Commit a2288d27 authored by Parag Salasakar's avatar Parag Salasakar

mips msa vp9 intra-pred optimization

intra pred - average improvement ~2x-3x

Change-Id: Ie3f7d6eded5ecb7ed7ee506ba8e4d98f93803b09
parent 78b434e8
...@@ -211,6 +211,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred4, NULL, NULL, NULL, NULL, ...@@ -211,6 +211,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred4, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, vp9_tm_predictor_4x4_neon) NULL, NULL, NULL, NULL, vp9_tm_predictor_4x4_neon)
#endif // HAVE_NEON #endif // HAVE_NEON
#if HAVE_MSA
INTRA_PRED_TEST(MSA, TestIntraPred4, vp9_dc_predictor_4x4_msa,
vp9_dc_left_predictor_4x4_msa, vp9_dc_top_predictor_4x4_msa,
vp9_dc_128_predictor_4x4_msa, vp9_v_predictor_4x4_msa,
vp9_h_predictor_4x4_msa, NULL, NULL, NULL, NULL, NULL,
NULL, vp9_tm_predictor_4x4_msa)
#endif // HAVE_MSA
// ----------------------------------------------------------------------------- // -----------------------------------------------------------------------------
// 8x8 // 8x8
...@@ -256,6 +264,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred8, vp9_dc_predictor_8x8_neon, ...@@ -256,6 +264,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred8, vp9_dc_predictor_8x8_neon,
#endif // HAVE_NEON #endif // HAVE_NEON
#if HAVE_MSA
INTRA_PRED_TEST(MSA, TestIntraPred8, vp9_dc_predictor_8x8_msa,
vp9_dc_left_predictor_8x8_msa, vp9_dc_top_predictor_8x8_msa,
vp9_dc_128_predictor_8x8_msa, vp9_v_predictor_8x8_msa,
vp9_h_predictor_8x8_msa, NULL, NULL, NULL, NULL, NULL,
NULL, vp9_tm_predictor_8x8_msa)
#endif // HAVE_MSA
// ----------------------------------------------------------------------------- // -----------------------------------------------------------------------------
// 16x16 // 16x16
...@@ -299,6 +315,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred16, vp9_dc_predictor_16x16_neon, ...@@ -299,6 +315,14 @@ INTRA_PRED_TEST(NEON, TestIntraPred16, vp9_dc_predictor_16x16_neon,
vp9_tm_predictor_16x16_neon) vp9_tm_predictor_16x16_neon)
#endif // HAVE_NEON #endif // HAVE_NEON
#if HAVE_MSA
INTRA_PRED_TEST(MSA, TestIntraPred16, vp9_dc_predictor_16x16_msa,
vp9_dc_left_predictor_16x16_msa, vp9_dc_top_predictor_16x16_msa,
vp9_dc_128_predictor_16x16_msa, vp9_v_predictor_16x16_msa,
vp9_h_predictor_16x16_msa, NULL, NULL, NULL, NULL, NULL,
NULL, vp9_tm_predictor_16x16_msa)
#endif // HAVE_MSA
// ----------------------------------------------------------------------------- // -----------------------------------------------------------------------------
// 32x32 // 32x32
...@@ -337,4 +361,12 @@ INTRA_PRED_TEST(NEON, TestIntraPred32, NULL, NULL, NULL, NULL, ...@@ -337,4 +361,12 @@ INTRA_PRED_TEST(NEON, TestIntraPred32, NULL, NULL, NULL, NULL,
NULL, NULL, NULL, NULL, NULL, vp9_tm_predictor_32x32_neon) NULL, NULL, NULL, NULL, NULL, vp9_tm_predictor_32x32_neon)
#endif // HAVE_NEON #endif // HAVE_NEON
#if HAVE_MSA
INTRA_PRED_TEST(MSA, TestIntraPred32, vp9_dc_predictor_32x32_msa,
vp9_dc_left_predictor_32x32_msa, vp9_dc_top_predictor_32x32_msa,
vp9_dc_128_predictor_32x32_msa, vp9_v_predictor_32x32_msa,
vp9_h_predictor_32x32_msa, NULL, NULL, NULL, NULL, NULL,
NULL, vp9_tm_predictor_32x32_msa)
#endif // HAVE_MSA
#include "test/test_libvpx.cc" #include "test/test_libvpx.cc"
This diff is collapsed.
