av_cold void ff_v210enc_init_x86(V210EncContext *s)
{
int cpu_flags = av_get_cpu_flags();
if (EXTERNAL_SSSE3(cpu_flags)) {
s->pack_line_8 = ff_v210_planar_pack_8_ssse3;
s->pack_line_10 = ff_v210_planar_pack_10_ssse3;
}
if (EXTERNAL_AVX(cpu_flags))
s->pack_line_8 = ff_v210_planar_pack_8_avx;
if (EXTERNAL_AVX2(cpu_flags)) {
s->pack_line_8 = ff_v210_planar_pack_8_avx2;
s->pack_line_10 = ff_v210_planar_pack_10_avx2;
s->sample_factor = 2;
}
}
av_cold void INIT_FUNC(VP9DSPContext *dsp)
{
#if HAVE_YASM
int cpu_flags = av_get_cpu_flags();
if (EXTERNAL_SSE2(cpu_flags)) {
init_subpel3(0, put, BPC, sse2);
init_subpel3(1, avg, BPC, sse2);
}
if (EXTERNAL_AVX2(cpu_flags)) {
init_subpel3_32_64(0, put, BPC, avx2);
init_subpel3_32_64(1, avg, BPC, avx2);
init_subpel2(2, 0, 16, put, BPC, avx2);
init_subpel2(2, 1, 16, avg, BPC, avx2);
}
#endif /* HAVE_YASM */
ff_vp9dsp_init_16bpp_x86(dsp);
}
av_cold void ff_vp9dsp_init_x86(VP9DSPContext *dsp)
{
#if HAVE_X86ASM
int cpu_flags = av_get_cpu_flags();
#define init_fpel(idx1, idx2, sz, type, opt) \
dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][0][0] = \
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][0][0] = \
dsp->mc[idx1][FILTER_8TAP_SHARP ][idx2][0][0] = \
dsp->mc[idx1][FILTER_BILINEAR ][idx2][0][0] = ff_vp9_ ## type ## sz ## _ ## opt
#define init_subpel1(idx1, idx2, idxh, idxv, sz, dir, type, opt) \
dsp->mc[idx1][FILTER_8TAP_SMOOTH][idx2][idxh][idxv] = type ## _8tap_smooth_ ## sz ## dir ## _ ## opt; \
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][idxh][idxv] = type ## _8tap_regular_ ## sz ## dir ## _ ## opt; \
dsp->mc[idx1][FILTER_8TAP_SHARP][idx2][idxh][idxv] = type ## _8tap_sharp_ ## sz ## dir ## _ ## opt
#define init_subpel2(idx1, idx2, sz, type, opt) \
init_subpel1(idx1, idx2, 1, 1, sz, hv, type, opt); \
init_subpel1(idx1, idx2, 0, 1, sz, v, type, opt); \
init_subpel1(idx1, idx2, 1, 0, sz, h, type, opt)
#define init_subpel3_32_64(idx, type, opt) \
init_subpel2(0, idx, 64, type, opt); \
init_subpel2(1, idx, 32, type, opt)
#define init_subpel3_8to64(idx, type, opt) \
init_subpel3_32_64(idx, type, opt); \
init_subpel2(2, idx, 16, type, opt); \
init_subpel2(3, idx, 8, type, opt)
#define init_subpel3(idx, type, opt) \
init_subpel3_8to64(idx, type, opt); \
init_subpel2(4, idx, 4, type, opt)
#define init_lpf(opt) do { \
dsp->loop_filter_16[0] = ff_vp9_loop_filter_h_16_16_##opt; \
dsp->loop_filter_16[1] = ff_vp9_loop_filter_v_16_16_##opt; \
dsp->loop_filter_mix2[0][0][0] = ff_vp9_loop_filter_h_44_16_##opt; \
dsp->loop_filter_mix2[0][0][1] = ff_vp9_loop_filter_v_44_16_##opt; \
dsp->loop_filter_mix2[0][1][0] = ff_vp9_loop_filter_h_48_16_##opt; \
dsp->loop_filter_mix2[0][1][1] = ff_vp9_loop_filter_v_48_16_##opt; \
dsp->loop_filter_mix2[1][0][0] = ff_vp9_loop_filter_h_84_16_##opt; \
dsp->loop_filter_mix2[1][0][1] = ff_vp9_loop_filter_v_84_16_##opt; \
dsp->loop_filter_mix2[1][1][0] = ff_vp9_loop_filter_h_88_16_##opt; \
dsp->loop_filter_mix2[1][1][1] = ff_vp9_loop_filter_v_88_16_##opt; \
} while (0)
if (EXTERNAL_MMX(cpu_flags)) {
init_fpel(4, 0, 4, put, mmx);
init_fpel(3, 0, 8, put, mmx);
}
if (EXTERNAL_MMXEXT(cpu_flags)) {
init_subpel2(4, 0, 4, put, mmxext);
init_subpel2(4, 1, 4, avg, mmxext);
init_fpel(4, 1, 4, avg, mmxext);
init_fpel(3, 1, 8, avg, mmxext);
}
if (EXTERNAL_SSE(cpu_flags)) {
init_fpel(2, 0, 16, put, sse);
init_fpel(1, 0, 32, put, sse);
init_fpel(0, 0, 64, put, sse);
}
if (EXTERNAL_SSE2(cpu_flags)) {
init_subpel3_8to64(0, put, sse2);
init_subpel3_8to64(1, avg, sse2);
init_fpel(2, 1, 16, avg, sse2);
init_fpel(1, 1, 32, avg, sse2);
init_fpel(0, 1, 64, avg, sse2);
init_lpf(sse2);
}
if (EXTERNAL_SSSE3(cpu_flags)) {
init_subpel3(0, put, ssse3);
init_subpel3(1, avg, ssse3);
init_lpf(ssse3);
}
if (EXTERNAL_AVX(cpu_flags)) {
init_fpel(1, 0, 32, put, avx);
init_fpel(0, 0, 64, put, avx);
init_lpf(avx);
}
