specialize qw/vpx_highbd_convolve_avg sse2 avx2 neon/;
add_proto qw/void vpx_highbd_convolve8/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8 neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8 avx2 neon/, "$sse2_x86_64";
add_proto qw/void vpx_highbd_convolve8_horiz/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8_horiz neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8_horiz avx2 neon/, "$sse2_x86_64";
add_proto qw/void vpx_highbd_convolve8_vert/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8_vert neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8_vert avx2 neon/, "$sse2_x86_64";
add_proto qw/void vpx_highbd_convolve8_avg/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8_avg neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8_avg avx2 neon/, "$sse2_x86_64";
add_proto qw/void vpx_highbd_convolve8_avg_horiz/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8_avg_horiz neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8_avg_horiz avx2 neon/, "$sse2_x86_64";
add_proto qw/void vpx_highbd_convolve8_avg_vert/, "const uint16_t *src, ptrdiff_t src_stride, uint16_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, int bps";
- specialize qw/vpx_highbd_convolve8_avg_vert neon/, "$sse2_x86_64";
+ specialize qw/vpx_highbd_convolve8_avg_vert avx2 neon/, "$sse2_x86_64";
} # CONFIG_VP9_HIGHBITDEPTH
#
if (vpx_config("CONFIG_VP9_ENCODER") eq "yes") {
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
- specialize qw/vpx_fdct4x4 sse2/;
+ specialize qw/vpx_fdct4x4 neon sse2/;
add_proto qw/void vpx_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct4x4_1 sse2/;
add_proto qw/void vpx_highbd_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
} else {
add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
- specialize qw/vpx_fdct4x4 sse2 msa/;
+ specialize qw/vpx_fdct4x4 neon sse2 msa/;
add_proto qw/void vpx_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct4x4_1 sse2/;
if (vpx_config("CONFIG_VP9") eq "yes") {
add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int stride";
-
add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int stride";
if (vpx_config("CONFIG_EMULATE_HARDWARE") ne "yes") {
+ # Note that there are more specializations appended when CONFIG_VP9_HIGHBITDEPTH is off.
specialize qw/vpx_idct4x4_16_add neon sse2/;
-
specialize qw/vpx_idct4x4_1_add neon sse2/;
-
specialize qw/vpx_idct8x8_64_add neon sse2 ssse3/;
-
specialize qw/vpx_idct8x8_12_add neon sse2 ssse3/;
-
specialize qw/vpx_idct8x8_1_add neon sse2/;
-
specialize qw/vpx_idct16x16_256_add neon sse2/;
-
specialize qw/vpx_idct16x16_38_add neon sse2/;
$vpx_idct16x16_38_add_sse2=vpx_idct16x16_256_add_sse2;
-
specialize qw/vpx_idct16x16_10_add neon sse2/;
-
specialize qw/vpx_idct16x16_1_add neon sse2/;
-
specialize qw/vpx_idct32x32_1024_add neon sse2 ssse3/;
-
specialize qw/vpx_idct32x32_135_add neon sse2 ssse3/;
$vpx_idct32x32_135_add_sse2=vpx_idct32x32_1024_add_sse2;
-
specialize qw/vpx_idct32x32_34_add neon sse2 ssse3/;
-
specialize qw/vpx_idct32x32_1_add neon sse2/;
+
+ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") ne "yes") {
+ # Note that these specializations appends to the above ones.
