[FFmpeg-cvslog] x86: hpeldsp: Use EXTERNAL_SSE2_FAST where appropriate
Diego Biurrun
git at videolan.org
Tue Jan 31 19:55:36 EET 2017
ffmpeg | branch: master | Diego Biurrun <diego at biurrun.de> | Mon Jan 11 16:04:17 2016 +0100| [c3e83ad3b7d75f3597f47ada2616ba4479665009] | committer: Diego Biurrun
x86: hpeldsp: Use EXTERNAL_SSE2_FAST where appropriate
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=c3e83ad3b7d75f3597f47ada2616ba4479665009
---
libavcodec/x86/hpeldsp_init.c | 15 ++++++---------
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/libavcodec/x86/hpeldsp_init.c b/libavcodec/x86/hpeldsp_init.c
index 6731428..a7a4515 100644
--- a/libavcodec/x86/hpeldsp_init.c
+++ b/libavcodec/x86/hpeldsp_init.c
@@ -229,15 +229,12 @@ static void hpeldsp_init_3dnow(HpelDSPContext *c, int flags, int cpu_flags)
#endif /* HAVE_AMD3DNOW_EXTERNAL */
}
-static void hpeldsp_init_sse2(HpelDSPContext *c, int flags, int cpu_flags)
+static void hpeldsp_init_sse2_fast(HpelDSPContext *c, int flags, int cpu_flags)
{
#if HAVE_SSE2_EXTERNAL
- if (!(cpu_flags & AV_CPU_FLAG_SSE2SLOW)) {
- // these functions are slower than mmx on AMD, but faster on Intel
- c->put_pixels_tab[0][0] = ff_put_pixels16_sse2;
- c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_sse2;
- c->avg_pixels_tab[0][0] = ff_avg_pixels16_sse2;
- }
+ c->put_pixels_tab[0][0] = ff_put_pixels16_sse2;
+ c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_sse2;
+ c->avg_pixels_tab[0][0] = ff_avg_pixels16_sse2;
#endif /* HAVE_SSE2_EXTERNAL */
}
@@ -254,8 +251,8 @@ av_cold void ff_hpeldsp_init_x86(HpelDSPContext *c, int flags)
if (EXTERNAL_MMXEXT(cpu_flags))
hpeldsp_init_mmxext(c, flags, cpu_flags);
- if (EXTERNAL_SSE2(cpu_flags))
- hpeldsp_init_sse2(c, flags, cpu_flags);
+ if (EXTERNAL_SSE2_FAST(cpu_flags))
+ hpeldsp_init_sse2_fast(c, flags, cpu_flags);
if (CONFIG_VP3_DECODER)
ff_hpeldsp_vp3_init_x86(c, cpu_flags, flags);
More information about the ffmpeg-cvslog
mailing list