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

Replace vp9_ prefix in 2D-DCT functions with vpx_

Browse Source 
Clean up the forward 2D-DCT function names in vpx_dsp.

Change-Id: I3117978596d198b690036e7eb05fe429caf3bc25
This commit is contained in:
Jingning Han 2015-07-28 15:57:40 -07:00
parent a7e9178d80
commit 4b5109cd73
25 changed files with 271 additions and 268 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
test/dct16x16_test.cc
View File

@ -272,7 +272,7 @@ typedef std::tr1::tuple<IdctFunc, IdctFunc, int, vpx_bit_depth_t>
void fdct16x16_ref(const int16_t *in, tran_low_t *out, int stride,
int /*tx_type*/) {
vp9_fdct16x16_c(in, out, stride);
vpx_fdct16x16_c(in, out, stride);
}
void idct16x16_ref(const tran_low_t *in, uint8_t *dest, int stride,
@ -822,14 +822,14 @@ using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_highbd_fdct16x16_c, &idct16x16_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct16x16_c, &idct16x16_12, 0, VPX_BITS_12),
make_tuple(&vp9_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
make_tuple(&vpx_highbd_fdct16x16_c, &idct16x16_10, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct16x16_c, &idct16x16_12, 0, VPX_BITS_12),
make_tuple(&vpx_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
#else
INSTANTIATE_TEST_CASE_P(
C, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
make_tuple(&vpx_fdct16x16_c, &vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
#endif // CONFIG_VP9_HIGHBITDEPTH
#if CONFIG_VP9_HIGHBITDEPTH
@ -862,7 +862,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
NEON, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_neon, 0, VPX_BITS_8)));
#endif
@ -870,7 +870,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_fdct16x16_sse2,
make_tuple(&vpx_fdct16x16_sse2,
&vp9_idct16x16_256_add_sse2, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
SSE2, Trans16x16HT,
@ -889,15 +889,15 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_highbd_fdct16x16_sse2,
make_tuple(&vpx_highbd_fdct16x16_sse2,
&idct16x16_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct16x16_c,
make_tuple(&vpx_highbd_fdct16x16_c,
&idct16x16_256_add_10_sse2, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct16x16_sse2,
make_tuple(&vpx_highbd_fdct16x16_sse2,
&idct16x16_12, 0, VPX_BITS_12),
make_tuple(&vp9_highbd_fdct16x16_c,
make_tuple(&vpx_highbd_fdct16x16_c,
&idct16x16_256_add_12_sse2, 0, VPX_BITS_12),
make_tuple(&vp9_fdct16x16_sse2,
make_tuple(&vpx_fdct16x16_sse2,
&vp9_idct16x16_256_add_c, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
SSE2, Trans16x16HT,
@ -926,7 +926,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
MSA, Trans16x16DCT,
::testing::Values(
make_tuple(&vp9_fdct16x16_msa,
make_tuple(&vpx_fdct16x16_msa,
&vp9_idct16x16_256_add_msa, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
MSA, Trans16x16HT,

48
test/dct32x32_test.cc
View File

@ -195,7 +195,7 @@ TEST_P(Trans32x32Test, CoeffCheck) {
input_block[j] = (rnd.Rand16() & mask_) - (rnd.Rand16() & mask_);
const int stride = 32;
vp9_fdct32x32_c(input_block, output_ref_block, stride);
vpx_fdct32x32_c(input_block, output_ref_block, stride);
ASM_REGISTER_STATE_CHECK(fwd_txfm_(input_block, output_block, stride));
if (version_ == 0) {
@ -232,7 +232,7 @@ TEST_P(Trans32x32Test, MemCheck) {
}
const int stride = 32;
vp9_fdct32x32_c(input_extreme_block, output_ref_block, stride);
vpx_fdct32x32_c(input_extreme_block, output_ref_block, stride);
ASM_REGISTER_STATE_CHECK(
fwd_txfm_(input_extreme_block, output_block, stride));
@ -315,25 +315,25 @@ using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_highbd_fdct32x32_c,
make_tuple(&vpx_highbd_fdct32x32_c,
&idct32x32_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct32x32_rd_c,
make_tuple(&vpx_highbd_fdct32x32_rd_c,
&idct32x32_10, 1, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct32x32_c,
make_tuple(&vpx_highbd_fdct32x32_c,
&idct32x32_12, 0, VPX_BITS_12),
make_tuple(&vp9_highbd_fdct32x32_rd_c,
make_tuple(&vpx_highbd_fdct32x32_rd_c,
&idct32x32_12, 1, VPX_BITS_12),
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_c,
make_tuple(&vpx_fdct32x32_rd_c,
&vp9_idct32x32_1024_add_c, 1, VPX_BITS_8)));
#else
INSTANTIATE_TEST_CASE_P(
C, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_c,
make_tuple(&vpx_fdct32x32_rd_c,
&vp9_idct32x32_1024_add_c, 1, VPX_BITS_8)));
#endif // CONFIG_VP9_HIGHBITDEPTH
@ -341,9 +341,9 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
NEON, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_neon, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_c,
make_tuple(&vpx_fdct32x32_rd_c,
&vp9_idct32x32_1024_add_neon, 1, VPX_BITS_8)));
#endif // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -351,9 +351,9 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_fdct32x32_sse2,
make_tuple(&vpx_fdct32x32_sse2,
&vp9_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_sse2,
make_tuple(&vpx_fdct32x32_rd_sse2,
&vp9_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
#endif // HAVE_SSE2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -361,15 +361,15 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_highbd_fdct32x32_sse2, &idct32x32_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct32x32_rd_sse2, &idct32x32_10, 1,
make_tuple(&vpx_highbd_fdct32x32_sse2, &idct32x32_10, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct32x32_rd_sse2, &idct32x32_10, 1,
VPX_BITS_10),
make_tuple(&vp9_highbd_fdct32x32_sse2, &idct32x32_12, 0, VPX_BITS_12),
make_tuple(&vp9_highbd_fdct32x32_rd_sse2, &idct32x32_12, 1,
make_tuple(&vpx_highbd_fdct32x32_sse2, &idct32x32_12, 0, VPX_BITS_12),
make_tuple(&vpx_highbd_fdct32x32_rd_sse2, &idct32x32_12, 1,
VPX_BITS_12),
make_tuple(&vp9_fdct32x32_sse2, &vp9_idct32x32_1024_add_c, 0,
make_tuple(&vpx_fdct32x32_sse2, &vp9_idct32x32_1024_add_c, 0,
VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_sse2, &vp9_idct32x32_1024_add_c, 1,
make_tuple(&vpx_fdct32x32_rd_sse2, &vp9_idct32x32_1024_add_c, 1,
VPX_BITS_8)));
#endif // HAVE_SSE2 && CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -377,9 +377,9 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
AVX2, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_fdct32x32_avx2,
make_tuple(&vpx_fdct32x32_avx2,
&vp9_idct32x32_1024_add_sse2, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_avx2,
make_tuple(&vpx_fdct32x32_rd_avx2,
&vp9_idct32x32_1024_add_sse2, 1, VPX_BITS_8)));
#endif // HAVE_AVX2 && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -387,9 +387,9 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
MSA, Trans32x32Test,
::testing::Values(
make_tuple(&vp9_fdct32x32_msa,
make_tuple(&vpx_fdct32x32_msa,
&vp9_idct32x32_1024_add_msa, 0, VPX_BITS_8),
make_tuple(&vp9_fdct32x32_rd_msa,
make_tuple(&vpx_fdct32x32_rd_msa,
&vp9_idct32x32_1024_add_msa, 1, VPX_BITS_8)));
#endif // HAVE_MSA && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
} // namespace

