#if HAVE_AVX2 && HAVE_SSSE3
filter8_1dfunction vp9_filter_block1d16_v8_avx2;
filter8_1dfunction vp9_filter_block1d16_h8_avx2;
filter8_1dfunction vp9_filter_block1d4_v8_ssse3;
#if ARCH_X86_64
filter8_1dfunction vp9_filter_block1d8_v8_intrin_ssse3;
filter8_1dfunction vp9_filter_block1d8_h8_intrin_ssse3;
filter8_1dfunction vp9_filter_block1d4_h8_intrin_ssse3;
#define vp9_filter_block1d8_v8_avx2 vp9_filter_block1d8_v8_intrin_ssse3
#define vp9_filter_block1d8_h8_avx2 vp9_filter_block1d8_h8_intrin_ssse3
#define vp9_filter_block1d4_h8_avx2 vp9_filter_block1d4_h8_intrin_ssse3
#else // ARCH_X86
filter8_1dfunction vp9_filter_block1d8_v8_ssse3;
filter8_1dfunction vp9_filter_block1d8_h8_ssse3;
filter8_1dfunction vp9_filter_block1d4_h8_ssse3;
#define vp9_filter_block1d8_v8_avx2 vp9_filter_block1d8_v8_ssse3
#define vp9_filter_block1d8_h8_avx2 vp9_filter_block1d8_h8_ssse3
#define vp9_filter_block1d4_h8_avx2 vp9_filter_block1d4_h8_ssse3
#endif // ARCH_X86_64 / ARCH_X86
filter8_1dfunction vp9_filter_block1d16_v2_ssse3;
filter8_1dfunction vp9_filter_block1d16_h2_ssse3;
filter8_1dfunction vp9_filter_block1d8_v2_ssse3;
filter8_1dfunction vp9_filter_block1d8_h2_ssse3;
filter8_1dfunction vp9_filter_block1d4_v2_ssse3;
filter8_1dfunction vp9_filter_block1d4_h2_ssse3;
#define vp9_filter_block1d4_v8_avx2 vp9_filter_block1d4_v8_ssse3
#define vp9_filter_block1d16_v2_avx2 vp9_filter_block1d16_v2_ssse3
#define vp9_filter_block1d16_h2_avx2 vp9_filter_block1d16_h2_ssse3
#define vp9_filter_block1d8_v2_avx2 vp9_filter_block1d8_v2_ssse3
#define vp9_filter_block1d8_h2_avx2 vp9_filter_block1d8_h2_ssse3
#define vp9_filter_block1d4_v2_avx2 vp9_filter_block1d4_v2_ssse3
#define vp9_filter_block1d4_h2_avx2 vp9_filter_block1d4_h2_ssse3
// void vp9_convolve8_horiz_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
// void vp9_convolve8_vert_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
FUN_CONV_1D(horiz, x_step_q4, filter_x, h, src, , avx2);
FUN_CONV_1D(vert, y_step_q4, filter_y, v, src - src_stride * 3, , avx2);
// void vp9_convolve8_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
FUN_CONV_2D(, avx2);
#endif // HAVE_AX2 && HAVE_SSSE3
#if HAVE_SSSE3
#if ARCH_X86_64
filter8_1dfunction vp9_filter_block1d16_v8_intrin_ssse3;

122
vp9/common/x86/vp9_subpixel_8t_intrin_avx2.c
View File

@ -8,7 +8,14 @@
* be found in the AUTHORS file in the root of the source tree.
*/
// Due to a header conflict between math.h and intrinsics includes with ceil()
// in certain configurations under vs9 this include needs to precede
// immintrin.h.
#include "./vp9_rtcd.h"
#include <immintrin.h>
#include "vp9/common/x86/convolve.h"
#include "vpx_ports/mem.h"
// filters for 16_h8 and 16_v8
@ -53,23 +60,23 @@ DECLARE_ALIGNED(32, static const uint8_t, filt4_global_avx2[32]) = {
# define MM256_BROADCASTSI128_SI256(x) _mm256_broadcastsi128_si256(x)
#endif // __clang__
void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
unsigned int src_pixels_per_line,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int output_height,
int16_t *filter) {
static void vp9_filter_block1d16_h8_avx2(const uint8_t *src_ptr,
ptrdiff_t src_pixels_per_line,
uint8_t *output_ptr,
ptrdiff_t output_pitch,
uint32_t output_height,
const int16_t *filter) {
__m128i filtersReg;
__m256i addFilterReg64, filt1Reg, filt2Reg, filt3Reg, filt4Reg;
__m256i firstFilters, secondFilters, thirdFilters, forthFilters;
__m256i srcRegFilt32b1_1, srcRegFilt32b2_1, srcRegFilt32b2, srcRegFilt32b3;
__m256i srcReg32b1, srcReg32b2, filtersReg32;
unsigned int i;
unsigned int src_stride, dst_stride;
ptrdiff_t src_stride, dst_stride;
// create a register with 0,64,0,64,0,64,0,64,0,64,0,64,0,64,0,64
addFilterReg64 = _mm256_set1_epi32((int)0x0400040u);
filtersReg = _mm_loadu_si128((__m128i *)filter);
filtersReg = _mm_loadu_si128((const __m128i *)filter);
// converting the 16 bit (short) to 8 bit (byte) and have the same data
// in both lanes of 128 bit register.