...@@ -743,6 +743,26 @@ ...@@ -743,6 +743,26 @@
CLIP_SH2_0_255(in2, in3); \ CLIP_SH2_0_255(in2, in3); \
} }
/* Description : Horizontal addition of unsigned byte vector elements
Arguments : Inputs - in0, in1
Outputs - out0, out1
Return Type - as per RTYPE
Details : Each unsigned odd byte element from 'in0' is added to
even unsigned byte element from 'in0' (pairwise) and the
halfword result is stored in 'out0'
*/
#define HADD_UB2(RTYPE, in0, in1, out0, out1) { \
out0 = (RTYPE)__msa_hadd_u_h((v16u8)in0, (v16u8)in0); \
out1 = (RTYPE)__msa_hadd_u_h((v16u8)in1, (v16u8)in1); \
}
#define HADD_UB2_UH(...) HADD_UB2(v8u16, __VA_ARGS__)
#define HADD_UB4(RTYPE, in0, in1, in2, in3, out0, out1, out2, out3) { \
HADD_UB2(RTYPE, in0, in1, out0, out1); \
HADD_UB2(RTYPE, in2, in3, out2, out3); \
}
#define HADD_UB4_UH(...) HADD_UB4(v8u16, __VA_ARGS__)
/* Description : Insert specified word elements from input vectors to 1 /* Description : Insert specified word elements from input vectors to 1
destination vector destination vector
Arguments : Inputs - in0, in1, in2, in3 (4 input vectors) Arguments : Inputs - in0, in1, in2, in3 (4 input vectors)
...@@ -755,6 +775,19 @@ ...@@ -755,6 +775,19 @@
} }
#define INSERT_W2_SB(...) INSERT_W2(v16i8, __VA_ARGS__) #define INSERT_W2_SB(...) INSERT_W2(v16i8, __VA_ARGS__)
/* Description : Insert specified double word elements from input vectors to 1
destination vector
Arguments : Inputs - in0, in1 (2 input vectors)
Outputs - out (output vector)
Return Type - as per RTYPE
*/
#define INSERT_D2(RTYPE, in0, in1, out) { \
out = (RTYPE)__msa_insert_d((v2i64)out, 0, in0); \
out = (RTYPE)__msa_insert_d((v2i64)out, 1, in1); \
}
#define INSERT_D2_UB(...) INSERT_D2(v16u8, __VA_ARGS__)
#define INSERT_D2_SB(...) INSERT_D2(v16i8, __VA_ARGS__)
/* Description : Interleave even byte elements from vectors /* Description : Interleave even byte elements from vectors
Arguments : Inputs - in0, in1, in2, in3 Arguments : Inputs - in0, in1, in2, in3
Outputs - out0, out1 Outputs - out0, out1
......
