Traneptora/FFmpeg
1
Fork 0
forked from FFmpeg/FFmpeg
Code Pull requests Activity

avcodec/x86/audiodsp: add scalarproduct avx2

Browse source
This commit is contained in:
Paul B Mahol 2022-09-12 18:53:31 +02:00
parent 72acff9f59
commit 37a503ac87
2 changed files with 24 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
libavcodec/x86/audiodsp.asm
View file

@ -44,6 +44,24 @@ cglobal scalarproduct_int16, 3,3,3, v1, v2, order
movd eax, m2 movd eax, m2
RET RET
%if HAVE_AVX2_EXTERNAL
INIT_YMM avx2
cglobal scalarproduct_int16, 3,3,2, v1, v2, order
add orderd, orderd
add v1q, orderq
add v2q, orderq
neg orderq
pxor m1, m1
.loop:
movu m0, [v1q + orderq]
pmaddwd m0, [v2q + orderq]
paddd m1, m0
add orderq, mmsize
jl .loop
HADDD m1, m0
movd eax, xm1
RET
%endif
;----------------------------------------------------------------------------- ;-----------------------------------------------------------------------------
; void ff_vector_clip_int32(int32_t *dst, const int32_t *src, int32_t min, ; void ff_vector_clip_int32(int32_t *dst, const int32_t *src, int32_t min,

6
libavcodec/x86/audiodsp_init.c
View file

@ -24,6 +24,9 @@
#include "libavutil/x86/cpu.h" #include "libavutil/x86/cpu.h"
#include "libavcodec/audiodsp.h" #include "libavcodec/audiodsp.h"
int32_t ff_scalarproduct_int16_avx2(const int16_t *v1, const int16_t *v2,
int order);
int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2, int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2,
int order); int order);
@ -53,4 +56,7 @@ av_cold void ff_audiodsp_init_x86(AudioDSPContext *c)
if (EXTERNAL_SSE4(cpu_flags)) if (EXTERNAL_SSE4(cpu_flags))
c->vector_clip_int32 = ff_vector_clip_int32_sse4; c->vector_clip_int32 = ff_vector_clip_int32_sse4;
if (EXTERNAL_AVX2_FAST(cpu_flags))
c->scalarproduct_int16 = ff_scalarproduct_int16_avx2;
} }