26
test/fdct4x4_test.cc
View File

@ -41,7 +41,7 @@ typedef std::tr1::tuple<FhtFunc, IhtFunc, int, vpx_bit_depth_t> Ht4x4Param;
void fdct4x4_ref(const int16_t *in, tran_low_t *out, int stride,
int tx_type) {
vp9_fdct4x4_c(in, out, stride);
vpx_fdct4x4_c(in, out, stride);
}
void fht4x4_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
@ -419,14 +419,14 @@ using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_12, 0, VPX_BITS_12),
make_tuple(&vp9_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_10, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_12, 0, VPX_BITS_12),
make_tuple(&vpx_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
#else
INSTANTIATE_TEST_CASE_P(
C, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
make_tuple(&vpx_fdct4x4_c, &vp9_idct4x4_16_add_c, 0, VPX_BITS_8)));
#endif // CONFIG_VP9_HIGHBITDEPTH
#if CONFIG_VP9_HIGHBITDEPTH
@ -473,7 +473,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
NEON, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_fdct4x4_c,
make_tuple(&vpx_fdct4x4_c,
&vp9_idct4x4_16_add_neon, 0, VPX_BITS_8)));
#endif // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -507,7 +507,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_fdct4x4_sse2,
make_tuple(&vpx_fdct4x4_sse2,
&vp9_idct4x4_16_add_sse2, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
SSE2, Trans4x4HT,
@ -522,11 +522,11 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_10_sse2, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct4x4_sse2, &idct4x4_10_sse2, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct4x4_c, &idct4x4_12_sse2, 0, VPX_BITS_12),
make_tuple(&vp9_highbd_fdct4x4_sse2, &idct4x4_12_sse2, 0, VPX_BITS_12),
make_tuple(&vp9_fdct4x4_sse2, &vp9_idct4x4_16_add_c, 0,
make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_10_sse2, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct4x4_sse2, &idct4x4_10_sse2, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct4x4_c, &idct4x4_12_sse2, 0, VPX_BITS_12),
make_tuple(&vpx_highbd_fdct4x4_sse2, &idct4x4_12_sse2, 0, VPX_BITS_12),
make_tuple(&vpx_fdct4x4_sse2, &vp9_idct4x4_16_add_c, 0,
VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
@ -542,7 +542,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
MSA, Trans4x4DCT,
::testing::Values(
make_tuple(&vp9_fdct4x4_msa, &vp9_idct4x4_16_add_msa, 0, VPX_BITS_8)));
make_tuple(&vpx_fdct4x4_msa, &vp9_idct4x4_16_add_msa, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
MSA, Trans4x4HT,
::testing::Values(

28
test/fdct8x8_test.cc
View File

@ -83,7 +83,7 @@ void reference_8x8_dct_2d(const int16_t input[kNumCoeffs],
void fdct8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
vp9_fdct8x8_c(in, out, stride);
vpx_fdct8x8_c(in, out, stride);
}
void fht8x8_ref(const int16_t *in, tran_low_t *out, int stride, int tx_type) {
@ -658,14 +658,14 @@ using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
make_tuple(&vp9_highbd_fdct8x8_c, &idct8x8_10, 0, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct8x8_c, &idct8x8_12, 0, VPX_BITS_12)));
make_tuple(&vpx_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
make_tuple(&vpx_highbd_fdct8x8_c, &idct8x8_10, 0, VPX_BITS_10),
make_tuple(&vpx_highbd_fdct8x8_c, &idct8x8_12, 0, VPX_BITS_12)));
#else
INSTANTIATE_TEST_CASE_P(
C, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8)));
make_tuple(&vpx_fdct8x8_c, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8)));
#endif // CONFIG_VP9_HIGHBITDEPTH
#if CONFIG_VP9_HIGHBITDEPTH
@ -698,7 +698,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
NEON, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_neon, &vp9_idct8x8_64_add_neon, 0,
make_tuple(&vpx_fdct8x8_neon, &vp9_idct8x8_64_add_neon, 0,
VPX_BITS_8)));
#endif // HAVE_NEON_ASM && !CONFIG_VP9_HIGHBITDEPTH && !CONFIG_EMULATE_HARDWARE
@ -716,7 +716,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_sse2, &vp9_idct8x8_64_add_sse2, 0,
make_tuple(&vpx_fdct8x8_sse2, &vp9_idct8x8_64_add_sse2, 0,
VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
SSE2, FwdTrans8x8HT,
@ -731,14 +731,14 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_sse2, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
make_tuple(&vp9_highbd_fdct8x8_c,
make_tuple(&vpx_fdct8x8_sse2, &vp9_idct8x8_64_add_c, 0, VPX_BITS_8),
make_tuple(&vpx_highbd_fdct8x8_c,
&idct8x8_64_add_10_sse2, 12, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct8x8_sse2,
make_tuple(&vpx_highbd_fdct8x8_sse2,
&idct8x8_64_add_10_sse2, 12, VPX_BITS_10),
make_tuple(&vp9_highbd_fdct8x8_c,
make_tuple(&vpx_highbd_fdct8x8_c,
&idct8x8_64_add_12_sse2, 12, VPX_BITS_12),
make_tuple(&vp9_highbd_fdct8x8_sse2,
make_tuple(&vpx_highbd_fdct8x8_sse2,
&idct8x8_64_add_12_sse2, 12, VPX_BITS_12)));
INSTANTIATE_TEST_CASE_P(
@ -769,7 +769,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSSE3, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_ssse3, &vp9_idct8x8_64_add_ssse3, 0,
make_tuple(&vpx_fdct8x8_ssse3, &vp9_idct8x8_64_add_ssse3, 0,
VPX_BITS_8)));
#endif
@ -777,7 +777,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
MSA, FwdTrans8x8DCT,
::testing::Values(
make_tuple(&vp9_fdct8x8_msa, &vp9_idct8x8_64_add_msa, 0, VPX_BITS_8)));
make_tuple(&vpx_fdct8x8_msa, &vp9_idct8x8_64_add_msa, 0, VPX_BITS_8)));
INSTANTIATE_TEST_CASE_P(
MSA, FwdTrans8x8HT,
::testing::Values(