+ specialize qw/vpx_idct4x4_16_add dspr2 msa/;
+ specialize qw/vpx_idct4x4_1_add dspr2 msa/;
+ specialize qw/vpx_idct8x8_64_add dspr2 msa/;
+ specialize qw/vpx_idct8x8_12_add dspr2 msa/;
+ specialize qw/vpx_idct8x8_1_add dspr2 msa/;
+ specialize qw/vpx_idct16x16_256_add dspr2 msa/;
+ specialize qw/vpx_idct16x16_38_add dspr2 msa/;
+ $vpx_idct16x16_38_add_dspr2=vpx_idct16x16_256_add_dspr2;
+ $vpx_idct16x16_38_add_msa=vpx_idct16x16_256_add_msa;
+ specialize qw/vpx_idct16x16_10_add dspr2 msa/;
+ specialize qw/vpx_idct16x16_1_add dspr2 msa/;
+ specialize qw/vpx_idct32x32_1024_add dspr2 msa/;
+ specialize qw/vpx_idct32x32_135_add dspr2 msa/;
+ $vpx_idct32x32_135_add_dspr2=vpx_idct32x32_1024_add_dspr2;
+ $vpx_idct32x32_135_add_msa=vpx_idct32x32_1024_add_msa;
+ specialize qw/vpx_idct32x32_34_add dspr2 msa/;
+ specialize qw/vpx_idct32x32_1_add dspr2 msa/;
+ specialize qw/vpx_iwht4x4_16_add msa sse2/;
+ specialize qw/vpx_iwht4x4_1_add msa/;
+ } # !CONFIG_VP9_HIGHBITDEPTH
} # !CONFIG_EMULATE_HARDWARE
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
# that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
specialize qw/vpx_iwht4x4_16_add sse2/;
- add_proto qw/void vpx_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
- specialize qw/vpx_highbd_idct4x4_1_add neon/;
-
- add_proto qw/void vpx_highbd_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
- specialize qw/vpx_highbd_idct8x8_1_add neon/;
-
- add_proto qw/void vpx_highbd_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct16x16_38_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
- specialize qw/vpx_highbd_idct16x16_1_add neon/;
-
- add_proto qw/void vpx_highbd_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct4x4_16_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct4x4_1_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ specialize qw/vpx_highbd_idct4x4_1_add neon sse2/;
+
+ add_proto qw/void vpx_highbd_idct8x8_64_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct8x8_12_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct8x8_1_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ specialize qw/vpx_highbd_idct8x8_1_add neon sse2/;
+
+ add_proto qw/void vpx_highbd_idct16x16_256_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct16x16_38_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct16x16_10_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct16x16_1_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ specialize qw/vpx_highbd_idct16x16_1_add neon sse2/;
+
+ add_proto qw/void vpx_highbd_idct32x32_1024_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct32x32_135_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct32x32_34_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_idct32x32_1_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
specialize qw/vpx_highbd_idct32x32_1_add neon sse2/;
- add_proto qw/void vpx_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
-
- add_proto qw/void vpx_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
+ add_proto qw/void vpx_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint16_t *dest, int stride, int bd";
if (vpx_config("CONFIG_EMULATE_HARDWARE") ne "yes") {
specialize qw/vpx_highbd_idct4x4_16_add neon sse2/;
-
specialize qw/vpx_highbd_idct8x8_64_add neon sse2/;
-
specialize qw/vpx_highbd_idct8x8_12_add neon sse2/;
-
specialize qw/vpx_highbd_idct16x16_256_add neon sse2/;
-
specialize qw/vpx_highbd_idct16x16_38_add neon sse2/;
$vpx_highbd_idct16x16_38_add_sse2=vpx_highbd_idct16x16_256_add_sse2;
-
specialize qw/vpx_highbd_idct16x16_10_add neon sse2/;
-
specialize qw/vpx_highbd_idct32x32_1024_add neon/;
-
specialize qw/vpx_highbd_idct32x32_135_add neon/;
-
specialize qw/vpx_highbd_idct32x32_34_add neon/;
} # !CONFIG_EMULATE_HARDWARE
-} else {
- if (vpx_config("CONFIG_EMULATE_HARDWARE") ne "yes") {
- specialize qw/vpx_idct4x4_16_add dspr2 msa/;
-
- specialize qw/vpx_idct4x4_1_add dspr2 msa/;
-
- specialize qw/vpx_idct8x8_64_add dspr2 msa/;
-
- specialize qw/vpx_idct8x8_12_add dspr2 msa/;
-
- specialize qw/vpx_idct8x8_1_add dspr2 msa/;
-
- specialize qw/vpx_idct16x16_256_add dspr2 msa/;
-
- specialize qw/vpx_idct16x16_38_add dspr2 msa/;
- $vpx_idct16x16_38_add_dspr2=vpx_idct16x16_256_add_dspr2;
- $vpx_idct16x16_38_add_msa=vpx_idct16x16_256_add_msa;
-
- specialize qw/vpx_idct16x16_10_add dspr2 msa/;
-
- specialize qw/vpx_idct16x16_1_add dspr2 msa/;
-
- specialize qw/vpx_idct32x32_1024_add dspr2 msa/;
-
- specialize qw/vpx_idct32x32_135_add dspr2 msa/;
- $vpx_idct32x32_135_add_dspr2=vpx_idct32x32_1024_add_dspr2;
- $vpx_idct32x32_135_add_msa=vpx_idct32x32_1024_add_msa;
-
- specialize qw/vpx_idct32x32_34_add dspr2 msa/;
-
