dilim/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2026-08-12 22:16:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

avcodec/x86/vc1dsp_mc: Add size 16 horizontal SSSE3 mc functions

Browse Source 
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_c:             309.1 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_mmxext:        177.3 ( 1.74x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc10_16_ssse3:          52.3 ( 5.91x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_c:             279.6 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_mmxext:        148.8 ( 1.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc20_16_ssse3:          52.1 ( 5.37x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_c:             332.6 ( 1.00x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_mmxext:        177.3 ( 1.88x)
vc1dsp.avg_vc1_mspel_pixels_tab_mc30_16_ssse3:          52.5 ( 6.33x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_c:             288.8 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_mmx:           170.3 ( 1.70x)
vc1dsp.put_vc1_mspel_pixels_tab_mc10_16_ssse3:          51.3 ( 5.63x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_c:             236.2 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_mmx:           144.1 ( 1.64x)
vc1dsp.put_vc1_mspel_pixels_tab_mc20_16_ssse3:          51.3 ( 4.61x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_c:             286.6 ( 1.00x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_mmx:           170.1 ( 1.69x)
vc1dsp.put_vc1_mspel_pixels_tab_mc30_16_ssse3:          51.2 ( 5.60x)

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This commit is contained in:
Andreas Rheinhardt
2026-05-04 20:28:37 +02:00
parent fd711adcb6
commit 3354cbf1b7
2 changed files with 32 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
libavcodec/x86/vc1dsp_init.c
View File

@@ -97,6 +97,9 @@ void ff_vc1_inv_trans_8x8_dc_mmxext(uint8_t *dest, ptrdiff_t linesize,
#define MSPEL_FUNCS_SIZE(X, Y, SIZE, XMM) \
MSPEL_FUNC(put, X, Y, SIZE, XMM); \
MSPEL_FUNC(avg, X, Y, SIZE, XMM)
#define MSPEL_FUNCS(X, Y, XMM) \
MSPEL_FUNCS_SIZE(X, Y, 8, XMM); \
MSPEL_FUNCS_SIZE(X, Y, 16, XMM)
av_cold void ff_vc1dsp_init_x86(VC1DSPContext *dsp)
{
@@ -141,9 +144,9 @@ av_cold void ff_vc1dsp_init_x86(VC1DSPContext *dsp)
dsp->put_no_rnd_vc1_chroma_pixels_tab[0] = ff_put_vc1_chroma_mc8_nornd_ssse3;
dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] = ff_avg_vc1_chroma_mc8_nornd_ssse3;
MSPEL_FUNCS_SIZE(1, 0, 8, ssse3);
MSPEL_FUNCS_SIZE(2, 0, 8, ssse3);
MSPEL_FUNCS_SIZE(3, 0, 8, ssse3);
MSPEL_FUNCS(1, 0, ssse3);
MSPEL_FUNCS(2, 0, ssse3);
MSPEL_FUNCS(3, 0, ssse3);
}
if (EXTERNAL_SSE4(cpu_flags)) {
dsp->vc1_h_loop_filter8 = ff_vc1_h_loop_filter8_sse4;

27
libavcodec/x86/vc1dsp_mc.asm
View File

@@ -200,6 +200,7 @@ HOR_16B_SHIFT2 OP_AVG, avg
%endif ; HAVE_MMX_INLINE
%define MOV8 movq
%define MOV16 movu
INIT_XMM ssse3
%macro HOR_8B 2
@@ -223,7 +224,7 @@ cglobal vc1_%1_mspel_mc30_%2, 4, 4, 6, dst, src, stride, rnd
vc1_%1_mspel_mc30_%2_after_prologue:
movd m0, rndd
WIN64_SPILL_XMM 7
WIN64_SPILL_XMM 7+(%2>>4)
%define hd rndd
mov hd, %2
SPLATW m0, m0
@@ -233,6 +234,7 @@ vc1_%1_mspel_mc30_%2_after_prologue:
MOV%2 m5, [srcq+1]
MOV%2 m6, [srcq+2]
%if %2 == 8
punpcklbw m3, m4
pmaddubsw m3, m1
%ifidn %1,avg
@@ -249,6 +251,26 @@ vc1_%1_mspel_mc30_%2_after_prologue:
pavgb m3, m4
%endif
movq [dstq], m3
%else
SBUTTERFLY bw, 3, 4, 7
pmaddubsw m3, m1
pmaddubsw m4, m1
SBUTTERFLY bw, 6, 5, 7
pmaddubsw m6, m2
pmaddubsw m5, m2
add srcq, strideq
psubw m3, m0
psubw m4, m0
paddw m3, m6
paddw m4, m5
psraw m3, 6
psraw m4, 6
packuswb m3, m4
%ifidn %1, avg
pavgb m3, [dstq]
%endif
mova [dstq], m3
%endif
add dstq, strideq
dec hd
jnz .loop
@@ -257,3 +279,6 @@ vc1_%1_mspel_mc30_%2_after_prologue:
HOR_8B put, 8
HOR_8B avg, 8
HOR_8B put, 16
HOR_8B avg, 16