if (EXTERNAL_AVX2(cpu_flags)) {
init_fpel(1, 1, 32, avg, avx2);
init_fpel(0, 1, 64, avg, avx2);
#if ARCH_X86_64 && HAVE_AVX2_EXTERNAL
init_subpel3_32_64(0, put, avx2);
init_subpel3_32_64(1, avg, avx2);
#endif /* ARCH_X86_64 && HAVE_AVX2_EXTERNAL */
}
#undef init_fpel
#undef init_subpel1
#undef init_subpel2
#undef init_subpel3
#endif /* HAVE_X86ASM */
}
av_cold void swri_audio_convert_init_x86(struct AudioConvert *ac,
enum AVSampleFormat out_fmt,
enum AVSampleFormat in_fmt,
int channels){
int mm_flags = av_get_cpu_flags();
ac->simd_f= NULL;
//FIXME add memcpy case
#define MULTI_CAPS_FUNC(flag, cap) \
if (EXTERNAL_##flag(mm_flags)) {\
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16P)\
ac->simd_f = ff_int16_to_int32_a_ ## cap;\
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32P)\
ac->simd_f = ff_int32_to_int16_a_ ## cap;\
}
MULTI_CAPS_FUNC(MMX, mmx)
MULTI_CAPS_FUNC(SSE2, sse2)
if(EXTERNAL_MMX(mm_flags)) {
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_mmx;
}
}
if(EXTERNAL_SSE(mm_flags)) {
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_sse;
}
}
if(EXTERNAL_SSE2(mm_flags)) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_int16_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int16_a_sse2;
if(channels == 2) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_2ch_float_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_2ch_float_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_2ch_float_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_2ch_float_to_int16_a_sse2;
}
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_6ch_float_to_int32_a_sse2;
}
}
if(EXTERNAL_SSSE3(mm_flags)) {
if(channels == 2) {
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int16_a_ssse3;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int32_a_ssse3;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_float_a_ssse3;
}
}
if(EXTERNAL_AVX(mm_flags)) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_int32_to_float_a_avx;
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_int32_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_6ch_float_to_int32_a_avx;
}
}
if(EXTERNAL_AVX2(mm_flags)) {
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int32_a_avx2;
}
}
av_cold void INIT_FUNC(VP9DSPContext *dsp)
{
#if HAVE_YASM
int cpu_flags = av_get_cpu_flags();
#define init_lpf_8_func(idx1, idx2, dir, wd, bpp, opt) \
dsp->loop_filter_8[idx1][idx2] = ff_vp9_loop_filter_##dir##_##wd##_##bpp##_##opt
#define init_lpf_16_func(idx, dir, bpp, opt) \
dsp->loop_filter_16[idx] = loop_filter_##dir##_16_##bpp##_##opt
#define init_lpf_mix2_func(idx1, idx2, idx3, dir, wd1, wd2, bpp, opt) \
dsp->loop_filter_mix2[idx1][idx2][idx3] = loop_filter_##dir##_##wd1##wd2##_##bpp##_##opt
#define init_lpf_funcs(bpp, opt) \
init_lpf_8_func(0, 0, h, 4, bpp, opt); \
init_lpf_8_func(0, 1, v, 4, bpp, opt); \
init_lpf_8_func(1, 0, h, 8, bpp, opt); \
init_lpf_8_func(1, 1, v, 8, bpp, opt); \
init_lpf_8_func(2, 0, h, 16, bpp, opt); \
init_lpf_8_func(2, 1, v, 16, bpp, opt); \
init_lpf_16_func(0, h, bpp, opt); \
init_lpf_16_func(1, v, bpp, opt); \
init_lpf_mix2_func(0, 0, 0, h, 4, 4, bpp, opt); \
init_lpf_mix2_func(0, 1, 0, h, 4, 8, bpp, opt); \
init_lpf_mix2_func(1, 0, 0, h, 8, 4, bpp, opt); \
init_lpf_mix2_func(1, 1, 0, h, 8, 8, bpp, opt); \
init_lpf_mix2_func(0, 0, 1, v, 4, 4, bpp, opt); \
init_lpf_mix2_func(0, 1, 1, v, 4, 8, bpp, opt); \
init_lpf_mix2_func(1, 0, 1, v, 8, 4, bpp, opt); \
init_lpf_mix2_func(1, 1, 1, v, 8, 8, bpp, opt)
if (EXTERNAL_MMXEXT(cpu_flags)) {
init_ipred_func(tm, TM_VP8, 4, BPC, mmxext);
}
if (EXTERNAL_SSE2(cpu_flags)) {
init_subpel3(0, put, BPC, sse2);
init_subpel3(1, avg, BPC, sse2);
init_lpf_funcs(BPC, sse2);
init_8_16_32_ipred_funcs(tm, TM_VP8, BPC, sse2);
}
if (EXTERNAL_SSSE3(cpu_flags)) {
init_lpf_funcs(BPC, ssse3);
}
if (EXTERNAL_AVX(cpu_flags)) {