- specialize qw/vpx_idct32x32_1_add dspr2 msa/;
-
- specialize qw/vpx_iwht4x4_16_add msa sse2/;
-
- specialize qw/vpx_iwht4x4_1_add msa/;
- } # !CONFIG_EMULATE_HARDWARE
} # CONFIG_VP9_HIGHBITDEPTH
} # CONFIG_VP9
# Single block SAD
#
add_proto qw/unsigned int vpx_sad64x64/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad64x64 avx2 neon msa sse2/;
+specialize qw/vpx_sad64x64 avx2 neon msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad64x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad64x32 avx2 msa sse2/;
+specialize qw/vpx_sad64x32 avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x64/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad32x64 avx2 msa sse2/;
+specialize qw/vpx_sad32x64 avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad32x32 avx2 neon msa sse2/;
+specialize qw/vpx_sad32x32 avx2 neon msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad32x16 avx2 msa sse2/;
+specialize qw/vpx_sad32x16 avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad16x32 msa sse2/;
+specialize qw/vpx_sad16x32 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad16x16 neon msa sse2/;
+specialize qw/vpx_sad16x16 neon msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-specialize qw/vpx_sad16x8 neon msa sse2/;
+specialize qw/vpx_sad16x8 neon msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad8x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
specialize qw/vpx_sad8x16 neon msa sse2/;
} # CONFIG_VP9_ENCODER
add_proto qw/unsigned int vpx_sad64x64_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad64x64_avg avx2 msa sse2/;
+specialize qw/vpx_sad64x64_avg avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad64x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad64x32_avg avx2 msa sse2/;
+specialize qw/vpx_sad64x32_avg avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x64_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad32x64_avg avx2 msa sse2/;
+specialize qw/vpx_sad32x64_avg avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad32x32_avg avx2 msa sse2/;
+specialize qw/vpx_sad32x32_avg avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad32x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad32x16_avg avx2 msa sse2/;
+specialize qw/vpx_sad32x16_avg avx2 msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad16x32_avg msa sse2/;
+specialize qw/vpx_sad16x32_avg msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad16x16_avg msa sse2/;
+specialize qw/vpx_sad16x16_avg msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad16x8_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-specialize qw/vpx_sad16x8_avg msa sse2/;
+specialize qw/vpx_sad16x8_avg msa sse2 vsx/;
add_proto qw/unsigned int vpx_sad8x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
specialize qw/vpx_sad8x16_avg msa sse2/;
# Multi-block SAD, comparing a reference to N independent blocks
#
add_proto qw/void vpx_sad64x64x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad64x64x4d avx2 neon msa sse2/;
+specialize qw/vpx_sad64x64x4d avx2 neon msa sse2 vsx/;
add_proto qw/void vpx_sad64x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad64x32x4d msa sse2/;
+specialize qw/vpx_sad64x32x4d msa sse2 vsx/;
add_proto qw/void vpx_sad32x64x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad32x64x4d msa sse2/;
+specialize qw/vpx_sad32x64x4d msa sse2 vsx/;
add_proto qw/void vpx_sad32x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad32x32x4d avx2 neon msa sse2/;
+specialize qw/vpx_sad32x32x4d avx2 neon msa sse2 vsx/;
add_proto qw/void vpx_sad32x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad32x16x4d msa sse2/;
+specialize qw/vpx_sad32x16x4d msa sse2 vsx/;
add_proto qw/void vpx_sad16x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad16x32x4d msa sse2/;
+specialize qw/vpx_sad16x32x4d msa sse2 vsx/;
add_proto qw/void vpx_sad16x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad16x16x4d neon msa sse2/;
+specialize qw/vpx_sad16x16x4d neon msa sse2 vsx/;
add_proto qw/void vpx_sad16x8x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
-specialize qw/vpx_sad16x8x4d msa sse2/;
+specialize qw/vpx_sad16x8x4d msa sse2 vsx/;
add_proto qw/void vpx_sad8x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
specialize qw/vpx_sad8x16x4d msa sse2/;
specialize qw/vpx_variance32x32 sse2 avx2 neon msa/;
add_proto qw/unsigned int vpx_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
- specialize qw/vpx_variance32x16 sse2 avx2 msa/;
+ specialize qw/vpx_variance32x16 sse2 avx2 neon msa/;
add_proto qw/unsigned int vpx_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
- specialize qw/vpx_variance16x32 sse2 msa/;
+ specialize qw/vpx_variance16x32 sse2 neon msa/;