...@@ -66,7 +66,7 @@ add_proto qw/void vp9_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, con ...@@ -66,7 +66,7 @@ add_proto qw/void vp9_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, con
specialize qw/vp9_d63_predictor_4x4/, "$ssse3_x86inc"; specialize qw/vp9_d63_predictor_4x4/, "$ssse3_x86inc";
add_proto qw/void vp9_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_h_predictor_4x4 neon dspr2/, "$ssse3_x86inc"; specialize qw/vp9_h_predictor_4x4 neon dspr2 msa/, "$ssse3_x86inc";
add_proto qw/void vp9_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d117_predictor_4x4/; specialize qw/vp9_d117_predictor_4x4/;
...@@ -78,22 +78,22 @@ add_proto qw/void vp9_d153_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, co ...@@ -78,22 +78,22 @@ add_proto qw/void vp9_d153_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, co
specialize qw/vp9_d153_predictor_4x4/, "$ssse3_x86inc"; specialize qw/vp9_d153_predictor_4x4/, "$ssse3_x86inc";
add_proto qw/void vp9_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_v_predictor_4x4 neon/, "$sse_x86inc"; specialize qw/vp9_v_predictor_4x4 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_tm_predictor_4x4 neon dspr2/, "$sse_x86inc"; specialize qw/vp9_tm_predictor_4x4 neon dspr2 msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_predictor_4x4 dspr2/, "$sse_x86inc"; specialize qw/vp9_dc_predictor_4x4 dspr2 msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_top_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_top_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_top_predictor_4x4/, "$sse_x86inc"; specialize qw/vp9_dc_top_predictor_4x4 msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_left_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_left_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_left_predictor_4x4/, "$sse_x86inc"; specialize qw/vp9_dc_left_predictor_4x4 msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_128_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_128_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_128_predictor_4x4/, "$sse_x86inc"; specialize qw/vp9_dc_128_predictor_4x4 msa/, "$sse_x86inc";
add_proto qw/void vp9_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d207_predictor_8x8/, "$ssse3_x86inc"; specialize qw/vp9_d207_predictor_8x8/, "$ssse3_x86inc";
...@@ -105,7 +105,7 @@ add_proto qw/void vp9_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, con ...@@ -105,7 +105,7 @@ add_proto qw/void vp9_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, con
specialize qw/vp9_d63_predictor_8x8/, "$ssse3_x86inc"; specialize qw/vp9_d63_predictor_8x8/, "$ssse3_x86inc";
add_proto qw/void vp9_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_h_predictor_8x8 neon dspr2/, "$ssse3_x86inc"; specialize qw/vp9_h_predictor_8x8 neon dspr2 msa/, "$ssse3_x86inc";
add_proto qw/void vp9_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d117_predictor_8x8/; specialize qw/vp9_d117_predictor_8x8/;
...@@ -117,22 +117,22 @@ add_proto qw/void vp9_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, co ...@@ -117,22 +117,22 @@ add_proto qw/void vp9_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, co
specialize qw/vp9_d153_predictor_8x8/, "$ssse3_x86inc"; specialize qw/vp9_d153_predictor_8x8/, "$ssse3_x86inc";
add_proto qw/void vp9_v_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_v_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_v_predictor_8x8 neon/, "$sse_x86inc"; specialize qw/vp9_v_predictor_8x8 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_tm_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_tm_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_tm_predictor_8x8 neon dspr2/, "$sse2_x86inc"; specialize qw/vp9_tm_predictor_8x8 neon dspr2 msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_predictor_8x8 dspr2 neon/, "$sse_x86inc"; specialize qw/vp9_dc_predictor_8x8 dspr2 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_top_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_top_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_top_predictor_8x8 neon/, "$sse_x86inc"; specialize qw/vp9_dc_top_predictor_8x8 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_left_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_left_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_left_predictor_8x8 neon/, "$sse_x86inc"; specialize qw/vp9_dc_left_predictor_8x8 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_dc_128_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_128_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_128_predictor_8x8 neon/, "$sse_x86inc"; specialize qw/vp9_dc_128_predictor_8x8 neon msa/, "$sse_x86inc";
add_proto qw/void vp9_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d207_predictor_16x16/, "$ssse3_x86inc"; specialize qw/vp9_d207_predictor_16x16/, "$ssse3_x86inc";
...@@ -144,7 +144,7 @@ add_proto qw/void vp9_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, c ...@@ -144,7 +144,7 @@ add_proto qw/void vp9_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, c
specialize qw/vp9_d63_predictor_16x16/, "$ssse3_x86inc"; specialize qw/vp9_d63_predictor_16x16/, "$ssse3_x86inc";