filtersReg =_mm_packs_epi16(filtersReg, filtersReg);
@ -104,9 +111,9 @@ void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
for (i = output_height; i > 1; i-=2) {
// load the 2 strides of source
srcReg32b1 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr-3)));
_mm_loadu_si128((const __m128i *)(src_ptr - 3)));
srcReg32b1 = _mm256_inserti128_si256(srcReg32b1,
_mm_loadu_si128((__m128i *)
_mm_loadu_si128((const __m128i *)
(src_ptr+src_pixels_per_line-3)), 1);
// filter the source buffer
@ -135,9 +142,9 @@ void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
// reading 2 strides of the next 16 bytes
// (part of it was being read by earlier read)
srcReg32b2 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+5)));
_mm_loadu_si128((const __m128i *)(src_ptr + 5)));
srcReg32b2 = _mm256_inserti128_si256(srcReg32b2,
_mm_loadu_si128((__m128i *)
_mm_loadu_si128((const __m128i *)
(src_ptr+src_pixels_per_line+5)), 1);
// add and saturate the results together
@ -202,7 +209,7 @@ void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
__m128i srcReg1, srcReg2, srcRegFilt1_1, srcRegFilt2_1;
__m128i srcRegFilt2, srcRegFilt3;
srcReg1 = _mm_loadu_si128((__m128i *)(src_ptr-3));
srcReg1 = _mm_loadu_si128((const __m128i *)(src_ptr - 3));
// filter the source buffer
srcRegFilt1_1 = _mm_shuffle_epi8(srcReg1,
@ -237,7 +244,7 @@ void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
// reading the next 16 bytes
// (part of it was being read by earlier read)
srcReg2 = _mm_loadu_si128((__m128i *)(src_ptr+5));
srcReg2 = _mm_loadu_si128((const __m128i *)(src_ptr + 5));
// add and saturate the results together
srcRegFilt1_1 = _mm_adds_epi16(srcRegFilt1_1,
@ -297,12 +304,12 @@ void vp9_filter_block1d16_h8_avx2(unsigned char *src_ptr,
}
}
void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
unsigned int src_pitch,
unsigned char *output_ptr,
unsigned int out_pitch,
unsigned int output_height,
int16_t *filter) {
static void vp9_filter_block1d16_v8_avx2(const uint8_t *src_ptr,
ptrdiff_t src_pitch,
uint8_t *output_ptr,
ptrdiff_t out_pitch,
uint32_t output_height,
const int16_t *filter) {
__m128i filtersReg;
__m256i addFilterReg64;
__m256i srcReg32b1, srcReg32b2, srcReg32b3, srcReg32b4, srcReg32b5;
@ -310,11 +317,11 @@ void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
__m256i srcReg32b11, srcReg32b12, filtersReg32;
__m256i firstFilters, secondFilters, thirdFilters, forthFilters;
unsigned int i;
unsigned int src_stride, dst_stride;
ptrdiff_t src_stride, dst_stride;
// create a register with 0,64,0,64,0,64,0,64,0,64,0,64,0,64,0,64
addFilterReg64 = _mm256_set1_epi32((int)0x0400040u);
filtersReg = _mm_loadu_si128((__m128i *)filter);
filtersReg = _mm_loadu_si128((const __m128i *)filter);
// converting the 16 bit (short) to 8 bit (byte) and have the
// same data in both lanes of 128 bit register.