init_lpf_funcs(BPC, avx);
}
if (EXTERNAL_AVX2(cpu_flags)) {
#if HAVE_AVX2_EXTERNAL
init_subpel3_32_64(0, put, BPC, avx2);
init_subpel3_32_64(1, avg, BPC, avx2);
init_subpel2(2, 0, 16, put, BPC, avx2);
init_subpel2(2, 1, 16, avg, BPC, avx2);
#endif
}
#endif /* HAVE_YASM */
ff_vp9dsp_init_16bpp_x86(dsp);
}
void ff_hevc_dsp_init_x86(HEVCDSPContext *c, const int bit_depth)
{
int cpu_flags = av_get_cpu_flags();
#define SET_LUMA_FUNCS(tabname, funcname, depth, cf) \
c->tabname[0] = funcname ## _4_ ## depth ## _ ## cf; \
c->tabname[1] = funcname ## _8_ ## depth ## _ ## cf; \
c->tabname[2] = funcname ## _12_ ## depth ## _ ## cf; \
c->tabname[3] = funcname ## _16_ ## depth ## _ ## cf; \
c->tabname[4] = funcname ## _24_ ## depth ## _ ## cf; \
c->tabname[5] = funcname ## _32_ ## depth ## _ ## cf; \
c->tabname[6] = funcname ## _48_ ## depth ## _ ## cf; \
c->tabname[7] = funcname ## _64_ ## depth ## _ ## cf;
#define SET_CHROMA_FUNCS(tabname, funcname, depth, cf) \
c->tabname[1] = funcname ## _4_ ## depth ## _ ## cf; \
c->tabname[3] = funcname ## _8_ ## depth ## _ ## cf; \
c->tabname[4] = funcname ## _12_ ## depth ## _ ## cf; \
c->tabname[5] = funcname ## _16_ ## depth ## _ ## cf; \
c->tabname[6] = funcname ## _24_ ## depth ## _ ## cf; \
c->tabname[7] = funcname ## _32_ ## depth ## _ ## cf;
#define SET_QPEL_FUNCS(v, h, depth, cf, name) SET_LUMA_FUNCS (put_hevc_qpel[v][h], name, depth, cf)
#define SET_EPEL_FUNCS(v, h, depth, cf, name) SET_CHROMA_FUNCS(put_hevc_epel[v][h], name, depth, cf)
if (bit_depth == 8) {
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct_4x4_dc_8_mmxext;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_mmxext;
}
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_sse2;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_8_sse2;
c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_sse2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_sse2;
c->idct[0] = ff_hevc_idct_4x4_8_sse2;
c->idct[1] = ff_hevc_idct_8x8_8_sse2;
SET_QPEL_FUNCS(0, 0, 8, sse2, ff_hevc_get_pixels);
SET_EPEL_FUNCS(0, 0, 8, sse2, ff_hevc_get_pixels);
SET_LUMA_FUNCS(put_unweighted_pred, ff_hevc_put_unweighted_pred, 8, sse2);
SET_LUMA_FUNCS(put_unweighted_pred_avg, ff_hevc_put_unweighted_pred_avg, 8, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_chroma, ff_hevc_put_unweighted_pred, 8, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_avg_chroma, ff_hevc_put_unweighted_pred_avg, 8, sse2);
}
if (EXTERNAL_SSSE3(cpu_flags)) {
SET_QPEL_FUNCS(0, 1, 8, ssse3, ff_hevc_qpel_h);
SET_QPEL_FUNCS(1, 0, 8, ssse3, ff_hevc_qpel_v);
SET_EPEL_FUNCS(0, 1, 8, ssse3, ff_hevc_epel_h);
SET_EPEL_FUNCS(1, 0, 8, ssse3, ff_hevc_epel_v);
}
if (EXTERNAL_AVX(cpu_flags)) {
c->idct[0] = ff_hevc_idct_4x4_8_avx;
c->idct[1] = ff_hevc_idct_8x8_8_avx;
}
} else if (bit_depth == 10) {
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct_4x4_dc_10_mmxext;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_mmxext;
}
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_sse2;
c->idct_dc[1] = ff_hevc_idct_8x8_dc_10_sse2;
c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_sse2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_sse2;
c->idct[0] = ff_hevc_idct_4x4_10_sse2;
c->idct[1] = ff_hevc_idct_8x8_10_sse2;
SET_QPEL_FUNCS(0, 0, 10, sse2, ff_hevc_get_pixels);
SET_EPEL_FUNCS(0, 0, 10, sse2, ff_hevc_get_pixels);
SET_LUMA_FUNCS(put_unweighted_pred, ff_hevc_put_unweighted_pred, 10, sse2);
SET_LUMA_FUNCS(put_unweighted_pred_avg, ff_hevc_put_unweighted_pred_avg, 10, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_chroma, ff_hevc_put_unweighted_pred, 10, sse2);
SET_CHROMA_FUNCS(put_unweighted_pred_avg_chroma, ff_hevc_put_unweighted_pred_avg, 10, sse2);
}
if (EXTERNAL_AVX(cpu_flags)) {
c->idct[0] = ff_hevc_idct_4x4_10_avx;
c->idct[1] = ff_hevc_idct_8x8_10_avx;
}
}
#if ARCH_X86_64
if (bit_depth == 8) {