56
test/partial_idct_test.cc
View File

@ -202,31 +202,31 @@ using std::tr1::make_tuple;
INSTANTIATE_TEST_CASE_P(
C, PartialIDctTest,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_34_add_c,
TX_32X32, 34),
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_1_add_c,
TX_32X32, 1),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_10_add_c,
TX_16X16, 10),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_1_add_c,
TX_16X16, 1),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_12_add_c,
TX_8X8, 12),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_1_add_c,
TX_8X8, 1),
make_tuple(&vp9_fdct4x4_c,
make_tuple(&vpx_fdct4x4_c,
&vp9_idct4x4_16_add_c,
&vp9_idct4x4_1_add_c,
TX_4X4, 1)));
@ -235,27 +235,27 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
NEON, PartialIDctTest,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_1_add_neon,
TX_32X32, 1),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_10_add_neon,
TX_16X16, 10),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_1_add_neon,
TX_16X16, 1),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_12_add_neon,
TX_8X8, 12),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_1_add_neon,
TX_8X8, 1),
make_tuple(&vp9_fdct4x4_c,
make_tuple(&vpx_fdct4x4_c,
&vp9_idct4x4_16_add_c,
&vp9_idct4x4_1_add_neon,
TX_4X4, 1)));
@ -265,31 +265,31 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSE2, PartialIDctTest,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_34_add_sse2,
TX_32X32, 34),
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_1_add_sse2,
TX_32X32, 1),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_10_add_sse2,
TX_16X16, 10),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_1_add_sse2,
TX_16X16, 1),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_12_add_sse2,
TX_8X8, 12),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_1_add_sse2,
TX_8X8, 1),
make_tuple(&vp9_fdct4x4_c,
make_tuple(&vpx_fdct4x4_c,
&vp9_idct4x4_16_add_c,
&vp9_idct4x4_1_add_sse2,
TX_4X4, 1)));
@ -300,7 +300,7 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
SSSE3_64, PartialIDctTest,
::testing::Values(
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_12_add_ssse3,
TX_8X8, 12)));
@ -310,31 +310,31 @@ INSTANTIATE_TEST_CASE_P(
INSTANTIATE_TEST_CASE_P(
MSA, PartialIDctTest,
::testing::Values(
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_34_add_msa,
TX_32X32, 34),
make_tuple(&vp9_fdct32x32_c,
make_tuple(&vpx_fdct32x32_c,
&vp9_idct32x32_1024_add_c,
&vp9_idct32x32_1_add_msa,
TX_32X32, 1),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_10_add_msa,
TX_16X16, 10),
make_tuple(&vp9_fdct16x16_c,
make_tuple(&vpx_fdct16x16_c,
&vp9_idct16x16_256_add_c,
&vp9_idct16x16_1_add_msa,
TX_16X16, 1),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_12_add_msa,
TX_8X8, 10),
make_tuple(&vp9_fdct8x8_c,
make_tuple(&vpx_fdct8x8_c,
&vp9_idct8x8_64_add_c,
&vp9_idct8x8_1_add_msa,
TX_8X8, 1),
make_tuple(&vp9_fdct4x4_c,
make_tuple(&vpx_fdct4x4_c,
&vp9_idct4x4_16_add_c,
&vp9_idct4x4_1_add_msa,
TX_4X4, 1)));

2
vp9/encoder/arm/neon/vp9_dct_neon.c
View File

@ -29,7 +29,7 @@ void vp9_fdct8x8_quant_neon(const int16_t *input, int stride,
int16_t temp_buffer[64];
(void)coeff_ptr;
vp9_fdct8x8_neon(input, temp_buffer, stride);
vpx_fdct8x8_neon(input, temp_buffer, stride);
vp9_quantize_fp_neon(temp_buffer, n_coeffs, skip_block, zbin_ptr, round_ptr,
quant_ptr, quant_shift_ptr, qcoeff_ptr, dqcoeff_ptr,
dequant_ptr, eob_ptr, scan_ptr, iscan_ptr);

6
vp9/encoder/vp9_dct.c
View File

@ -531,7 +531,7 @@ static const transform_2d FHT_16[] = {
void vp9_fht4x4_c(const int16_t *input, tran_low_t *output,
int stride, int tx_type) {
if (tx_type == DCT_DCT) {
vp9_fdct4x4_c(input, output, stride);
vpx_fdct4x4_c(input, output, stride);
} else {
tran_low_t out[4 * 4];
int i, j;
@ -675,7 +675,7 @@ void vp9_fdct8x8_quant_c(const int16_t *input, int stride,
void vp9_fht8x8_c(const int16_t *input, tran_low_t *output,
int stride, int tx_type) {
if (tx_type == DCT_DCT) {
vp9_fdct8x8_c(input, output, stride);
vpx_fdct8x8_c(input, output, stride);
} else {
tran_low_t out[64];
int i, j;
@ -761,7 +761,7 @@ void vp9_fwht4x4_c(const int16_t *input, tran_low_t *output, int stride) {
void vp9_fht16x16_c(const int16_t *input, tran_low_t *output,
int stride, int tx_type) {
if (tx_type == DCT_DCT) {
vp9_fdct16x16_c(input, output, stride);
vpx_fdct16x16_c(input, output, stride);
} else {
tran_low_t out[256];
int i, j;

6
vp9/encoder/vp9_encodeframe.c
View File

@ -3885,13 +3885,13 @@ static void encode_frame_internal(VP9_COMP *cpi) {
#if CONFIG_VP9_HIGHBITDEPTH
if (cm->use_highbitdepth)
x->fwd_txm4x4 = xd->lossless ? vp9_highbd_fwht4x4 : vp9_highbd_fdct4x4;
x->fwd_txm4x4 = xd->lossless ? vp9_highbd_fwht4x4 : vpx_highbd_fdct4x4;
else
x->fwd_txm4x4 = xd->lossless ? vp9_fwht4x4 : vp9_fdct4x4;
x->fwd_txm4x4 = xd->lossless ? vp9_fwht4x4 : vpx_fdct4x4;
x->highbd_itxm_add = xd->lossless ? vp9_highbd_iwht4x4_add :
vp9_highbd_idct4x4_add;
#else
x->fwd_txm4x4 = xd->lossless ? vp9_fwht4x4 : vp9_fdct4x4;
x->fwd_txm4x4 = xd->lossless ? vp9_fwht4x4 : vpx_fdct4x4;
#endif // CONFIG_VP9_HIGHBITDEPTH
x->itxm_add = xd->lossless ? vp9_iwht4x4_add : vp9_idct4x4_add;