filtersReg =_mm_packs_epi16(filtersReg, filtersReg);
@ -344,19 +351,19 @@ void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
// load 16 bytes 7 times in stride of src_pitch
srcReg32b1 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr)));
_mm_loadu_si128((const __m128i *)(src_ptr)));
srcReg32b2 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch)));
srcReg32b3 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*2)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 2)));
srcReg32b4 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*3)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 3)));
srcReg32b5 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*4)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 4)));
srcReg32b6 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*5)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 5)));
srcReg32b7 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*6)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 6)));
// have each consecutive loads on the same 256 register
srcReg32b1 = _mm256_inserti128_si256(srcReg32b1,
@ -393,11 +400,11 @@ void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
// load the last 2 loads of 16 bytes and have every two
// consecutive loads in the same 256 bit register
srcReg32b8 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*7)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 7)));
srcReg32b7 = _mm256_inserti128_si256(srcReg32b7,
_mm256_castsi256_si128(srcReg32b8), 1);
srcReg32b9 = _mm256_castsi128_si256(
_mm_loadu_si128((__m128i *)(src_ptr+src_pitch*8)));
_mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 8)));
srcReg32b8 = _mm256_inserti128_si256(srcReg32b8,
_mm256_castsi256_si128(srcReg32b9), 1);
@ -476,7 +483,7 @@ void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
__m128i srcRegFilt1, srcRegFilt3, srcRegFilt4, srcRegFilt5;
__m128i srcRegFilt6, srcRegFilt7, srcRegFilt8;
// load the last 16 bytes
srcRegFilt8 = _mm_loadu_si128((__m128i *)(src_ptr+src_pitch*7));
srcRegFilt8 = _mm_loadu_si128((const __m128i *)(src_ptr + src_pitch * 7));
// merge the last 2 results together
srcRegFilt4 = _mm_unpacklo_epi8(
@ -542,3 +549,54 @@ void vp9_filter_block1d16_v8_avx2(unsigned char *src_ptr,
_mm_store_si128((__m128i*)output_ptr, srcRegFilt1);
}
}
#if HAVE_AVX2 && HAVE_SSSE3
filter8_1dfunction vp9_filter_block1d4_v8_ssse3;
#if ARCH_X86_64
filter8_1dfunction vp9_filter_block1d8_v8_intrin_ssse3;
filter8_1dfunction vp9_filter_block1d8_h8_intrin_ssse3;
filter8_1dfunction vp9_filter_block1d4_h8_intrin_ssse3;
#define vp9_filter_block1d8_v8_avx2 vp9_filter_block1d8_v8_intrin_ssse3
#define vp9_filter_block1d8_h8_avx2 vp9_filter_block1d8_h8_intrin_ssse3
#define vp9_filter_block1d4_h8_avx2 vp9_filter_block1d4_h8_intrin_ssse3
#else // ARCH_X86
filter8_1dfunction vp9_filter_block1d8_v8_ssse3;
filter8_1dfunction vp9_filter_block1d8_h8_ssse3;
filter8_1dfunction vp9_filter_block1d4_h8_ssse3;
#define vp9_filter_block1d8_v8_avx2 vp9_filter_block1d8_v8_ssse3
#define vp9_filter_block1d8_h8_avx2 vp9_filter_block1d8_h8_ssse3
#define vp9_filter_block1d4_h8_avx2 vp9_filter_block1d4_h8_ssse3
#endif // ARCH_X86_64
filter8_1dfunction vp9_filter_block1d16_v2_ssse3;
filter8_1dfunction vp9_filter_block1d16_h2_ssse3;
filter8_1dfunction vp9_filter_block1d8_v2_ssse3;
filter8_1dfunction vp9_filter_block1d8_h2_ssse3;
filter8_1dfunction vp9_filter_block1d4_v2_ssse3;
filter8_1dfunction vp9_filter_block1d4_h2_ssse3;
#define vp9_filter_block1d4_v8_avx2 vp9_filter_block1d4_v8_ssse3
#define vp9_filter_block1d16_v2_avx2 vp9_filter_block1d16_v2_ssse3
#define vp9_filter_block1d16_h2_avx2 vp9_filter_block1d16_h2_ssse3
#define vp9_filter_block1d8_v2_avx2 vp9_filter_block1d8_v2_ssse3
#define vp9_filter_block1d8_h2_avx2 vp9_filter_block1d8_h2_ssse3
#define vp9_filter_block1d4_v2_avx2 vp9_filter_block1d4_v2_ssse3
#define vp9_filter_block1d4_h2_avx2 vp9_filter_block1d4_h2_ssse3
// void vp9_convolve8_horiz_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
// void vp9_convolve8_vert_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
FUN_CONV_1D(horiz, x_step_q4, filter_x, h, src, , avx2);
FUN_CONV_1D(vert, y_step_q4, filter_y, v, src - src_stride * 3, , avx2);
// void vp9_convolve8_avx2(const uint8_t *src, ptrdiff_t src_stride,
// uint8_t *dst, ptrdiff_t dst_stride,
// const int16_t *filter_x, int x_step_q4,
// const int16_t *filter_y, int y_step_q4,
// int w, int h);
FUN_CONV_2D(, avx2);
#endif // HAVE_AX2 && HAVE_SSSE3