if (EXTERNAL_SSE2(cpu_flags)) {
c->idct[2] = ff_hevc_idct_16x16_8_sse2;
c->idct[3] = ff_hevc_idct_32x32_8_sse2;
}
if (EXTERNAL_SSSE3(cpu_flags)) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_ssse3;
}
if (EXTERNAL_SSE4(cpu_flags)) {
SET_LUMA_FUNCS(weighted_pred, ff_hevc_put_weighted_pred, 8, sse4);
SET_CHROMA_FUNCS(weighted_pred_chroma, ff_hevc_put_weighted_pred, 8, sse4);
SET_LUMA_FUNCS(weighted_pred_avg, ff_hevc_put_weighted_pred_avg, 8, sse4);
SET_CHROMA_FUNCS(weighted_pred_avg_chroma, ff_hevc_put_weighted_pred_avg, 8, sse4);
}
if (EXTERNAL_AVX(cpu_flags)) {
#if HAVE_AVX_EXTERNAL
SET_QPEL_FUNCS(1, 1, 8, avx, hevc_qpel_hv);
SET_EPEL_FUNCS(1, 1, 8, avx, hevc_epel_hv);
#endif /* HAVE_AVX_EXTERNAL */
c->idct[2] = ff_hevc_idct_16x16_8_avx;
c->idct[3] = ff_hevc_idct_32x32_8_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct_16x16_dc_8_avx2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_8_avx2;
}
} else if (bit_depth == 10) {
if (EXTERNAL_SSE2(cpu_flags)) {
c->idct[2] = ff_hevc_idct_16x16_10_sse2;
c->idct[3] = ff_hevc_idct_32x32_10_sse2;
}
if (EXTERNAL_SSSE3(cpu_flags)) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_ssse3;
}
if (EXTERNAL_SSE4(cpu_flags)) {
SET_LUMA_FUNCS(weighted_pred, ff_hevc_put_weighted_pred, 10, sse4);
SET_CHROMA_FUNCS(weighted_pred_chroma, ff_hevc_put_weighted_pred, 10, sse4);
SET_LUMA_FUNCS(weighted_pred_avg, ff_hevc_put_weighted_pred_avg, 10, sse4);
SET_CHROMA_FUNCS(weighted_pred_avg_chroma, ff_hevc_put_weighted_pred_avg, 10, sse4);
}
if (EXTERNAL_AVX(cpu_flags)) {
#if HAVE_AVX_EXTERNAL
SET_QPEL_FUNCS(0, 1, 10, avx, ff_hevc_qpel_h);
SET_QPEL_FUNCS(1, 0, 10, avx, ff_hevc_qpel_v);
SET_QPEL_FUNCS(1, 1, 10, avx, hevc_qpel_hv);
SET_EPEL_FUNCS(0, 1, 10, avx, ff_hevc_epel_h);
SET_EPEL_FUNCS(1, 0, 10, avx, ff_hevc_epel_v);
SET_EPEL_FUNCS(1, 1, 10, avx, hevc_epel_hv);
#endif /* HAVE_AVX_EXTERNAL */
c->idct[2] = ff_hevc_idct_16x16_10_avx;
c->idct[3] = ff_hevc_idct_32x32_10_avx;
}
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct_16x16_dc_10_avx2;
c->idct_dc[3] = ff_hevc_idct_32x32_dc_10_avx2;
}
}
#endif /* ARCH_X86_64 */
}
void ff_hevc_dsp_init_x86(HEVCDSPContext *c, const int bit_depth)
{
int cpu_flags = av_get_cpu_flags();
if (bit_depth == 8) {
#if (HAVE_MMXEXT_EXTERNAL == 1)
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct4x4_dc_8_mmxext;
c->idct_dc[1] = ff_hevc_idct8x8_dc_8_mmxext;
c->transform_add[0] = ff_hevc_transform_add4_8_mmxext;
}
#endif
#if (HAVE_SSE2_EXTERNAL == 1)
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_sse2;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_sse2;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_sse2;
}
c->idct_dc[1] = ff_hevc_idct8x8_dc_8_sse2;
c->idct_dc[2] = ff_hevc_idct16x16_dc_8_sse2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_8_sse2;
c->transform_add[1] = ff_hevc_transform_add8_8_sse2;
c->transform_add[2] = ff_hevc_transform_add16_8_sse2;
c->transform_add[3] = ff_hevc_transform_add32_8_sse2;
}
#endif
#if (HAVE_SSSE3_EXTERNAL == 1)
if (EXTERNAL_SSSE3(cpu_flags) && ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_ssse3;
}
#endif
#if (HAVE_SSE4_EXTERNAL == 1)
if (EXTERNAL_SSE4(cpu_flags) && ARCH_X86_64) {
EPEL_LINKS(c->put_hevc_epel, 0, 0, pel_pixels, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 0, 1, epel_h, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 0, epel_v, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 1, epel_hv, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 0, pel_pixels, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 1, qpel_h, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 0, qpel_v, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 1, qpel_hv, 8, sse4);
}
#endif
#if (HAVE_AVX_EXTERNAL == 1)
