diff --git a/test/fdct8x8_test.cc b/test/fdct8x8_test.cc index 7edb4d0420413b63ee2afcc707c367a923941175..d5f4793bd5bff8def45e8f5e6425b2236f26c0d3 100644 --- a/test/fdct8x8_test.cc +++ b/test/fdct8x8_test.cc @@ -21,7 +21,7 @@ extern "C" { #include "vp9/common/vp9_entropy.h" #include "./vp9_rtcd.h" -void vp9_short_idct8x8_add_c(int16_t *input, uint8_t *output, int pitch); +void vp9_idct8x8_64_add_c(int16_t *input, uint8_t *output, int pitch); } #include "vpx/vpx_integer.h" @@ -296,7 +296,7 @@ using std::tr1::make_tuple; INSTANTIATE_TEST_CASE_P( C, FwdTrans8x8DCT, ::testing::Values( - make_tuple(&vp9_short_fdct8x8_c, &vp9_short_idct8x8_add_c, 0))); + make_tuple(&vp9_short_fdct8x8_c, &vp9_idct8x8_64_add_c, 0))); INSTANTIATE_TEST_CASE_P( C, FwdTrans8x8HT, ::testing::Values( @@ -309,7 +309,7 @@ INSTANTIATE_TEST_CASE_P( INSTANTIATE_TEST_CASE_P( SSE2, FwdTrans8x8DCT, ::testing::Values( - make_tuple(&vp9_short_fdct8x8_sse2, &vp9_short_idct8x8_add_sse2, 0))); + make_tuple(&vp9_short_fdct8x8_sse2, &vp9_idct8x8_64_add_sse2, 0))); INSTANTIATE_TEST_CASE_P( SSE2, FwdTrans8x8HT, ::testing::Values( diff --git a/test/idct8x8_test.cc b/test/idct8x8_test.cc index fc8129e04a780ad388238dc3a340be75341e767b..d8c61ffb2c554df860872e4ddd5c01a7dde8cb63 100644 --- a/test/idct8x8_test.cc +++ b/test/idct8x8_test.cc @@ -126,7 +126,7 @@ TEST(VP9Idct8x8Test, AccuracyCheck) { reference_dct_2d(input, output_r); for (int j = 0; j < 64; ++j) coeff[j] = round(output_r[j]); - vp9_short_idct8x8_add_c(coeff, dst, 8); + vp9_idct8x8_64_add_c(coeff, dst, 8); for (int j = 0; j < 64; ++j) { const int diff = dst[j] - src[j]; const int error = diff * diff; diff --git a/vp9/common/arm/neon/vp9_short_idct8x8_1_add_neon.asm b/vp9/common/arm/neon/vp9_short_idct8x8_1_add_neon.asm index 923804f90a5d1437d6f7eec4f951f44526a3f61b..421d202d403ead10b672091e09c95db13e58b760 100644 --- a/vp9/common/arm/neon/vp9_short_idct8x8_1_add_neon.asm +++ b/vp9/common/arm/neon/vp9_short_idct8x8_1_add_neon.asm @@ -8,21 +8,21 @@ ; - EXPORT |vp9_short_idct8x8_1_add_neon| + EXPORT |vp9_idct8x8_1_add_neon| ARM REQUIRE8 PRESERVE8 AREA ||.text||, CODE, READONLY, ALIGN=2 -;void vp9_short_idct8x8_1_add_neon(int16_t *input, uint8_t *dest, +;void vp9_idct8x8_1_add_neon(int16_t *input, uint8_t *dest, ; int dest_stride) ; ; r0 int16_t input ; r1 uint8_t *dest ; r2 int dest_stride) -|vp9_short_idct8x8_1_add_neon| PROC +|vp9_idct8x8_1_add_neon| PROC ldrsh r0, [r0] ; generate cospi_16_64 = 11585 @@ -83,6 +83,6 @@ vst1.64 {d31}, [r12], r2 bx lr - ENDP ; |vp9_short_idct8x8_1_add_neon| + ENDP ; |vp9_idct8x8_1_add_neon| END diff --git a/vp9/common/arm/neon/vp9_short_idct8x8_add_neon.asm b/vp9/common/arm/neon/vp9_short_idct8x8_add_neon.asm index