Implement sse2 and ssse3 versions for all sub_pixel_variance sizes.

Overall speedup around 5% (bus @ 1500kbps first 50 frames 4min10 ->
3min58). Specific changes to timings for each function compared to
original assembly-optimized versions (or just new version timings if
no previous assembly-optimized version was available):

sse2   4x4:    99 ->   82 cycles
sse2   4x8:           128 cycles
sse2   8x4:           121 cycles
sse2   8x8:   149 ->  129 cycles
sse2   8x16:  235 ->  245 cycles (?)
sse2  16x8:   269 ->  203 cycles
sse2  16x16:  441 ->  349 cycles
sse2  16x32:          641 cycles
sse2  32x16:          643 cycles
sse2  32x32: 1733 -> 1154 cycles
sse2  32x64:         2247 cycles
sse2  64x32:         2323 cycles
sse2  64x64: 6984 -> 4442 cycles

ssse3  4x4:           100 cycles (?)
ssse3  4x8:           103 cycles
ssse3  8x4:            71 cycles
ssse3  8x8:           147 cycles
ssse3  8x16:          158 cycles
ssse3 16x8:   188 ->  162 cycles
ssse3 16x16:  316 ->  273 cycles
ssse3 16x32:          535 cycles
ssse3 32x16:          564 cycles
ssse3 32x32:          973 cycles
ssse3 32x64:         1930 cycles
ssse3 64x32:         1922 cycles
ssse3 64x64:         3760 cycles

Change-Id: I81ff6fe51daf35a40d19785167004664d7e0c59d

1 files changed, 19 insertions, 23 deletions

diff --git a/vp9/common/vp9_rtcd_defs.sh b/vp9/common/vp9_rtcd_defs.sh
index a405aab8d..575b619a1 100644
--- a/vp9/common/vp9_rtcd_defs.sh
+++ b/vp9/common/vp9_rtcd_defs.sh
@@ -266,85 +266,81 @@ prototype unsigned int vp9_variance4x4 "const uint8_t *src_ptr, int source_strid
 specialize vp9_variance4x4 mmx sse2
 
 prototype unsigned int vp9_sub_pixel_variance64x64 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance64x64 sse2
+specialize vp9_sub_pixel_variance64x64 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance64x64 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance64x64
 
 prototype unsigned int vp9_sub_pixel_variance32x64 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance32x64
+specialize vp9_sub_pixel_variance32x64 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance32x64 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance32x64
 
 prototype unsigned int vp9_sub_pixel_variance64x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance64x32
+specialize vp9_sub_pixel_variance64x32 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance64x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance64x32
 
 prototype unsigned int vp9_sub_pixel_variance32x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance32x16
+specialize vp9_sub_pixel_variance32x16 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance32x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance32x16
 
 prototype unsigned int vp9_sub_pixel_variance16x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance16x32
+specialize vp9_sub_pixel_variance16x32 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance16x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance16x32
 
 prototype unsigned int vp9_sub_pixel_variance32x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance32x32 sse2
+specialize vp9_sub_pixel_variance32x32 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance32x32 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance32x32
 
 prototype unsigned int vp9_sub_pixel_variance16x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance16x16 sse2 mmx ssse3
+specialize vp9_sub_pixel_variance16x16 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance16x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance16x16
 
 prototype unsigned int vp9_sub_pixel_variance8x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance8x16 sse2 mmx
-vp9_sub_pixel_variance8x16_sse2=vp9_sub_pixel_variance8x16_wmt
+specialize vp9_sub_pixel_variance8x16 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance8x16 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance8x16
 
 prototype unsigned int vp9_sub_pixel_variance16x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance16x8 sse2 mmx ssse3
-vp9_sub_pixel_variance16x8_sse2=vp9_sub_pixel_variance16x8_ssse3;
-vp9_sub_pixel_variance16x8_sse2=vp9_sub_pixel_variance16x8_wmt
+specialize vp9_sub_pixel_variance16x8 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance16x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance16x8
 
 prototype unsigned int vp9_sub_pixel_variance8x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance8x8 sse2 mmx
-vp9_sub_pixel_variance8x8_sse2=vp9_sub_pixel_variance8x8_wmt
+specialize vp9_sub_pixel_variance8x8 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance8x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance8x8
 
 # TODO(jingning): need to convert 8x4/4x8 functions into mmx/sse form
 prototype unsigned int vp9_sub_pixel_variance8x4 "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance8x4
+specialize vp9_sub_pixel_variance8x4 sse2 ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance8x4 "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance8x4
 
 prototype unsigned int vp9_sub_pixel_variance4x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance4x8
+specialize vp9_sub_pixel_variance4x8 sse ssse3
 
 prototype unsigned int vp9_sub_pixel_avg_variance4x8 "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance4x8
 
 prototype unsigned int vp9_sub_pixel_variance4x4 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_sub_pixel_variance4x4 sse2 mmx
-vp9_sub_pixel_variance4x4_sse2=vp9_sub_pixel_variance4x4_wmt
+specialize vp9_sub_pixel_variance4x4 sse ssse3
+#vp9_sub_pixel_variance4x4_sse2=vp9_sub_pixel_variance4x4_wmt
 
 prototype unsigned int vp9_sub_pixel_avg_variance4x4 "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"
 specialize vp9_sub_pixel_avg_variance4x4
@@ -390,15 +386,15 @@ prototype unsigned int vp9_sad4x4 "const uint8_t *src_ptr, int source_stride, co
 specialize vp9_sad4x4 mmx sse
 
 prototype unsigned int vp9_variance_halfpixvar16x16_h "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_variance_halfpixvar16x16_h mmx sse2
+specialize vp9_variance_halfpixvar16x16_h sse2
 vp9_variance_halfpixvar16x16_h_sse2=vp9_variance_halfpixvar16x16_h_wmt
 
 prototype unsigned int vp9_variance_halfpixvar16x16_v "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_variance_halfpixvar16x16_v mmx sse2
+specialize vp9_variance_halfpixvar16x16_v sse2
 vp9_variance_halfpixvar16x16_v_sse2=vp9_variance_halfpixvar16x16_v_wmt
 
 prototype unsigned int vp9_variance_halfpixvar16x16_hv "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
-specialize vp9_variance_halfpixvar16x16_hv mmx sse2
+specialize vp9_variance_halfpixvar16x16_hv sse2
 vp9_variance_halfpixvar16x16_hv_sse2=vp9_variance_halfpixvar16x16_hv_wmt
 
 prototype unsigned int vp9_variance_halfpixvar64x64_h "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"
@@ -507,8 +503,8 @@ specialize vp9_sad4x8x4d sse
 prototype void vp9_sad4x4x4d "const uint8_t *src_ptr, int  src_stride, const uint8_t* const ref_ptr[], int  ref_stride, unsigned int *sad_array"
 specialize vp9_sad4x4x4d sse
 
-prototype unsigned int vp9_sub_pixel_mse16x16 "const uint8_t *src_ptr, int  src_pixels_per_line, int  xoffset, int  yoffset, const uint8_t *dst_ptr, int dst_pixels_per_line, unsigned int *sse"
-specialize vp9_sub_pixel_mse16x16 sse2 mmx
+#prototype unsigned int vp9_sub_pixel_mse16x16 "const uint8_t *src_ptr, int  src_pixels_per_line, int  xoffset, int  yoffset, const uint8_t *dst_ptr, int dst_pixels_per_line, unsigned int *sse"
+#specialize vp9_sub_pixel_mse16x16 sse2 mmx
 
 prototype unsigned int vp9_mse16x16 "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse"
 specialize vp9_mse16x16 mmx sse2