add_proto qw/unsigned int vpx_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
specialize qw/vpx_variance16x16 sse2 avx2 neon msa/;
specialize qw/vpx_variance8x8 sse2 neon msa/;
add_proto qw/unsigned int vpx_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
- specialize qw/vpx_variance8x4 sse2 msa/;
+ specialize qw/vpx_variance8x4 sse2 neon msa/;
add_proto qw/unsigned int vpx_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
- specialize qw/vpx_variance4x8 sse2 msa/;
+ specialize qw/vpx_variance4x8 sse2 neon msa/;
add_proto qw/unsigned int vpx_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
- specialize qw/vpx_variance4x4 sse2 msa/;
+ specialize qw/vpx_variance4x4 sse2 neon msa/;
#
# Specialty Variance
specialize qw/vpx_mse8x8 sse2 msa/;
add_proto qw/unsigned int vpx_get_mb_ss/, "const int16_t *";
- specialize qw/vpx_get_mb_ss sse2 msa/;
+ specialize qw/vpx_get_mb_ss sse2 msa vsx/;
add_proto qw/unsigned int vpx_get4x4sse_cs/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride";
- specialize qw/vpx_get4x4sse_cs neon msa/;
+ specialize qw/vpx_get4x4sse_cs neon msa vsx/;
add_proto qw/void vpx_comp_avg_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride";
- specialize qw/vpx_comp_avg_pred sse2/;
+ specialize qw/vpx_comp_avg_pred neon sse2 vsx/;
#
# Subpixel Variance
specialize qw/vpx_sub_pixel_variance64x64 avx2 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance64x32 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance64x32 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance32x64 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance32x64 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/vpx_sub_pixel_variance32x32 avx2 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance32x16 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance32x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance16x32 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance16x32 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/vpx_sub_pixel_variance16x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance16x8 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance16x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance8x16 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance8x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
specialize qw/vpx_sub_pixel_variance8x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance8x4 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance8x4 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance4x8 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance4x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
- specialize qw/vpx_sub_pixel_variance4x4 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_variance4x4 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance64x64 avx2 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance64x64 neon avx2 msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance64x32 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance64x32 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance32x64 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance32x64 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance32x32 avx2 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance32x32 neon avx2 msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance32x16 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance32x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance16x32 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance16x32 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance16x16 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance16x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance16x8 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance16x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance8x16 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance8x16 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance8x8 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance8x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance8x4 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance8x4 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance4x8 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance4x8 neon msa sse2 ssse3/;
add_proto qw/uint32_t vpx_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
- specialize qw/vpx_sub_pixel_avg_variance4x4 msa sse2 ssse3/;
+ specialize qw/vpx_sub_pixel_avg_variance4x4 neon msa sse2 ssse3/;
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
add_proto qw/unsigned int vpx_highbd_12_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";