if (EXTERNAL_AVX(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_avx;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_avx;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_avx;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_avx;
}
c->transform_add[1] = ff_hevc_transform_add8_8_avx;
c->transform_add[2] = ff_hevc_transform_add16_8_avx;
c->transform_add[3] = ff_hevc_transform_add32_8_avx;
}
#endif
#if (HAVE_AVX2_EXTERNAL == 1)
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct16x16_dc_8_avx2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_8_avx2;
c->transform_add[3] = ff_hevc_transform_add32_8_avx2;
}
#endif
} else if (bit_depth == 10) {
#if (HAVE_MMXEXT_EXTERNAL == 1)
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->transform_add[0] = ff_hevc_transform_add4_10_mmxext;
c->idct_dc[0] = ff_hevc_idct4x4_dc_10_mmxext;
c->idct_dc[1] = ff_hevc_idct8x8_dc_10_mmxext;
}
#endif
#if (HAVE_SSE2_EXTERNAL == 1)
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_sse2;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_sse2;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_sse2;
}
c->idct_dc[1] = ff_hevc_idct8x8_dc_10_sse2;
c->idct_dc[2] = ff_hevc_idct16x16_dc_10_sse2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_10_sse2;
c->transform_add[1] = ff_hevc_transform_add8_10_sse2;
c->transform_add[2] = ff_hevc_transform_add16_10_sse2;
c->transform_add[3] = ff_hevc_transform_add32_10_sse2;
}
#endif
#if (HAVE_SSSE3_EXTERNAL == 1)
if (EXTERNAL_SSSE3(cpu_flags) && ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_ssse3;
}
#endif
#if (HAVE_SSE4_EXTERNAL == 1)
if (EXTERNAL_SSE4(cpu_flags) && ARCH_X86_64) {
EPEL_LINKS(c->put_hevc_epel, 0, 0, pel_pixels, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 0, 1, epel_h, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 0, epel_v, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 1, epel_hv, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 0, pel_pixels, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 1, qpel_h, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 0, qpel_v, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 1, qpel_hv, 10, sse4);
}
#endif
#if (HAVE_AVX_EXTERNAL == 1)
if (EXTERNAL_AVX(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_avx;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_avx;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_avx;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_avx;
}
}
#endif
#if (HAVE_AVX2_EXTERNAL == 1)
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct16x16_dc_10_avx2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_10_avx2;
c->transform_add[2] = ff_hevc_transform_add16_10_avx2;
c->transform_add[3] = ff_hevc_transform_add32_10_avx2;
}
#endif
} else if (bit_depth == 12) {
#if (HAVE_MMXEXT_EXTERNAL == 1)
if (EXTERNAL_MMXEXT(cpu_flags)) {
c->idct_dc[0] = ff_hevc_idct4x4_dc_12_mmxext;
c->idct_dc[1] = ff_hevc_idct8x8_dc_12_mmxext;
}
#endif
#if (HAVE_SSE2_EXTERNAL == 1)
if (EXTERNAL_SSE2(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_12_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_12_sse2;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_12_sse2;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_12_sse2;
}
c->idct_dc[1] = ff_hevc_idct8x8_dc_12_sse2;
c->idct_dc[2] = ff_hevc_idct16x16_dc_12_sse2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_12_sse2;
}
#endif
#if (HAVE_SSSE3_EXTERNAL == 1)
if (EXTERNAL_SSSE3(cpu_flags) && ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_12_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_12_ssse3;
}
#endif
#if (HAVE_SSE4_EXTERNAL == 1)
if (EXTERNAL_SSE4(cpu_flags) && ARCH_X86_64) {
EPEL_LINKS(c->put_hevc_epel, 0, 0, pel_pixels, 12, sse4);
