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

Merge "vp9_subpel_variance_impl_intrin_avx2.c: make some tables static"

Browse Source
This commit is contained in:
James Zern 2014-02-20 16:03:23 -08:00 committed by Gerrit Code Review
commit e36cfc91a4
1 changed files with 27 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

53
vp9/encoder/x86/vp9_subpel_variance_impl_intrin_avx2.c
View File

@ -12,7 +12,7 @@
#include "vpx_ports/mem.h" #include "vpx_ports/mem.h"
#include "vp9/encoder/vp9_variance.h" #include "vp9/encoder/vp9_variance.h"
DECLARE_ALIGNED(32, const unsigned char, vp9_bilinear_filters_avx2[512])= { DECLARE_ALIGNED(32, static const uint8_t, bilinear_filters_avx2[512]) = {
16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0,
16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0, 16, 0,
15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1,
@ -44,7 +44,8 @@ DECLARE_ALIGNED(32, const unsigned char, vp9_bilinear_filters_avx2[512])= {
2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14,
2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14, 2, 14,
1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15,
1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15}; 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15, 1, 15
};
unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src, unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
int src_stride, int src_stride,
@ -137,12 +138,12 @@ unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
int64_t y_offset64; int64_t y_offset64;
y_offset64 = y_offset; y_offset64 = y_offset;
y_offset64 <<= 5; y_offset64 <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + y_offset64)); (__m256i const *)(bilinear_filters_avx2 + y_offset64));
#else #else
y_offset <<= 5; y_offset <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + y_offset)); (__m256i const *)(bilinear_filters_avx2 + y_offset));
#endif #endif
pw8 = _mm256_set1_epi16(8); pw8 = _mm256_set1_epi16(8);
for (i = 0; i < height ; i++) { for (i = 0; i < height ; i++) {
@ -283,12 +284,12 @@ unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
int64_t y_offset64; int64_t y_offset64;
y_offset64 = y_offset; y_offset64 = y_offset;
y_offset64 <<= 5; y_offset64 <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2+y_offset64)); (__m256i const *)(bilinear_filters_avx2 + y_offset64));
#else #else
y_offset <<= 5; y_offset <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + y_offset)); (__m256i const *)(bilinear_filters_avx2 + y_offset));
#endif #endif
pw8 = _mm256_set1_epi16(8); pw8 = _mm256_set1_epi16(8);
// load source and another source starting from the next // load source and another source starting from the next
@ -354,12 +355,12 @@ unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
int64_t x_offset64; int64_t x_offset64;
x_offset64 = x_offset; x_offset64 = x_offset;
x_offset64 <<= 5; x_offset64 <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2+x_offset64)); (__m256i const *)(bilinear_filters_avx2 + x_offset64));
#else #else
x_offset <<= 5; x_offset <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + x_offset)); (__m256i const *)(bilinear_filters_avx2 + x_offset));
#endif #endif
pw8 = _mm256_set1_epi16(8); pw8 = _mm256_set1_epi16(8);
for (i = 0; i < height ; i++) { for (i = 0; i < height ; i++) {
@ -413,12 +414,12 @@ unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
int64_t x_offset64; int64_t x_offset64;
x_offset64 = x_offset; x_offset64 = x_offset;
x_offset64 <<= 5; x_offset64 <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2+x_offset64)); (__m256i const *)(bilinear_filters_avx2 + x_offset64));
#else #else
x_offset <<= 5; x_offset <<= 5;
filter = _mm256_load_si256((__m256i const *) filter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + x_offset)); (__m256i const *)(bilinear_filters_avx2 + x_offset));
#endif #endif
pw8 = _mm256_set1_epi16(8); pw8 = _mm256_set1_epi16(8);
// load source and another source starting from the next // load source and another source starting from the next
@ -508,17 +509,17 @@ unsigned int vp9_sub_pixel_variance32xh_avx2(const uint8_t *src,
x_offset64 <<= 5; x_offset64 <<= 5;
y_offset64 = y_offset; y_offset64 = y_offset;
y_offset64 <<= 5; y_offset64 <<= 5;
xfilter = _mm256_load_si256((__m256i const *) xfilter = _mm256_load_si256(
(vp9_bilinear_filters_avx2+x_offset64)); (__m256i const *)(bilinear_filters_avx2 + x_offset64));
yfilter = _mm256_load_si256((__m256i const *) yfilter = _mm256_load_si256(
(vp9_bilinear_filters_avx2+y_offset64)); (__m256i const *)(bilinear_filters_avx2 + y_offset64));
#else #else
x_offset <<= 5; x_offset <<= 5;
xfilter = _mm256_load_si256((__m256i const *) xfilter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + x_offset)); (__m256i const *)(bilinear_filters_avx2 + x_offset));
y_offset <<= 5; y_offset <<= 5;
yfilter = _mm256_load_si256((__m256i const *) yfilter = _mm256_load_si256(
(vp9_bilinear_filters_avx2 + y_offset)); (__m256i const *)(bilinear_filters_avx2 + y_offset));
#endif #endif
pw8 = _mm256_set1_epi16(8); pw8 = _mm256_set1_epi16(8);
// load source and another source starting from the next // load source and another source starting from the next