c02251a3dab49f9582bb9380d5946cfeedbfa588..54764008bdf853eb716e10e91194f3a674876b7c 100644 --- a/vp9/common/arm/neon/vp9_short_idct8x8_add_neon.asm +++ b/vp9/common/arm/neon/vp9_short_idct8x8_add_neon.asm @@ -8,8 +8,8 @@ ; be found in the AUTHORS file in the root of the source tree. ; - EXPORT |vp9_short_idct8x8_add_neon| - EXPORT |vp9_short_idct8x8_10_add_neon| + EXPORT |vp9_idct8x8_64_add_neon| + EXPORT |vp9_idct8x8_10_add_neon| ARM REQUIRE8 PRESERVE8 @@ -198,13 +198,13 @@ MEND AREA Block, CODE, READONLY ; name this block of code -;void vp9_short_idct8x8_add_neon(int16_t *input, uint8_t *dest, int dest_stride) +;void vp9_idct8x8_64_add_neon(int16_t *input, uint8_t *dest, int dest_stride) ; ; r0 int16_t input ; r1 uint8_t *dest ; r2 int dest_stride) -|vp9_short_idct8x8_add_neon| PROC +|vp9_idct8x8_64_add_neon| PROC push {r4-r9} vpush {d8-d15} vld1.s16 {q8,q9}, [r0]! @@ -308,15 +308,15 @@ vpop {d8-d15} pop {r4-r9} bx lr - ENDP ; |vp9_short_idct8x8_add_neon| + ENDP ; |vp9_idct8x8_64_add_neon| -;void vp9_short_idct8x8_10_add_neon(int16_t *input, uint8_t *dest, int dest_stride) +;void vp9_idct8x8_10_add_neon(int16_t *input, uint8_t *dest, int dest_stride) ; ; r0 int16_t input ; r1 uint8_t *dest ; r2 int dest_stride) -|vp9_short_idct8x8_10_add_neon| PROC +|vp9_idct8x8_10_add_neon| PROC push {r4-r9} vpush {d8-d15} vld1.s16 {q8,q9}, [r0]! @@ -514,6 +514,6 @@ vpop {d8-d15} pop {r4-r9} bx lr - ENDP ; |vp9_short_idct8x8_10_add_neon| + ENDP ; |vp9_idct8x8_10_add_neon| END diff --git a/vp9/common/vp9_idct.c b/vp9/common/vp9_idct.c index 46363700764d4b5b0231d9cd4570dcbec7604c82..dea923724fed503e9fda8da294aaa3d32edbef9c 100644 --- a/vp9/common/vp9_idct.c +++ b/vp9/common/vp9_idct.c @@ -201,7 +201,7 @@ static void idct8_1d(int16_t *input, int16_t *output) { output[7] = step1[0] - step1[7]; } -void vp9_short_idct8x8_add_c(int16_t *input, uint8_t *dest, int dest_stride) { +void vp9_idct8x8_64_add_c(int16_t *input, uint8_t *dest, int dest_stride) { int16_t out[8 * 8]; int16_t *outptr = out; int i, j; @@ -225,7 +225,7 @@ void vp9_short_idct8x8_add_c(int16_t *input, uint8_t *dest, int dest_stride) { } } -void vp9_short_idct8x8_1_add_c(int16_t *input, uint8_t *dest, int dest_stride) { +void vp9_idct8x8_1_add_c(int16_t *input, uint8_t *dest, int dest_stride) { int i, j; int a1; int16_t out = dct_const_round_shift(input[0] * cospi_16_64); @@ -420,7 +420,7 @@ void vp9_short_iht8x8_add_c(int16_t *input, uint8_t *dest, int dest_stride, + dest[j * dest_stride + i]); } } -void vp9_short_idct8x8_10_add_c(int16_t *input, uint8_t *dest, +void vp9_idct8x8_10_add_c(int16_t *input, uint8_t *dest, int dest_stride) { int16_t out[8 * 8] = { 0 }; int16_t *outptr = out; @@ -1301,7 +1301,7 @@ void vp9_iwht4x4_add(int16_t *input, uint8_t *dest, int stride, int eob) { vp9_iwht4x4_1_add(input, dest, stride); } -void vp9_idct_add_8x8(int16_t *input, uint8_t *dest, int stride, int eob) { +void vp9_idct8x8_add(int16_t *input, uint8_t *dest, int stride, int eob) { // If dc is 1, then input[0] is the reconstructed value, do not need // dequantization. Also, when dc is 1, dc is counted in eobs, namely eobs >=1. @@ -1312,11 +1312,11 @@ void vp9_idct_add_8x8(int16_t *input, uint8_t *dest, int stride, int eob) { if (eob) { if (eob == 1) // DC only DCT coefficient - vp9_short_idct8x8_1_add(input, dest, stride); + vp9_idct8x8_1_add(input, dest, stride); else if (eob <= 10) - vp9_short_idct8x8_10_add(input, dest, stride); + vp9_idct8x8_10_add(input, dest, stride); else - vp9_short_idct8x8_add(input, dest, stride); + vp9_idct8x8_64_add(input, dest, stride); } } @@ -1355,7 +1355,7 @@ void vp9_iht_add(TX_TYPE tx_type, int16_t *input, uint8_t *dest, int stride, void vp9_iht_add_8x8(TX_TYPE tx_type, int16_t *input, uint8_t *dest, int stride, int eob) { if (tx_type == DCT_DCT) { - vp9_idct_add_8x8(input, dest, stride, eob); + vp9_idct8x8_add(input, dest, stride, eob); } else { if (eob > 0) { vp9_short_iht8x8_add(input, dest, stride, tx_type); diff --git a/vp9/common/vp9_idct.h b/vp9/common/vp9_idct.h index a15b6d36f4032cad57e03557c4f4d25c06052b7d..41519ce44cbc1fa84f2f0e9b16719a94ea2034a8 100644 --- a/vp9/common/vp9_idct.h +++ b/vp9/common/vp9_idct.h @@ -90,7 +90,7 @@ typedef struct { void vp9_idct4x4_add(int16_t *input, uint8_t *dest, int stride, int eob); void vp9_iwht4x4_add(int16_t *input, uint8_t *dest, int stride, int eob); -void vp9_idct_add_8x8(int16_t *input, uint8_t *dest, int stride, int eob); +void vp9_idct8x8_add(int16_t *input, uint8_t *dest, int stride, int eob); void vp9_idct_add_16x16(int16_t *input, uint8_t *dest, int stride, int eob); void vp9_idct_add_32x32(int16_t *input, uint8_t *dest, int stride, int eob); diff --git a/vp9/common/vp9_rtcd_defs.sh b/vp9/common/vp9_rtcd_defs.sh index 67dced2102162fdc98e9328785adefe4eadba273..244064f60d43d732077e6da97bfb6e42686351b8 100644 --- a/vp9/common/vp9_rtcd_defs.sh +++ b/vp9/common/vp9_rtcd_defs.sh @@ -273,14 +273,14 @@ specialize vp9_idct4x4_1_add sse2 neon prototype void vp9_idct4x4_16_add "int16_t *input, uint8_t *dest, int dest_stride" specialize vp9_idct4x4_16_add sse2 neon -prototype void vp9_short_idct8x8_1_add "int16_t *input, uint8_t *dest, int dest_stride" -specialize vp9_short_idct8x8_1_add sse2 neon +prototype void vp9_idct8x8_1_add "int16_t *input, uint8_t *dest, int dest_stride" +specialize vp9_idct8x8_1_add sse2 neon -prototype void vp9_short_idct8x8_add "int16_t *input, uint8_t *dest, int dest_stride" -specialize vp9_short_idct8x8_add sse2 neon +prototype void vp9_idct8x8_64_add "int16_t *input, uint8_t *dest, int dest_stride" +specialize vp9_idct8x8_64_add sse2 neon -prototype void vp9_short_idct8x8_10_add "int16_t *input, uint8_t *dest, int dest_stride" -specialize vp9_short_idct8x8_10_add sse2 neon +prototype void vp9_idct8x8_10_add "int16_t *input, uint8_t *dest, int dest_stride" +specialize vp9_idct8x8_10_add sse2 neon prototype void vp9_short_idct16x16_1_add "int16_t *input, uint8_t *dest, int dest_stride" specialize vp9_short_idct16x16_1_add sse2 neon diff --git a/vp9/common/x86/vp9_idct_intrin_sse2.c b/vp9/common/x86/vp9_idct_intrin_sse2.c index f03af33782e434a1ba991ccc5f2a52e0be1ce5a2..84338051b0f040217e39d5db9a73e20c5df5f8c3 100644 --- a/vp9/common/x86/vp9_idct_intrin_sse2.c +++ b/vp9/common/x86/vp9_idct_intrin_sse2.c @@ -529,7 +529,7 @@ void vp9_short_iht4x4_add_sse2(int16_t *input, uint8_t *dest, int stride, dest += stride; \ } -void vp9_short_idct8x8_add_sse2(int16_t *input, uint8_t *dest, int stride) { +void vp9_idct8x8_64_add_sse2(int16_t *input, uint8_t *dest, int stride) { const __m128i zero = _mm_setzero_si128(); const __m128i rounding = _mm_set1_epi32(DCT_CONST_ROUNDING); const __m128i final_rounding = _mm_set1_epi16(1<<4); @@ -597,7 +597,7 @@ void vp9_short_idct8x8_add_sse2(int16_t *input, uint8_t *dest, int stride) { RECON_AND_STORE(dest, in7); } -void vp9_short_idct8x8_1_add_sse2(int16_t *input, uint8_t *dest, int stride) { +void vp9_idct8x8_1_add_sse2(int16_t *input, uint8_t *dest, int stride) { __m128i dc_value; const __m128i zero = _mm_setzero_si128(); int a; @@ -985,7 +985,7 @@ void vp9_short_iht8x8_add_sse2(int16_t *input, uint8_t *dest, int stride, RECON_AND_STORE(dest, in[7]); } -void vp9_short_idct8x8_10_add_sse2(int16_t *input, uint8_t *dest, int stride) { +void vp9_idct8x8_10_add_sse2(int16_t *input, uint8_t *dest, int stride) { const __m128i zero = _mm_setzero_si128(); const __m128i rounding = _mm_set1_epi32(DCT_CONST_ROUNDING); const __m128i final_rounding = _mm_set1_epi16(1<<4); diff --git a/vp9/encoder/vp9_encodemb.c b/vp9/encoder/vp9_encodemb.c index 1f36f161865f4ffbe2594eb476ca39e85a6b81b8..994d0c8106c1859fae5503cb0392173d76230747 100644 --- a/vp9/encoder/vp9_encodemb.c +++ b/vp9/encoder/vp9_encodemb.c @@ -460,7 +460,7 @@ static void encode_block(int plane, int block, BLOCK_SIZE plane_bsize, vp9_idct_add_16x16(dqcoeff, dst, pd->dst.stride, pd->eobs[block]); break; case TX_8X8: - vp9_idct_add_8x8(dqcoeff, dst, pd->dst.stride, pd->eobs[block]); + vp9_idct8x8_add(dqcoeff, dst, pd->dst.stride, pd->eobs[block]); break; case TX_4X4: // this is like vp9_short_idct4x4 but has a special case around eob<=1