EPEL_LINKS(c->put_hevc_epel, 0, 1, epel_h, 12, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 0, epel_v, 12, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 1, epel_hv, 12, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 0, pel_pixels, 12, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 1, qpel_h, 12, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 0, qpel_v, 12, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 1, qpel_hv, 12, sse4);
}
#endif
#if (HAVE_AVX_EXTERNAL == 1)
if (EXTERNAL_AVX(cpu_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_12_avx;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_12_avx;
if (ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_12_avx;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_12_avx;
}
}
#endif
#if (HAVE_AVX2_EXTERNAL == 1)
if (EXTERNAL_AVX2(cpu_flags)) {
c->idct_dc[2] = ff_hevc_idct16x16_dc_12_avx2;
c->idct_dc[3] = ff_hevc_idct32x32_dc_12_avx2;
}
#endif
}
}
av_cold void ff_vp9dsp_init_x86(VP9DSPContext *dsp)
{
#if HAVE_YASM
int cpu_flags = av_get_cpu_flags();
#define init_fpel(idx1, idx2, sz, type, opt) \
dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][0][0] = \
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][0][0] = \
dsp->mc[idx1][FILTER_8TAP_SHARP ][idx2][0][0] = \
dsp->mc[idx1][FILTER_BILINEAR ][idx2][0][0] = ff_vp9_##type##sz##_##opt
#define init_subpel1(idx1, idx2, idxh, idxv, sz, dir, type, opt) \
dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][idxh][idxv] = type##_8tap_smooth_##sz##dir##_##opt; \
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][idxh][idxv] = type##_8tap_regular_##sz##dir##_##opt; \
dsp->mc[idx1][FILTER_8TAP_SHARP ][idx2][idxh][idxv] = type##_8tap_sharp_##sz##dir##_##opt
#define init_subpel2(idx1, idx2, sz, type, opt) \
init_subpel1(idx1, idx2, 1, 1, sz, hv, type, opt); \
init_subpel1(idx1, idx2, 0, 1, sz, v, type, opt); \
init_subpel1(idx1, idx2, 1, 0, sz, h, type, opt)
#define init_subpel3_32_64(idx, type, opt) \
init_subpel2(0, idx, 64, type, opt); \
init_subpel2(1, idx, 32, type, opt)
#define init_subpel3_8to64(idx, type, opt) \
init_subpel3_32_64(idx, type, opt); \
init_subpel2(2, idx, 16, type, opt); \
init_subpel2(3, idx, 8, type, opt)
#define init_subpel3(idx, type, opt) \
init_subpel3_8to64(idx, type, opt); \
init_subpel2(4, idx, 4, type, opt)
#define init_lpf(opt) do { \
if (ARCH_X86_64) { \
dsp->loop_filter_16[0] = ff_vp9_loop_filter_h_16_16_##opt; \
dsp->loop_filter_16[1] = ff_vp9_loop_filter_v_16_16_##opt; \
dsp->loop_filter_mix2[0][0][0] = ff_vp9_loop_filter_h_44_16_##opt; \
dsp->loop_filter_mix2[0][0][1] = ff_vp9_loop_filter_v_44_16_##opt; \
dsp->loop_filter_mix2[0][1][0] = ff_vp9_loop_filter_h_48_16_##opt; \
dsp->loop_filter_mix2[0][1][1] = ff_vp9_loop_filter_v_48_16_##opt; \
dsp->loop_filter_mix2[1][0][0] = ff_vp9_loop_filter_h_84_16_##opt; \
dsp->loop_filter_mix2[1][0][1] = ff_vp9_loop_filter_v_84_16_##opt; \
dsp->loop_filter_mix2[1][1][0] = ff_vp9_loop_filter_h_88_16_##opt; \
dsp->loop_filter_mix2[1][1][1] = ff_vp9_loop_filter_v_88_16_##opt; \
} \
} while (0)
#define init_ipred(sz, opt, t, e) \
dsp->intra_pred[TX_##sz##X##sz][e##_PRED] = ff_vp9_ipred_##t##_##sz##x##sz##_##opt
#define ff_vp9_ipred_hd_4x4_ssse3 ff_vp9_ipred_hd_4x4_mmxext
#define ff_vp9_ipred_vl_4x4_ssse3 ff_vp9_ipred_vl_4x4_mmxext
#define init_dir_tm_ipred(sz, opt) do { \
init_ipred(sz, opt, dl, DIAG_DOWN_LEFT); \
init_ipred(sz, opt, dr, DIAG_DOWN_RIGHT); \
init_ipred(sz, opt, hd, HOR_DOWN); \
init_ipred(sz, opt, vl, VERT_LEFT); \
init_ipred(sz, opt, hu, HOR_UP); \
init_ipred(sz, opt, tm, TM_VP8); \
init_ipred(sz, opt, vr, VERT_RIGHT); \
} while (0)
#define init_dir_tm_h_ipred(sz, opt) do { \
init_dir_tm_ipred(sz, opt); \
init_ipred(sz, opt, h, HOR); \
} while (0)
#define init_dc_ipred(sz, opt) do { \
init_ipred(sz, opt, dc, DC); \
init_ipred(sz, opt, dc_left, LEFT_DC); \
init_ipred(sz, opt, dc_top, TOP_DC); \
} while (0)
#define init_all_ipred(sz, opt) do { \
init_dc_ipred(sz, opt); \
init_dir_tm_h_ipred(sz, opt); \
