flac_wasted_33_c: 214.1
flac_wasted_33_sse4: 133.6
flac_wasted_33_avx2: 93.1
Signed-off-by: James Almer <jamr...@gmail.com>
---
libavcodec/x86/flacdsp.asm | 24 ++++++++++++++++++++++++
libavcodec/x86/flacdsp_init.c | 6 ++++++
2 files changed, 30 insertions(+)
diff --git a/libavcodec/x86/flacdsp.asm b/libavcodec/x86/flacdsp.asm
index 3a940059c7..84cd4dd465 100644
--- a/libavcodec/x86/flacdsp.asm
+++ b/libavcodec/x86/flacdsp.asm
@@ -104,6 +104,30 @@ ALIGN 16
jl .loop
RET
+%macro WASTED_33 1
+cglobal flac_wasted_33, 4,4,2, decoded, residuals, wasted, len
+ shl lend, 2
+ lea decodedq, [decodedq+lenq*2]
+ add residualsq, lenq
+ neg lenq
+ movd xm1, wastedd
+ALIGN 16
+.loop:
+ pmovsxdq m0, [residualsq+lenq]
+ psllq m0, xm1
+ mov%1 [decodedq+lenq*2], m0
+ add lenq, mmsize / 2
+ jl .loop
+ RET
+%endmacro
+
+INIT_XMM sse4
+WASTED_33 a
+%if HAVE_AVX2_EXTERNAL
+INIT_YMM avx2
+WASTED_33 u
+%endif
+
;----------------------------------------------------------------------------------
;void ff_flac_decorrelate_[lrm]s_16_sse2(uint8_t **out, int32_t **in, int
channels,
; int len, int shift);
diff --git a/libavcodec/x86/flacdsp_init.c b/libavcodec/x86/flacdsp_init.c
index 67aa118760..22482f8787 100644
--- a/libavcodec/x86/flacdsp_init.c
+++ b/libavcodec/x86/flacdsp_init.c
@@ -31,6 +31,8 @@ void ff_flac_lpc_32_xop(int32_t *samples, const int
coeffs[32], int order,
int qlevel, int len);
void ff_flac_wasted_32_sse2(int32_t *decoded, int wasted, int len);
+void ff_flac_wasted_33_sse4(int64_t *decoded, const int32_t *residual, int
wasted, int len);
+void ff_flac_wasted_33_avx2(int64_t *decoded, const int32_t *residual, int
wasted, int len);
#define DECORRELATE_FUNCS(fmt, opt)
\
void ff_flac_decorrelate_ls_##fmt##_##opt(uint8_t **out, int32_t **in, int
channels, \
@@ -100,6 +102,7 @@ av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum
AVSampleFormat fmt, int
if (EXTERNAL_SSE4(cpu_flags)) {
c->lpc16 = ff_flac_lpc_16_sse4;
c->lpc32 = ff_flac_lpc_32_sse4;
+ c->wasted33 = ff_flac_wasted_33_sse4;
}
if (EXTERNAL_AVX(cpu_flags)) {
if (fmt == AV_SAMPLE_FMT_S16) {
@@ -117,5 +120,8 @@ av_cold void ff_flacdsp_init_x86(FLACDSPContext *c, enum
AVSampleFormat fmt, int
if (EXTERNAL_XOP(cpu_flags)) {
c->lpc32 = ff_flac_lpc_32_xop;
}
+ if (EXTERNAL_AVX2_FAST(cpu_flags)) {
+ c->wasted33 = ff_flac_wasted_33_avx2;
+ }
#endif /* HAVE_X86ASM */
}
--
2.45.0
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