38
vp9/encoder/vp9_encodemb.c
View File

@ -308,18 +308,18 @@ static INLINE void fdct32x32(int rd_transform,
const int16_t *src, tran_low_t *dst,
int src_stride) {
if (rd_transform)
vp9_fdct32x32_rd(src, dst, src_stride);
vpx_fdct32x32_rd(src, dst, src_stride);
else
vp9_fdct32x32(src, dst, src_stride);
vpx_fdct32x32(src, dst, src_stride);
}
#if CONFIG_VP9_HIGHBITDEPTH
static INLINE void highbd_fdct32x32(int rd_transform, const int16_t *src,
tran_low_t *dst, int src_stride) {
if (rd_transform)
vp9_highbd_fdct32x32_rd(src, dst, src_stride);
vpx_highbd_fdct32x32_rd(src, dst, src_stride);
else
vp9_highbd_fdct32x32(src, dst, src_stride);
vpx_highbd_fdct32x32(src, dst, src_stride);
}
#endif // CONFIG_VP9_HIGHBITDEPTH
@ -351,14 +351,14 @@ void vp9_xform_quant_fp(MACROBLOCK *x, int plane, int block,
scan_order->iscan);
break;
case TX_16X16:
vp9_highbd_fdct16x16(src_diff, coeff, diff_stride);
vpx_highbd_fdct16x16(src_diff, coeff, diff_stride);
vp9_highbd_quantize_fp(coeff, 256, x->skip_block, p->zbin, p->round_fp,
p->quant_fp, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
scan_order->scan, scan_order->iscan);
break;
case TX_8X8:
vp9_highbd_fdct8x8(src_diff, coeff, diff_stride);
vpx_highbd_fdct8x8(src_diff, coeff, diff_stride);
vp9_highbd_quantize_fp(coeff, 64, x->skip_block, p->zbin, p->round_fp,
p->quant_fp, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
@ -387,7 +387,7 @@ void vp9_xform_quant_fp(MACROBLOCK *x, int plane, int block,
scan_order->iscan);
break;
case TX_16X16:
vp9_fdct16x16(src_diff, coeff, diff_stride);
vpx_fdct16x16(src_diff, coeff, diff_stride);
vp9_quantize_fp(coeff, 256, x->skip_block, p->zbin, p->round_fp,
p->quant_fp, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
@ -433,19 +433,19 @@ void vp9_xform_quant_dc(MACROBLOCK *x, int plane, int block,
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
switch (tx_size) {
case TX_32X32:
vp9_highbd_fdct32x32_1(src_diff, coeff, diff_stride);
vpx_highbd_fdct32x32_1(src_diff, coeff, diff_stride);
vp9_highbd_quantize_dc_32x32(coeff, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
break;
case TX_16X16:
vp9_highbd_fdct16x16_1(src_diff, coeff, diff_stride);
vpx_highbd_fdct16x16_1(src_diff, coeff, diff_stride);
vp9_highbd_quantize_dc(coeff, 256, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
break;
case TX_8X8:
vp9_highbd_fdct8x8_1(src_diff, coeff, diff_stride);
vpx_highbd_fdct8x8_1(src_diff, coeff, diff_stride);
vp9_highbd_quantize_dc(coeff, 64, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
@ -465,19 +465,19 @@ void vp9_xform_quant_dc(MACROBLOCK *x, int plane, int block,
switch (tx_size) {
case TX_32X32:
vp9_fdct32x32_1(src_diff, coeff, diff_stride);
vpx_fdct32x32_1(src_diff, coeff, diff_stride);
vp9_quantize_dc_32x32(coeff, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
break;
case TX_16X16:
vp9_fdct16x16_1(src_diff, coeff, diff_stride);
vpx_fdct16x16_1(src_diff, coeff, diff_stride);
vp9_quantize_dc(coeff, 256, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
break;
case TX_8X8:
vp9_fdct8x8_1(src_diff, coeff, diff_stride);
vpx_fdct8x8_1(src_diff, coeff, diff_stride);
vp9_quantize_dc(coeff, 64, x->skip_block, p->round,
p->quant_fp[0], qcoeff, dqcoeff,
pd->dequant[0], eob);
@ -521,14 +521,14 @@ void vp9_xform_quant(MACROBLOCK *x, int plane, int block,
scan_order->scan, scan_order->iscan);
break;
case TX_16X16:
vp9_highbd_fdct16x16(src_diff, coeff, diff_stride);
vpx_highbd_fdct16x16(src_diff, coeff, diff_stride);
vp9_highbd_quantize_b(coeff, 256, x->skip_block, p->zbin, p->round,
p->quant, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
scan_order->scan, scan_order->iscan);
break;
case TX_8X8:
vp9_highbd_fdct8x8(src_diff, coeff, diff_stride);
vpx_highbd_fdct8x8(src_diff, coeff, diff_stride);
vp9_highbd_quantize_b(coeff, 64, x->skip_block, p->zbin, p->round,
p->quant, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
@ -557,14 +557,14 @@ void vp9_xform_quant(MACROBLOCK *x, int plane, int block,
scan_order->iscan);
break;
case TX_16X16:
vp9_fdct16x16(src_diff, coeff, diff_stride);
vpx_fdct16x16(src_diff, coeff, diff_stride);
vp9_quantize_b(coeff, 256, x->skip_block, p->zbin, p->round,
p->quant, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
scan_order->scan, scan_order->iscan);
break;
case TX_8X8:
vp9_fdct8x8(src_diff, coeff, diff_stride);
vpx_fdct8x8(src_diff, coeff, diff_stride);
vp9_quantize_b(coeff, 64, x->skip_block, p->zbin, p->round,
p->quant, p->quant_shift, qcoeff, dqcoeff,
pd->dequant, eob,
@ -829,7 +829,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
vpx_highbd_subtract_block(16, 16, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
if (tx_type == DCT_DCT)
vp9_highbd_fdct16x16(src_diff, coeff, diff_stride);
vpx_highbd_fdct16x16(src_diff, coeff, diff_stride);
else
vp9_highbd_fht16x16(src_diff, coeff, diff_stride, tx_type);
vp9_highbd_quantize_b(coeff, 256, x->skip_block, p->zbin, p->round,
@ -847,7 +847,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
vpx_highbd_subtract_block(8, 8, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
if (tx_type == DCT_DCT)
vp9_highbd_fdct8x8(src_diff, coeff, diff_stride);
vpx_highbd_fdct8x8(src_diff, coeff, diff_stride);
else
vp9_highbd_fht8x8(src_diff, coeff, diff_stride, tx_type);
vp9_highbd_quantize_b(coeff, 64, x->skip_block, p->zbin, p->round,

6
vp9/encoder/vp9_encoder.c
View File

@ -4306,13 +4306,13 @@ int vp9_get_compressed_data(VP9_COMP *cpi, unsigned int *frame_flags,
#if CONFIG_VP9_HIGHBITDEPTH
if (cpi->oxcf.use_highbitdepth)
cpi->td.mb.fwd_txm4x4 = lossless ?
vp9_highbd_fwht4x4 : vp9_highbd_fdct4x4;
vp9_highbd_fwht4x4 : vpx_highbd_fdct4x4;
else
cpi->td.mb.fwd_txm4x4 = lossless ? vp9_fwht4x4 : vp9_fdct4x4;
cpi->td.mb.fwd_txm4x4 = lossless ? vp9_fwht4x4 : vpx_fdct4x4;
cpi->td.mb.highbd_itxm_add = lossless ? vp9_highbd_iwht4x4_add :
vp9_highbd_idct4x4_add;
#else
cpi->td.mb.fwd_txm4x4 = lossless ? vp9_fwht4x4 : vp9_fdct4x4;
cpi->td.mb.fwd_txm4x4 = lossless ? vp9_fwht4x4 : vpx_fdct4x4;
#endif // CONFIG_VP9_HIGHBITDEPTH
cpi->td.mb.itxm_add = lossless ? vp9_iwht4x4_add : vp9_idct4x4_add;
vp9_first_pass(cpi, source);

2
vp9/encoder/vp9_pickmode.c
View File

@ -611,7 +611,7 @@ static void block_yrd(VP9_COMP *cpi, MACROBLOCK *x, int *rate, int64_t *dist,
switch (tx_size) {
case TX_32X32:
vp9_fdct32x32_rd(src_diff, coeff, diff_stride);
vpx_fdct32x32_rd(src_diff, coeff, diff_stride);
vp9_quantize_fp_32x32(coeff, 1024, x->skip_block, p->zbin,
p->round_fp, p->quant_fp, p->quant_shift,
qcoeff, dqcoeff, pd->dequant, eob,

2
vp9/encoder/vp9_psnrhvs.c
View File

@ -26,7 +26,7 @@
void od_bin_fdct8x8(tran_low_t *y, int ystride, const int16_t *x, int xstride) {
(void) xstride;
vp9_fdct8x8(x, y, ystride);
vpx_fdct8x8(x, y, ystride);
}
/* Normalized inverse quantization matrix for 8x8 DCT at the point of

2
vp9/encoder/vp9_rdopt.c
View File

@ -846,7 +846,7 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x,
const TX_TYPE tx_type = get_tx_type_4x4(PLANE_TYPE_Y, xd, block);
const scan_order *so = &vp9_scan_orders[TX_4X4][tx_type];
if (tx_type == DCT_DCT)
vp9_highbd_fdct4x4(src_diff, coeff, 8);
vpx_highbd_fdct4x4(src_diff, coeff, 8);
else
vp9_highbd_fht4x4(src_diff, coeff, 8, tx_type);
vp9_regular_quantize_b_4x4(x, 0, block, so->scan, so->iscan);

6
vp9/encoder/x86/vp9_dct_sse2.c
View File

@ -157,7 +157,7 @@ void vp9_fht4x4_sse2(const int16_t *input, tran_low_t *output,
switch (tx_type) {
case DCT_DCT:
vp9_fdct4x4_sse2(input, output, stride);
vpx_fdct4x4_sse2(input, output, stride);
break;
case ADST_DCT:
load_buffer_4x4(input, in, stride);
@ -1141,7 +1141,7 @@ void vp9_fht8x8_sse2(const int16_t *input, tran_low_t *output,
switch (tx_type) {
case DCT_DCT:
vp9_fdct8x8_sse2(input, output, stride);
vpx_fdct8x8_sse2(input, output, stride);
break;
case ADST_DCT:
load_buffer_8x8(input, in, stride);
@ -2028,7 +2028,7 @@ void vp9_fht16x16_sse2(const int16_t *input, tran_low_t *output,
switch (tx_type) {
case DCT_DCT:
vp9_fdct16x16_sse2(input, output, stride);
vpx_fdct16x16_sse2(input, output, stride);
break;
case ADST_DCT:
load_buffer_16x16(input, in0, in1, stride);

4
vpx_dsp/arm/fwd_txfm_neon.c
View File

@ -13,7 +13,7 @@
#include "./vpx_config.h"
#include "vpx_dsp/txfm_common.h"
void vp9_fdct8x8_neon(const int16_t *input, int16_t *final_output, int stride) {
void vpx_fdct8x8_neon(const int16_t *input, int16_t *final_output, int stride) {
int i;
// stage 1
int16x8_t input_0 = vshlq_n_s16(vld1q_s16(&input[0 * stride]), 2);
@ -202,7 +202,7 @@ void vp9_fdct8x8_neon(const int16_t *input, int16_t *final_output, int stride) {
}
}
void vp9_fdct8x8_1_neon(const int16_t *input, int16_t *output, int stride) {
void vpx_fdct8x8_1_neon(const int16_t *input, int16_t *output, int stride) {
int r;
int16x8_t sum = vld1q_s16(&input[0]);
for (r = 1; r < 8; ++r) {

60
vpx_dsp/fwd_txfm.c
View File

@ -10,7 +10,7 @@
#include "vpx_dsp/fwd_txfm.h"
void vp9_fdct4x4_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct4x4_c(const int16_t *input, tran_low_t *output, int stride) {
// The 2D transform is done with two passes which are actually pretty
// similar. In the first one, we transform the columns and transpose
// the results. In the second one, we transform the rows. To achieve that,
@ -77,7 +77,7 @@ void vp9_fdct4x4_c(const int16_t *input, tran_low_t *output, int stride) {
}
}
void vp9_fdct4x4_1_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct4x4_1_c(const int16_t *input, tran_low_t *output, int stride) {
int r, c;
tran_low_t sum = 0;
for (r = 0; r < 4; ++r)
@ -88,7 +88,7 @@ void vp9_fdct4x4_1_c(const int16_t *input, tran_low_t *output, int stride) {
output[1] = 0;
}
void vp9_fdct8x8_c(const int16_t *input, tran_low_t *final_output, int stride) {
void vpx_fdct8x8_c(const int16_t *input, tran_low_t *final_output, int stride) {
int i, j;
tran_low_t intermediate[64];
int pass;
@ -174,7 +174,7 @@ void vp9_fdct8x8_c(const int16_t *input, tran_low_t *final_output, int stride) {
}
}
void vp9_fdct8x8_1_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct8x8_1_c(const int16_t *input, tran_low_t *output, int stride) {
int r, c;
tran_low_t sum = 0;
for (r = 0; r < 8; ++r)
@ -185,7 +185,7 @@ void vp9_fdct8x8_1_c(const int16_t *input, tran_low_t *output, int stride) {
output[1] = 0;
}
void vp9_fdct16x16_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct16x16_c(const int16_t *input, tran_low_t *output, int stride) {
// The 2D transform is done with two passes which are actually pretty
// similar. In the first one, we transform the columns and transpose
// the results. In the second one, we transform the rows. To achieve that,
@ -365,7 +365,7 @@ void vp9_fdct16x16_c(const int16_t *input, tran_low_t *output, int stride) {
}
}
void vp9_fdct16x16_1_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct16x16_1_c(const int16_t *input, tran_low_t *output, int stride) {
int r, c;
tran_low_t sum = 0;
for (r = 0; r < 16; ++r)
@ -389,7 +389,7 @@ static INLINE tran_high_t half_round_shift(tran_high_t input) {
return rv;
}
void vp9_fdct32(const tran_high_t *input, tran_high_t *output, int round) {
void vpx_fdct32(const tran_high_t *input, tran_high_t *output, int round) {
tran_high_t step[32];
// Stage 1
step[0] = input[0] + input[(32 - 1)];
@ -712,7 +712,7 @@ void vp9_fdct32(const tran_high_t *input, tran_high_t *output, int round) {
output[31] = dct_32_round(step[31] * cospi_31_64 + step[16] * -cospi_1_64);
}
void vp9_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
void vpx_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
int i, j;
tran_high_t output[32 * 32];
@ -721,7 +721,7 @@ void vp9_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = input[j * stride + i] * 4;
vp9_fdct32(temp_in, temp_out, 0);
vpx_fdct32(temp_in, temp_out, 0);
for (j = 0; j < 32; ++j)
output[j * 32 + i] = (temp_out[j] + 1 + (temp_out[j] > 0)) >> 2;
}
@ -731,7 +731,7 @@ void vp9_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = output[j + i * 32];
vp9_fdct32(temp_in, temp_out, 0);
vpx_fdct32(temp_in, temp_out, 0);
for (j = 0; j < 32; ++j)
out[j + i * 32] =
(tran_low_t)((temp_out[j] + 1 + (temp_out[j] < 0)) >> 2);
@ -741,7 +741,7 @@ void vp9_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
// Note that although we use dct_32_round in dct32 computation flow,
// this 2d fdct32x32 for rate-distortion optimization loop is operating
// within 16 bits precision.
void vp9_fdct32x32_rd_c(const int16_t *input, tran_low_t *out, int stride) {
void vpx_fdct32x32_rd_c(const int16_t *input, tran_low_t *out, int stride) {
int i, j;
tran_high_t output[32 * 32];
@ -750,7 +750,7 @@ void vp9_fdct32x32_rd_c(const int16_t *input, tran_low_t *out, int stride) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = input[j * stride + i] * 4;
vp9_fdct32(temp_in, temp_out, 0);
vpx_fdct32(temp_in, temp_out, 0);
for (j = 0; j < 32; ++j)
// TODO(cd): see quality impact of only doing
// output[j * 32 + i] = (temp_out[j] + 1) >> 2;
@ -763,13 +763,13 @@ void vp9_fdct32x32_rd_c(const int16_t *input, tran_low_t *out, int stride) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = output[j + i * 32];
vp9_fdct32(temp_in, temp_out, 1);
vpx_fdct32(temp_in, temp_out, 1);
for (j = 0; j < 32; ++j)
out[j + i * 32] = (tran_low_t)temp_out[j];
}
}
void vp9_fdct32x32_1_c(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct32x32_1_c(const int16_t *input, tran_low_t *output, int stride) {
int r, c;
tran_low_t sum = 0;
for (r = 0; r < 32; ++r)
@ -781,42 +781,42 @@ void vp9_fdct32x32_1_c(const int16_t *input, tran_low_t *output, int stride) {
}
#if CONFIG_VP9_HIGHBITDEPTH
void vp9_highbd_fdct4x4_c(const int16_t *input, tran_low_t *output,
void vpx_highbd_fdct4x4_c(const int16_t *input, tran_low_t *output,
int stride) {
vp9_fdct4x4_c(input, output, stride);
vpx_fdct4x4_c(input, output, stride);
}
void vp9_highbd_fdct8x8_c(const int16_t *input, tran_low_t *final_output,
void vpx_highbd_fdct8x8_c(const int16_t *input, tran_low_t *final_output,
int stride) {
vp9_fdct8x8_c(input, final_output, stride);
vpx_fdct8x8_c(input, final_output, stride);
}
void vp9_highbd_fdct8x8_1_c(const int16_t *input, tran_low_t *final_output,
void vpx_highbd_fdct8x8_1_c(const int16_t *input, tran_low_t *final_output,
int stride) {
vp9_fdct8x8_1_c(input, final_output, stride);
vpx_fdct8x8_1_c(input, final_output, stride);
}
void vp9_highbd_fdct16x16_c(const int16_t *input, tran_low_t *output,
void vpx_highbd_fdct16x16_c(const int16_t *input, tran_low_t *output,
int stride) {
vp9_fdct16x16_c(input, output, stride);
vpx_fdct16x16_c(input, output, stride);
}
void vp9_highbd_fdct16x16_1_c(const int16_t *input, tran_low_t *output,
void vpx_highbd_fdct16x16_1_c(const int16_t *input, tran_low_t *output,
int stride) {
vp9_fdct16x16_1_c(input, output, stride);
vpx_fdct16x16_1_c(input, output, stride);
}
void vp9_highbd_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
vp9_fdct32x32_c(input, out, stride);
void vpx_highbd_fdct32x32_c(const int16_t *input, tran_low_t *out, int stride) {
vpx_fdct32x32_c(input, out, stride);
}
void vp9_highbd_fdct32x32_rd_c(const int16_t *input, tran_low_t *out,
void vpx_highbd_fdct32x32_rd_c(const int16_t *input, tran_low_t *out,
int stride) {
vp9_fdct32x32_rd_c(input, out, stride);
vpx_fdct32x32_rd_c(input, out, stride);
}
void vp9_highbd_fdct32x32_1_c(const int16_t *input, tran_low_t *out,
void vpx_highbd_fdct32x32_1_c(const int16_t *input, tran_low_t *out,
int stride) {
vp9_fdct32x32_1_c(input, out, stride);
vpx_fdct32x32_1_c(input, out, stride);
}
#endif // CONFIG_VP9_HIGHBITDEPTH

2
vpx_dsp/fwd_txfm.h
View File

@ -21,5 +21,5 @@ static INLINE tran_high_t fdct_round_shift(tran_high_t input) {
return rv;
}
void vp9_fdct32(const tran_high_t *input, tran_high_t *output, int round);
void vpx_fdct32(const tran_high_t *input, tran_high_t *output, int round);
#endif // VPX_DSP_FWD_TXFM_H_

6
vpx_dsp/mips/fwd_dct32x32_msa.c
View File

@ -675,7 +675,7 @@ static void fdct32x8_1d_row_4x(int16_t *tmp_buf_big, int16_t *tmp_buf,
fdct8x32_1d_row_transpose_store(tmp_buf, output);
}
void vp9_fdct32x32_msa(const int16_t *input, int16_t *output,
void vpx_fdct32x32_msa(const int16_t *input, int16_t *output,
int32_t src_stride) {
int32_t i;
DECLARE_ALIGNED(32, int16_t, tmp_buf_big[1024]);
@ -913,7 +913,7 @@ static void fdct32x8_1d_row_rd(int16_t *tmp_buf_big, int16_t *tmp_buf,
fdct8x32_1d_row_transpose_store(tmp_buf, output);
}
void vp9_fdct32x32_rd_msa(const int16_t *input, int16_t *out,
void vpx_fdct32x32_rd_msa(const int16_t *input, int16_t *out,
int32_t src_stride) {
int32_t i;
DECLARE_ALIGNED(32, int16_t, tmp_buf_big[1024]);
@ -932,7 +932,7 @@ void vp9_fdct32x32_rd_msa(const int16_t *input, int16_t *out,
}
}
void vp9_fdct32x32_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
void vpx_fdct32x32_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
out[1] = 0;
out[0] = LD_HADD(input, stride);

10
vpx_dsp/mips/fwd_txfm_msa.c
View File

@ -166,7 +166,7 @@ void fdct16x8_1d_row(int16_t *input, int16_t *output) {
ST_SH8(tmp4, in4, tmp5, in5, tmp6, in6, tmp7, in7, output + 8, 16);
}
void vp9_fdct4x4_msa(const int16_t *input, int16_t *output,
void vpx_fdct4x4_msa(const int16_t *input, int16_t *output,
int32_t src_stride) {
v8i16 in0, in1, in2, in3;
@ -196,7 +196,7 @@ void vp9_fdct4x4_msa(const int16_t *input, int16_t *output,
ST_SH2(in0, in2, output, 8);
}
void vp9_fdct8x8_msa(const int16_t *input, int16_t *output,
void vpx_fdct8x8_msa(const int16_t *input, int16_t *output,
int32_t src_stride) {
v8i16 in0, in1, in2, in3, in4, in5, in6, in7;
@ -215,12 +215,12 @@ void vp9_fdct8x8_msa(const int16_t *input, int16_t *output,
ST_SH8(in0, in1, in2, in3, in4, in5, in6, in7, output, 8);
}
void vp9_fdct8x8_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
void vpx_fdct8x8_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
out[0] = LD_HADD(input, stride);
out[1] = 0;
}
void vp9_fdct16x16_msa(const int16_t *input, int16_t *output,
void vpx_fdct16x16_msa(const int16_t *input, int16_t *output,
int32_t src_stride) {
int32_t i;
DECLARE_ALIGNED(32, int16_t, tmp_buf[16 * 16]);
@ -236,7 +236,7 @@ void vp9_fdct16x16_msa(const int16_t *input, int16_t *output,
}
}
void vp9_fdct16x16_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
void vpx_fdct16x16_1_msa(const int16_t *input, int16_t *out, int32_t stride) {
out[1] = 0;
out[0] = LD_HADD(input, stride);

104
vpx_dsp/vpx_dsp_rtcd_defs.pl
View File

@ -129,83 +129,83 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
#
if (vpx_config("CONFIG_VP9_ENCODER") eq "yes") {
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
add_proto qw/void vp9_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct4x4 sse2/;
add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct4x4 sse2/;
add_proto qw/void vp9_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct4x4_1 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 vp9_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct8x8 sse2/;
add_proto qw/void vpx_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct8x8 sse2/;
add_proto qw/void vp9_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct8x8_1 sse2/;
add_proto qw/void vpx_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct8x8_1 sse2/;
add_proto qw/void vp9_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct16x16 sse2/;
add_proto qw/void vpx_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct16x16 sse2/;
add_proto qw/void vp9_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct16x16_1 sse2/;
add_proto qw/void vpx_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct16x16_1 sse2/;
add_proto qw/void vp9_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32 sse2/;
add_proto qw/void vpx_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32 sse2/;
add_proto qw/void vp9_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32_rd sse2/;
add_proto qw/void vpx_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32_rd sse2/;
add_proto qw/void vp9_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32_1 sse2/;
add_proto qw/void vpx_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32_1 sse2/;
add_proto qw/void vp9_highbd_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct4x4 sse2/;
add_proto qw/void vpx_highbd_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct4x4 sse2/;
add_proto qw/void vp9_highbd_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct8x8 sse2/;
add_proto qw/void vpx_highbd_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct8x8 sse2/;
add_proto qw/void vp9_highbd_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct8x8_1/;
add_proto qw/void vpx_highbd_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct8x8_1/;
add_proto qw/void vp9_highbd_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct16x16 sse2/;
add_proto qw/void vpx_highbd_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct16x16 sse2/;
add_proto qw/void vp9_highbd_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct16x16_1/;
add_proto qw/void vpx_highbd_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct16x16_1/;
add_proto qw/void vp9_highbd_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct32x32 sse2/;
add_proto qw/void vpx_highbd_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct32x32 sse2/;
add_proto qw/void vp9_highbd_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct32x32_rd sse2/;
add_proto qw/void vpx_highbd_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct32x32_rd sse2/;
add_proto qw/void vp9_highbd_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_highbd_fdct32x32_1/;
add_proto qw/void vpx_highbd_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_highbd_fdct32x32_1/;
} else {
add_proto qw/void vp9_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct4x4 sse2 msa/;
add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct4x4 sse2 msa/;
add_proto qw/void vp9_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct4x4_1 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 vp9_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct8x8 sse2 neon msa/, "$ssse3_x86_64_x86inc";
add_proto qw/void vpx_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct8x8 sse2 neon msa/, "$ssse3_x86_64_x86inc";
add_proto qw/void vp9_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct8x8_1 sse2 neon msa/;
add_proto qw/void vpx_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct8x8_1 sse2 neon msa/;
add_proto qw/void vp9_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct16x16 sse2 msa/;
add_proto qw/void vpx_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct16x16 sse2 msa/;
add_proto qw/void vp9_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct16x16_1 sse2 msa/;
add_proto qw/void vpx_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct16x16_1 sse2 msa/;
add_proto qw/void vp9_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32 sse2 avx2 msa/;
add_proto qw/void vpx_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32 sse2 avx2 msa/;
add_proto qw/void vp9_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32_rd sse2 avx2 msa/;
add_proto qw/void vpx_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32_rd sse2 avx2 msa/;
add_proto qw/void vp9_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vp9_fdct32x32_1 sse2 msa/;
add_proto qw/void vpx_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
specialize qw/vpx_fdct32x32_1 sse2 msa/;
} # CONFIG_VP9_HIGHBITDEPTH
} # CONFIG_VP9_ENCODER

16
vpx_dsp/x86/fwd_dct32x32_impl_sse2.h
View File

@ -21,34 +21,34 @@
#define ADD_EPI16 _mm_adds_epi16
#define SUB_EPI16 _mm_subs_epi16
#if FDCT32x32_HIGH_PRECISION
void vp9_fdct32x32_rows_c(const int16_t *intermediate, tran_low_t *out) {
void vpx_fdct32x32_rows_c(const int16_t *intermediate, tran_low_t *out) {
int i, j;
for (i = 0; i < 32; ++i) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = intermediate[j * 32 + i];
vp9_fdct32(temp_in, temp_out, 0);
vpx_fdct32(temp_in, temp_out, 0);
for (j = 0; j < 32; ++j)
out[j + i * 32] =
(tran_low_t)((temp_out[j] + 1 + (temp_out[j] < 0)) >> 2);
}
}
#define HIGH_FDCT32x32_2D_C vp9_highbd_fdct32x32_c
#define HIGH_FDCT32x32_2D_ROWS_C vp9_fdct32x32_rows_c
#define HIGH_FDCT32x32_2D_C vpx_highbd_fdct32x32_c
#define HIGH_FDCT32x32_2D_ROWS_C vpx_fdct32x32_rows_c
#else
void vp9_fdct32x32_rd_rows_c(const int16_t *intermediate, tran_low_t *out) {
void vpx_fdct32x32_rd_rows_c(const int16_t *intermediate, tran_low_t *out) {
int i, j;
for (i = 0; i < 32; ++i) {
tran_high_t temp_in[32], temp_out[32];
for (j = 0; j < 32; ++j)
temp_in[j] = intermediate[j * 32 + i];
vp9_fdct32(temp_in, temp_out, 1);
vpx_fdct32(temp_in, temp_out, 1);
for (j = 0; j < 32; ++j)
out[j + i * 32] = (tran_low_t)temp_out[j];
}
}
#define HIGH_FDCT32x32_2D_C vp9_highbd_fdct32x32_rd_c
#define HIGH_FDCT32x32_2D_ROWS_C vp9_fdct32x32_rd_rows_c
#define HIGH_FDCT32x32_2D_C vpx_highbd_fdct32x32_rd_c
#define HIGH_FDCT32x32_2D_ROWS_C vpx_fdct32x32_rd_rows_c
#endif // FDCT32x32_HIGH_PRECISION
#else
#define ADD_EPI16 _mm_add_epi16

4
vpx_dsp/x86/fwd_txfm_avx2.c
View File

@ -10,13 +10,13 @@
#include "./vpx_config.h"
#define FDCT32x32_2D_AVX2 vp9_fdct32x32_rd_avx2
#define FDCT32x32_2D_AVX2 vpx_fdct32x32_rd_avx2
#define FDCT32x32_HIGH_PRECISION 0
#include "vpx_dsp/x86/fwd_dct32x32_impl_avx2.h"
#undef FDCT32x32_2D_AVX2
#undef FDCT32x32_HIGH_PRECISION
#define FDCT32x32_2D_AVX2 vp9_fdct32x32_avx2
#define FDCT32x32_2D_AVX2 vpx_fdct32x32_avx2
#define FDCT32x32_HIGH_PRECISION 1
#include "vpx_dsp/x86/fwd_dct32x32_impl_avx2.h" // NOLINT
#undef FDCT32x32_2D_AVX2

48
vpx_dsp/x86/fwd_txfm_impl_sse2.h
View File

@ -106,7 +106,7 @@ void FDCT4x4_2D(const int16_t *input, tran_low_t *output, int stride) {
_mm_cmplt_epi16(in1, _mm_set1_epi16(0xfc00)));
test = _mm_movemask_epi8(_mm_or_si128(cmp0, cmp1));
if (test) {
vp9_highbd_fdct4x4_c(input, output, stride);
vpx_highbd_fdct4x4_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -177,7 +177,7 @@ void FDCT4x4_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x2(&x0, &x1);
if (overflow) {
vp9_highbd_fdct4x4_c(input, output, stride);
vpx_highbd_fdct4x4_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -199,7 +199,7 @@ void FDCT4x4_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x2(&t0, &t1);
if (overflow) {
vp9_highbd_fdct4x4_c(input, output, stride);
vpx_highbd_fdct4x4_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -238,7 +238,7 @@ void FDCT4x4_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x2(&x0, &x1);
if (overflow) {
vp9_highbd_fdct4x4_c(input, output, stride);
vpx_highbd_fdct4x4_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -322,7 +322,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x8(&q0, &q1, &q2, &q3,
&q4, &q5, &q6, &q7);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
}
@ -337,7 +337,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&r0, &r1, &r2, &r3);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -380,7 +380,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&res0, &res4, &res2, &res6);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -410,7 +410,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x2(&r0, &r1);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -423,7 +423,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&x0, &x1, &x2, &x3);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -466,7 +466,7 @@ void FDCT8x8_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&res1, &res7, &res5, &res3);
if (overflow) {
vp9_highbd_fdct8x8_c(input, output, stride);
vpx_highbd_fdct8x8_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -729,7 +729,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x8(&input0, &input1, &input2, &input3,
&input4, &input5, &input6, &input7);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -750,7 +750,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
&step1_4, &step1_5,
&step1_6, &step1_7);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -770,7 +770,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x8(&q0, &q1, &q2, &q3,
&q4, &q5, &q6, &q7);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -784,7 +784,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&r0, &r1, &r2, &r3);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -806,7 +806,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&res00, &res08, &res04, &res12);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -827,7 +827,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x2(&r0, &r1);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -840,7 +840,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&x0, &x1, &x2, &x3);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -863,7 +863,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x4(&res02, &res14,
&res10, &res06);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -891,7 +891,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x4(&step2_2, &step2_3, &step2_5,
&step2_4);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -912,7 +912,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
&step3_4, &step3_5,
&step3_6, &step3_7);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -935,7 +935,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
overflow = check_epi16_overflow_x4(&step2_1, &step2_2, &step2_6,
&step2_5);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -956,7 +956,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
&step1_4, &step1_5,
&step1_6, &step1_7);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -978,7 +978,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&res01, &res09, &res15, &res07);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH
@ -999,7 +999,7 @@ void FDCT16x16_2D(const int16_t *input, tran_low_t *output, int stride) {
#if DCT_HIGH_BIT_DEPTH
overflow = check_epi16_overflow_x4(&res05, &res13, &res11, &res03);
if (overflow) {
vp9_highbd_fdct16x16_c(input, output, stride);
vpx_highbd_fdct16x16_c(input, output, stride);
return;
}
#endif // DCT_HIGH_BIT_DEPTH

28
vpx_dsp/x86/fwd_txfm_sse2.c
View File

@ -14,7 +14,7 @@
#include "vpx_dsp/vpx_dsp_common.h"
#include "vpx_dsp/x86/fwd_txfm_sse2.h"
void vp9_fdct4x4_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct4x4_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
__m128i in0, in1;
__m128i tmp;
const __m128i zero = _mm_setzero_si128();
@ -43,7 +43,7 @@ void vp9_fdct4x4_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
store_output(&in0, output);
}
void vp9_fdct8x8_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
void vpx_fdct8x8_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
__m128i in0 = _mm_load_si128((const __m128i *)(input + 0 * stride));
__m128i in1 = _mm_load_si128((const __m128i *)(input + 1 * stride));
__m128i in2 = _mm_load_si128((const __m128i *)(input + 2 * stride));
@ -83,7 +83,7 @@ void vp9_fdct8x8_1_sse2(const int16_t *input, tran_low_t *output, int stride) {
store_output(&in1, output);
}
void vp9_fdct16x16_1_sse2(const int16_t *input, tran_low_t *output,
void vpx_fdct16x16_1_sse2(const int16_t *input, tran_low_t *output,
int stride) {
__m128i in0, in1, in2, in3;
__m128i u0, u1;
@ -152,7 +152,7 @@ void vp9_fdct16x16_1_sse2(const int16_t *input, tran_low_t *output,
store_output(&in1, output);
}
void vp9_fdct32x32_1_sse2(const int16_t *input, tran_low_t *output,
void vpx_fdct32x32_1_sse2(const int16_t *input, tran_low_t *output,
int stride) {
__m128i in0, in1, in2, in3;
__m128i u0, u1;
@ -225,21 +225,21 @@ void vp9_fdct32x32_1_sse2(const int16_t *input, tran_low_t *output,
}
#define DCT_HIGH_BIT_DEPTH 0
#define FDCT4x4_2D vp9_fdct4x4_sse2
#define FDCT8x8_2D vp9_fdct8x8_sse2
#define FDCT16x16_2D vp9_fdct16x16_sse2
#define FDCT4x4_2D vpx_fdct4x4_sse2
#define FDCT8x8_2D vpx_fdct8x8_sse2
#define FDCT16x16_2D vpx_fdct16x16_sse2
#include "vpx_dsp/x86/fwd_txfm_impl_sse2.h"
#undef FDCT4x4_2D
#undef FDCT8x8_2D
#undef FDCT16x16_2D
#define FDCT32x32_2D vp9_fdct32x32_rd_sse2
#define FDCT32x32_2D vpx_fdct32x32_rd_sse2
#define FDCT32x32_HIGH_PRECISION 0
#include "vpx_dsp/x86/fwd_dct32x32_impl_sse2.h"
#undef FDCT32x32_2D
#undef FDCT32x32_HIGH_PRECISION
#define FDCT32x32_2D vp9_fdct32x32_sse2
#define FDCT32x32_2D vpx_fdct32x32_sse2
#define FDCT32x32_HIGH_PRECISION 1
#include "vpx_dsp/x86/fwd_dct32x32_impl_sse2.h" // NOLINT
#undef FDCT32x32_2D
@ -248,21 +248,21 @@ void vp9_fdct32x32_1_sse2(const int16_t *input, tran_low_t *output,
#if CONFIG_VP9_HIGHBITDEPTH
#define DCT_HIGH_BIT_DEPTH 1
#define FDCT4x4_2D vp9_highbd_fdct4x4_sse2
#define FDCT8x8_2D vp9_highbd_fdct8x8_sse2
#define FDCT16x16_2D vp9_highbd_fdct16x16_sse2
#define FDCT4x4_2D vpx_highbd_fdct4x4_sse2
#define FDCT8x8_2D vpx_highbd_fdct8x8_sse2
#define FDCT16x16_2D vpx_highbd_fdct16x16_sse2
#include "vpx_dsp/x86/fwd_txfm_impl_sse2.h" // NOLINT
#undef FDCT4x4_2D
#undef FDCT8x8_2D
#undef FDCT16x16_2D
#define FDCT32x32_2D vp9_highbd_fdct32x32_rd_sse2
#define FDCT32x32_2D vpx_highbd_fdct32x32_rd_sse2
#define FDCT32x32_HIGH_PRECISION 0
#include "vpx_dsp/x86/fwd_dct32x32_impl_sse2.h" // NOLINT
#undef FDCT32x32_2D
#undef FDCT32x32_HIGH_PRECISION
#define FDCT32x32_2D vp9_highbd_fdct32x32_sse2
#define FDCT32x32_2D vpx_highbd_fdct32x32_sse2
#define FDCT32x32_HIGH_PRECISION 1
#include "vpx_dsp/x86/fwd_dct32x32_impl_sse2.h" // NOLINT
#undef FDCT32x32_2D

3
vpx_dsp/x86/fwd_txfm_ssse3_x86_64.asm
View File

@ -7,6 +7,9 @@
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
%define program_name vpx
%include "third_party/x86inc/x86inc.asm"
; This file provides SSSE3 version of the forward transformation. Part