} while (0)
if (EXTERNAL_MMX(cpu_flags)) {
init_fpel(4, 0, 4, put, mmx);
init_fpel(3, 0, 8, put, mmx);
dsp->itxfm_add[4 /* lossless */][DCT_DCT] =
dsp->itxfm_add[4 /* lossless */][ADST_DCT] =
dsp->itxfm_add[4 /* lossless */][DCT_ADST] =
dsp->itxfm_add[4 /* lossless */][ADST_ADST] = ff_vp9_iwht_iwht_4x4_add_mmx;
init_ipred(8, mmx, v, VERT);
}
if (EXTERNAL_MMXEXT(cpu_flags)) {
init_subpel2(4, 0, 4, put, mmxext);
init_subpel2(4, 1, 4, avg, mmxext);
init_fpel(4, 1, 4, avg, mmxext);
init_fpel(3, 1, 8, avg, mmxext);
dsp->itxfm_add[TX_4X4][DCT_DCT] = ff_vp9_idct_idct_4x4_add_mmxext;
init_dc_ipred(4, mmxext);
init_dc_ipred(8, mmxext);
init_dir_tm_ipred(4, mmxext);
}
if (EXTERNAL_SSE(cpu_flags)) {
init_fpel(2, 0, 16, put, sse);
init_fpel(1, 0, 32, put, sse);
init_fpel(0, 0, 64, put, sse);
init_ipred(16, sse, v, VERT);
init_ipred(32, sse, v, VERT);
}
if (EXTERNAL_SSE2(cpu_flags)) {
init_subpel3_8to64(0, put, sse2);
init_subpel3_8to64(1, avg, sse2);
init_fpel(2, 1, 16, avg, sse2);
init_fpel(1, 1, 32, avg, sse2);
init_fpel(0, 1, 64, avg, sse2);
init_lpf(sse2);
dsp->itxfm_add[TX_4X4][ADST_DCT] = ff_vp9_idct_iadst_4x4_add_sse2;
dsp->itxfm_add[TX_4X4][DCT_ADST] = ff_vp9_iadst_idct_4x4_add_sse2;
dsp->itxfm_add[TX_4X4][ADST_ADST] = ff_vp9_iadst_iadst_4x4_add_sse2;
dsp->itxfm_add[TX_8X8][DCT_DCT] = ff_vp9_idct_idct_8x8_add_sse2;
dsp->itxfm_add[TX_8X8][ADST_DCT] = ff_vp9_idct_iadst_8x8_add_sse2;
dsp->itxfm_add[TX_8X8][DCT_ADST] = ff_vp9_iadst_idct_8x8_add_sse2;
dsp->itxfm_add[TX_8X8][ADST_ADST] = ff_vp9_iadst_iadst_8x8_add_sse2;
dsp->itxfm_add[TX_16X16][DCT_DCT] = ff_vp9_idct_idct_16x16_add_sse2;
dsp->itxfm_add[TX_16X16][ADST_DCT] = ff_vp9_idct_iadst_16x16_add_sse2;
dsp->itxfm_add[TX_16X16][DCT_ADST] = ff_vp9_iadst_idct_16x16_add_sse2;
dsp->itxfm_add[TX_16X16][ADST_ADST] = ff_vp9_iadst_iadst_16x16_add_sse2;
dsp->itxfm_add[TX_32X32][ADST_ADST] =
dsp->itxfm_add[TX_32X32][ADST_DCT] =
dsp->itxfm_add[TX_32X32][DCT_ADST] =
dsp->itxfm_add[TX_32X32][DCT_DCT] = ff_vp9_idct_idct_32x32_add_sse2;
init_dc_ipred(16, sse2);
init_dc_ipred(32, sse2);
init_dir_tm_h_ipred(8, sse2);
init_dir_tm_h_ipred(16, sse2);
init_dir_tm_h_ipred(32, sse2);
init_ipred(4, sse2, h, HOR);
}
if (EXTERNAL_SSSE3(cpu_flags)) {
init_subpel3(0, put, ssse3);
init_subpel3(1, avg, ssse3);
dsp->itxfm_add[TX_4X4][DCT_DCT] = ff_vp9_idct_idct_4x4_add_ssse3;
dsp->itxfm_add[TX_4X4][ADST_DCT] = ff_vp9_idct_iadst_4x4_add_ssse3;
dsp->itxfm_add[TX_4X4][DCT_ADST] = ff_vp9_iadst_idct_4x4_add_ssse3;
dsp->itxfm_add[TX_4X4][ADST_ADST] = ff_vp9_iadst_iadst_4x4_add_ssse3;
dsp->itxfm_add[TX_8X8][DCT_DCT] = ff_vp9_idct_idct_8x8_add_ssse3;
dsp->itxfm_add[TX_8X8][ADST_DCT] = ff_vp9_idct_iadst_8x8_add_ssse3;
dsp->itxfm_add[TX_8X8][DCT_ADST] = ff_vp9_iadst_idct_8x8_add_ssse3;
dsp->itxfm_add[TX_8X8][ADST_ADST] = ff_vp9_iadst_iadst_8x8_add_ssse3;
dsp->itxfm_add[TX_16X16][DCT_DCT] = ff_vp9_idct_idct_16x16_add_ssse3;
dsp->itxfm_add[TX_16X16][ADST_DCT] = ff_vp9_idct_iadst_16x16_add_ssse3;
dsp->itxfm_add[TX_16X16][DCT_ADST] = ff_vp9_iadst_idct_16x16_add_ssse3;
dsp->itxfm_add[TX_16X16][ADST_ADST] = ff_vp9_iadst_iadst_16x16_add_ssse3;
dsp->itxfm_add[TX_32X32][ADST_ADST] =
dsp->itxfm_add[TX_32X32][ADST_DCT] =
dsp->itxfm_add[TX_32X32][DCT_ADST] =
dsp->itxfm_add[TX_32X32][DCT_DCT] = ff_vp9_idct_idct_32x32_add_ssse3;
init_lpf(ssse3);
init_all_ipred(4, ssse3);
init_all_ipred(8, ssse3);
init_all_ipred(16, ssse3);
init_all_ipred(32, ssse3);
}
if (EXTERNAL_AVX(cpu_flags)) {
dsp->itxfm_add[TX_8X8][DCT_DCT] = ff_vp9_idct_idct_8x8_add_avx;
dsp->itxfm_add[TX_8X8][ADST_DCT] = ff_vp9_idct_iadst_8x8_add_avx;
dsp->itxfm_add[TX_8X8][DCT_ADST] = ff_vp9_iadst_idct_8x8_add_avx;
dsp->itxfm_add[TX_8X8][ADST_ADST] = ff_vp9_iadst_iadst_8x8_add_avx;
dsp->itxfm_add[TX_16X16][DCT_DCT] = ff_vp9_idct_idct_16x16_add_avx;
dsp->itxfm_add[TX_16X16][ADST_DCT] = ff_vp9_idct_iadst_16x16_add_avx;