add_proto qw/void vp9_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_h_predictor_16x16 neon dspr2/, "$ssse3_x86inc"; specialize qw/vp9_h_predictor_16x16 neon dspr2 msa/, "$ssse3_x86inc";
add_proto qw/void vp9_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d117_predictor_16x16/; specialize qw/vp9_d117_predictor_16x16/;
...@@ -156,22 +156,22 @@ add_proto qw/void vp9_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, ...@@ -156,22 +156,22 @@ add_proto qw/void vp9_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride,
specialize qw/vp9_d153_predictor_16x16/, "$ssse3_x86inc"; specialize qw/vp9_d153_predictor_16x16/, "$ssse3_x86inc";
add_proto qw/void vp9_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_v_predictor_16x16 neon/, "$sse2_x86inc"; specialize qw/vp9_v_predictor_16x16 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_tm_predictor_16x16 neon/, "$sse2_x86inc"; specialize qw/vp9_tm_predictor_16x16 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_predictor_16x16 dspr2 neon/, "$sse2_x86inc"; specialize qw/vp9_dc_predictor_16x16 dspr2 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_top_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_top_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_top_predictor_16x16 neon/, "$sse2_x86inc"; specialize qw/vp9_dc_top_predictor_16x16 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_left_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_left_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_left_predictor_16x16 neon/, "$sse2_x86inc"; specialize qw/vp9_dc_left_predictor_16x16 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_128_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_128_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_128_predictor_16x16 neon/, "$sse2_x86inc"; specialize qw/vp9_dc_128_predictor_16x16 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d207_predictor_32x32/, "$ssse3_x86inc"; specialize qw/vp9_d207_predictor_32x32/, "$ssse3_x86inc";
...@@ -183,7 +183,7 @@ add_proto qw/void vp9_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, c ...@@ -183,7 +183,7 @@ add_proto qw/void vp9_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, c
specialize qw/vp9_d63_predictor_32x32/, "$ssse3_x86inc"; specialize qw/vp9_d63_predictor_32x32/, "$ssse3_x86inc";
add_proto qw/void vp9_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_h_predictor_32x32 neon/, "$ssse3_x86inc"; specialize qw/vp9_h_predictor_32x32 neon msa/, "$ssse3_x86inc";
add_proto qw/void vp9_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_d117_predictor_32x32/; specialize qw/vp9_d117_predictor_32x32/;
...@@ -195,22 +195,22 @@ add_proto qw/void vp9_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, ...@@ -195,22 +195,22 @@ add_proto qw/void vp9_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride,
specialize qw/vp9_d153_predictor_32x32/; specialize qw/vp9_d153_predictor_32x32/;
add_proto qw/void vp9_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_v_predictor_32x32 neon/, "$sse2_x86inc"; specialize qw/vp9_v_predictor_32x32 neon msa/, "$sse2_x86inc";
add_proto qw/void vp9_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_tm_predictor_32x32 neon/, "$sse2_x86_64"; specialize qw/vp9_tm_predictor_32x32 neon msa/, "$sse2_x86_64";
add_proto qw/void vp9_dc_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_predictor_32x32/, "$sse2_x86inc"; specialize qw/vp9_dc_predictor_32x32 msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_top_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_top_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_top_predictor_32x32/, "$sse2_x86inc"; specialize qw/vp9_dc_top_predictor_32x32 msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_left_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_left_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_left_predictor_32x32/, "$sse2_x86inc"; specialize qw/vp9_dc_left_predictor_32x32 msa/, "$sse2_x86inc";
add_proto qw/void vp9_dc_128_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; add_proto qw/void vp9_dc_128_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vp9_dc_128_predictor_32x32/, "$sse2_x86inc"; specialize qw/vp9_dc_128_predictor_32x32 msa/, "$sse2_x86inc";
# #
# Loopfilter # Loopfilter
......
...@@ -146,6 +146,7 @@ VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct8x8_msa.c ...@@ -146,6 +146,7 @@ VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct8x8_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct16x16_msa.c VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct16x16_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct32x32_msa.c VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct32x32_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct_msa.h VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_idct_msa.h
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_intra_predict_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_4_msa.c VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_4_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_8_msa.c VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_8_msa.c
VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_16_msa.c VP9_COMMON_SRCS-$(HAVE_MSA) += common/mips/msa/vp9_loopfilter_16_msa.c
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment