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

Merge "Add "vpx_" prefix to 2 idct x86 functions"

Browse Source
This commit is contained in:
Linfeng Zhang 2018-01-23 18:28:59 +00:00 committed by Gerrit Code Review
commit 8fd648c78a
7 changed files with 23 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vp9/common/x86/vp9_highbd_iht8x8_add_sse4.c
View File

@ -204,12 +204,12 @@ void vp9_highbd_iht8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
io_short[7] = _mm_packs_epi32(io[11], io[15]);
if (tx_type == DCT_DCT || tx_type == ADST_DCT) {
idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
} else {
iadst8_sse2(io_short);
}
if (tx_type == DCT_DCT || tx_type == DCT_ADST) {
idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
} else {
iadst8_sse2(io_short);
}
@ -218,8 +218,8 @@ void vp9_highbd_iht8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
__m128i temp[4];
if (tx_type == DCT_DCT || tx_type == ADST_DCT) {
highbd_idct8x8_half1d(io);
highbd_idct8x8_half1d(&io[8]);
vpx_highbd_idct8x8_half1d_sse4_1(io);
vpx_highbd_idct8x8_half1d_sse4_1(&io[8]);
} else {
highbd_iadst8_sse4_1(io);
highbd_iadst8_sse4_1(&io[8]);
@ -235,12 +235,12 @@ void vp9_highbd_iht8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
io[7] = io[11];
if (tx_type == DCT_DCT || tx_type == DCT_ADST) {
highbd_idct8x8_half1d(io);
vpx_highbd_idct8x8_half1d_sse4_1(io);
io[8] = temp[0];
io[9] = temp[1];
io[10] = temp[2];
io[11] = temp[3];
highbd_idct8x8_half1d(&io[8]);
vpx_highbd_idct8x8_half1d_sse4_1(&io[8]);
} else {
highbd_iadst8_sse4_1(io);
io[8] = temp[0];

8
vp9/common/x86/vp9_idct_intrin_sse2.c
View File

@ -68,16 +68,16 @@ void vp9_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int stride,
switch (tx_type) {
case 0: // DCT_DCT
idct8_sse2(in);
idct8_sse2(in);
vpx_idct8_sse2(in);
vpx_idct8_sse2(in);
break;
case 1: // ADST_DCT
idct8_sse2(in);
vpx_idct8_sse2(in);
iadst8_sse2(in);
break;
case 2: // DCT_ADST
iadst8_sse2(in);
idct8_sse2(in);
vpx_idct8_sse2(in);
break;
case 3: // ADST_ADST
iadst8_sse2(in);

4
vpx_dsp/x86/highbd_idct8x8_add_sse2.c
View File

@ -124,8 +124,8 @@ void vpx_highbd_idct8x8_64_add_sse2(const tran_low_t *input, uint16_t *dest,
io_short[6] = _mm_packs_epi32(io[10], io[14]);
io_short[7] = _mm_packs_epi32(io[11], io[15]);
idct8_sse2(io_short);
idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
round_shift_8x8(io_short, io);
} else {
__m128i temp[4];

14
vpx_dsp/x86/highbd_idct8x8_add_sse4.c
View File

@ -17,7 +17,7 @@
#include "vpx_dsp/x86/inv_txfm_ssse3.h"
#include "vpx_dsp/x86/transpose_sse2.h"
void highbd_idct8x8_half1d(__m128i *const io) {
void vpx_highbd_idct8x8_half1d_sse4_1(__m128i *const io) {
__m128i step1[8], step2[8];
transpose_32bit_4x4x2(io, io);
@ -126,13 +126,13 @@ void vpx_highbd_idct8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
io_short[6] = _mm_packs_epi32(io[10], io[14]);
io_short[7] = _mm_packs_epi32(io[11], io[15]);
idct8_sse2(io_short);
idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
vpx_idct8_sse2(io_short);
round_shift_8x8(io_short, io);
} else {
__m128i temp[4];
highbd_idct8x8_half1d(io);
vpx_highbd_idct8x8_half1d_sse4_1(io);
io[8] = _mm_load_si128((const __m128i *)(input + 4 * 8 + 0));
io[12] = _mm_load_si128((const __m128i *)(input + 4 * 8 + 4));
@ -142,7 +142,7 @@ void vpx_highbd_idct8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
io[14] = _mm_load_si128((const __m128i *)(input + 6 * 8 + 4));
io[11] = _mm_load_si128((const __m128i *)(input + 7 * 8 + 0));
io[15] = _mm_load_si128((const __m128i *)(input + 7 * 8 + 4));
highbd_idct8x8_half1d(&io[8]);
vpx_highbd_idct8x8_half1d_sse4_1(&io[8]);
temp[0] = io[4];
temp[1] = io[5];
@ -152,13 +152,13 @@ void vpx_highbd_idct8x8_64_add_sse4_1(const tran_low_t *input, uint16_t *dest,
io[5] = io[9];
io[6] = io[10];
io[7] = io[11];
highbd_idct8x8_half1d(io);
vpx_highbd_idct8x8_half1d_sse4_1(io);
io[8] = temp[0];
io[9] = temp[1];
io[10] = temp[2];
io[11] = temp[3];
highbd_idct8x8_half1d(&io[8]);
vpx_highbd_idct8x8_half1d_sse4_1(&io[8]);
highbd_idct8x8_final_round(io);
}

2
vpx_dsp/x86/highbd_inv_txfm_sse4.h
View File

@ -106,6 +106,6 @@ static INLINE void highbd_idct4_sse4_1(__m128i *const io) {
io[3] = _mm_sub_epi32(step[0], step[3]); // step[0] - step[3]
}
void highbd_idct8x8_half1d(__m128i *const io);
void vpx_highbd_idct8x8_half1d_sse4_1(__m128i *const io);
#endif // VPX_DSP_X86_HIGHBD_INV_TXFM_SSE4_H_

4
vpx_dsp/x86/inv_txfm_sse2.c
View File

@ -165,7 +165,7 @@ void vpx_idct8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest,
// 2-D
for (i = 0; i < 2; i++) {
idct8_sse2(in);
vpx_idct8_sse2(in);
}
write_buffer_8x8(in, dest, stride);
@ -221,7 +221,7 @@ void vpx_idct8x8_1_add_sse2(const tran_low_t *input, uint8_t *dest,
recon_and_store_8_dual(dest, dc_value, stride);
}
void idct8_sse2(__m128i *const in) {
void vpx_idct8_sse2(__m128i *const in) {
// 8x8 Transpose is copied from vpx_fdct8x8_sse2()
transpose_16bit_8x8(in, in);

2
vpx_dsp/x86/inv_txfm_sse2.h
View File

@ -697,7 +697,7 @@ static INLINE void idct32_8x32_quarter_3_4_stage_4_to_7(
}
void idct4_sse2(__m128i *const in);
void idct8_sse2(__m128i *const in);
void vpx_idct8_sse2(__m128i *const in);
void idct16_sse2(__m128i *const in0, __m128i *const in1);
void iadst4_sse2(__m128i *const in);
void iadst8_sse2(__m128i *const in);