dsp->itxfm_add[TX_16X16][DCT_ADST] = ff_vp9_iadst_idct_16x16_add_avx;
dsp->itxfm_add[TX_16X16][ADST_ADST] = ff_vp9_iadst_iadst_16x16_add_avx;
dsp->itxfm_add[TX_32X32][ADST_ADST] =
dsp->itxfm_add[TX_32X32][ADST_DCT] =
dsp->itxfm_add[TX_32X32][DCT_ADST] =
dsp->itxfm_add[TX_32X32][DCT_DCT] = ff_vp9_idct_idct_32x32_add_avx;
init_fpel(1, 0, 32, put, avx);
init_fpel(0, 0, 64, put, avx);
init_lpf(avx);
init_dir_tm_h_ipred(8, avx);
init_dir_tm_h_ipred(16, avx);
init_dir_tm_h_ipred(32, avx);
init_ipred(32, avx, v, VERT);
}
if (EXTERNAL_AVX2(cpu_flags)) {
init_fpel(1, 1, 32, avg, avx2);
init_fpel(0, 1, 64, avg, avx2);
if (ARCH_X86_64) {
#if ARCH_X86_64 && HAVE_AVX2_EXTERNAL
init_subpel3_32_64(0, put, avx2);
init_subpel3_32_64(1, avg, avx2);
#endif
}
init_dc_ipred(32, avx2);
init_ipred(32, avx2, h, HOR);
init_ipred(32, avx2, tm, TM_VP8);
}
#undef init_fpel
#undef init_subpel1
#undef init_subpel2
#undef init_subpel3
#endif /* HAVE_YASM */
}
void ff_hevcdsp_init_x86(HEVCDSPContext *c, const int bit_depth)
{
int mm_flags = av_get_cpu_flags();
if (bit_depth == 8) {
if (EXTERNAL_MMXEXT(mm_flags)) {
c->transform_dc_add[0] = ff_hevc_idct4_dc_add_8_mmxext;
c->transform_dc_add[1] = ff_hevc_idct8_dc_add_8_mmxext;
}
if (EXTERNAL_SSE2(mm_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_8_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_8_sse2;
c->transform_dc_add[2] = ff_hevc_idct16_dc_add_8_sse2;
c->transform_dc_add[3] = ff_hevc_idct32_dc_add_8_sse2;
}
if (EXTERNAL_SSSE3(mm_flags) && ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_8_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_8_ssse3;
}
if (EXTERNAL_SSE4(mm_flags) && ARCH_X86_64) {
EPEL_LINKS(c->put_hevc_epel, 0, 0, pel_pixels, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 0, 1, epel_h, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 0, epel_v, 8, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 1, epel_hv, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 0, pel_pixels, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 1, qpel_h, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 0, qpel_v, 8, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 1, qpel_hv, 8, sse4);
}
if (EXTERNAL_AVX2(mm_flags)) {
c->transform_dc_add[3] = ff_hevc_idct32_dc_add_8_avx2;
}
} else if (bit_depth == 10) {
if (EXTERNAL_MMXEXT(mm_flags)) {
c->transform_dc_add[0] = ff_hevc_idct4_dc_add_10_mmxext;
}
if (EXTERNAL_SSE2(mm_flags)) {
c->hevc_v_loop_filter_chroma = ff_hevc_v_loop_filter_chroma_10_sse2;
c->hevc_h_loop_filter_chroma = ff_hevc_h_loop_filter_chroma_10_sse2;
c->transform_dc_add[1] = ff_hevc_idct8_dc_add_10_sse2;
c->transform_dc_add[2] = ff_hevc_idct16_dc_add_10_sse2;
c->transform_dc_add[3] = ff_hevc_idct32_dc_add_10_sse2;
}
if (EXTERNAL_SSSE3(mm_flags) && ARCH_X86_64) {
c->hevc_v_loop_filter_luma = ff_hevc_v_loop_filter_luma_10_ssse3;
c->hevc_h_loop_filter_luma = ff_hevc_h_loop_filter_luma_10_ssse3;
}
if (EXTERNAL_SSE4(mm_flags) && ARCH_X86_64) {
EPEL_LINKS(c->put_hevc_epel, 0, 0, pel_pixels, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 0, 1, epel_h, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 0, epel_v, 10, sse4);
EPEL_LINKS(c->put_hevc_epel, 1, 1, epel_hv, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 0, pel_pixels, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 0, 1, qpel_h, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 0, qpel_v, 10, sse4);
QPEL_LINKS(c->put_hevc_qpel, 1, 1, qpel_hv, 10, sse4);
}
if (EXTERNAL_AVX(mm_flags)) {
c->transform_dc_add[1] = ff_hevc_idct8_dc_add_10_avx;
c->transform_dc_add[2] = ff_hevc_idct16_dc_add_10_avx;
c->transform_dc_add[3] = ff_hevc_idct32_dc_add_10_avx;
}
if (EXTERNAL_AVX2(mm_flags)) {
c->transform_dc_add[2] = ff_hevc_idct16_dc_add_10_avx2;
c->transform_dc_add[3] = ff_hevc_idct32_dc_add_10_avx2;
}
}
}
