[FFmpeg-devel] [PATCH 8/9] x86: sbrdsp: implement SSE neg_odd_64
Christophe Gisquet
christophe.gisquet at gmail.com
Thu Apr 4 21:45:52 CEST 2013
Timing on Arrandale:
C SSE
Win32: 57 44
Win64: 47 38
Unrolling and not storing mask both save some cycles.
---
libavcodec/x86/sbrdsp.asm | 22 ++++++++++++++++++++++
libavcodec/x86/sbrdsp_init.c | 2 ++
2 files changed, 24 insertions(+)
diff --git a/libavcodec/x86/sbrdsp.asm b/libavcodec/x86/sbrdsp.asm
index fdeaa2b..4e36cec 100644
--- a/libavcodec/x86/sbrdsp.asm
+++ b/libavcodec/x86/sbrdsp.asm
@@ -24,6 +24,7 @@
SECTION_RODATA
; mask equivalent for multiply by -1.0 1.0
ps_mask times 2 dd 1<<31, 0
+ps_mask2 times 2 dd 0, 1<<31
ps_neg times 4 dd 1<<31
ps_noise0 times 2 dd 1.0, 0.0,
ps_noise2 times 2 dd -1.0, 0.0
@@ -401,3 +402,24 @@ apply_noise_main:
add count, 16
jl .loop
RET
+
+INIT_XMM sse
+cglobal sbr_neg_odd_64, 1,2,4,z
+ lea r1q, [zq+256]
+.loop:
+ mova m0, [zq+ 0]
+ mova m1, [zq+16]
+ mova m2, [zq+32]
+ mova m3, [zq+48]
+ xorps m0, [ps_mask2]
+ xorps m1, [ps_mask2]
+ xorps m2, [ps_mask2]
+ xorps m3, [ps_mask2]
+ mova [zq+ 0], m0
+ mova [zq+16], m1
+ mova [zq+32], m2
+ mova [zq+48], m3
+ add zq, 64
+ cmp zq, r1q
+ jne .loop
+ REP_RET
diff --git a/libavcodec/x86/sbrdsp_init.c b/libavcodec/x86/sbrdsp_init.c
index 78c8f4c..9c80897 100644
--- a/libavcodec/x86/sbrdsp_init.c
+++ b/libavcodec/x86/sbrdsp_init.c
@@ -48,6 +48,7 @@ void ff_sbr_hf_apply_noise_2_sse(float (*Y)[2], const float *s_m,
void ff_sbr_hf_apply_noise_3_sse(float (*Y)[2], const float *s_m,
const float *q_filt, int noise,
int kx, int m_max);
+void ff_sbr_neg_odd_64_sse(float *z);
av_cold void ff_sbrdsp_init_x86(SBRDSPContext *s)
{
@@ -66,5 +67,6 @@ av_cold void ff_sbrdsp_init_x86(SBRDSPContext *s)
s->hf_apply_noise[1] = ff_sbr_hf_apply_noise_1_sse;
s->hf_apply_noise[2] = ff_sbr_hf_apply_noise_2_sse;
s->hf_apply_noise[3] = ff_sbr_hf_apply_noise_3_sse;
+ s->neg_odd_64 = ff_sbr_neg_odd_64_sse;
}
}
--
1.8.0.msysgit.0
More information about the ffmpeg-devel
mailing list