[FFmpeg-cvslog] swscale/x86/rgb2rgb: disable rgb24toyv12_mmxext for x86_64
Ramiro Polla
git at videolan.org
Sat Sep 7 00:14:55 EEST 2024
ffmpeg | branch: master | Ramiro Polla <ramiro.polla at gmail.com> | Wed Aug 28 20:03:39 2024 +0200| [caaec2ea957290941eecfe5d87baf5c0a500b450] | committer: Ramiro Polla
swscale/x86/rgb2rgb: disable rgb24toyv12_mmxext for x86_64
The mmxext implementation is slower than the C version in x86_64.
m32 m64
rgb24toyv12_16_200_c: 24942.7 14812.6
rgb24toyv12_16_200_mmxext: 17857.2 ( 1.40x) 17400.4 ( 0.85x)
rgb24toyv12_128_60_c: 56892.9 35616.9
rgb24toyv12_128_60_mmxext: 40730.9 ( 1.40x) 39610.4 ( 0.90x)
rgb24toyv12_512_16_c: 58402.7 37209.4
rgb24toyv12_512_16_mmxext: 44842.4 ( 1.30x) 41136.2 ( 0.90x)
rgb24toyv12_1920_4_c: 54827.4 34737.4
rgb24toyv12_1920_4_mmxext: 51169.9 ( 1.07x) 34818.9 ( 1.00x)
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=caaec2ea957290941eecfe5d87baf5c0a500b450
---
libswscale/x86/rgb2rgb.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/libswscale/x86/rgb2rgb.c b/libswscale/x86/rgb2rgb.c
index 1617c33338..4943023342 100644
--- a/libswscale/x86/rgb2rgb.c
+++ b/libswscale/x86/rgb2rgb.c
@@ -1480,7 +1480,7 @@ static inline void planar2x_mmxext(const uint8_t *src, uint8_t *dst, int srcWidt
* others are ignored in the C version.
* FIXME: Write HQ version.
*/
-#if HAVE_7REGS
+#if ARCH_X86_32 && HAVE_7REGS
static inline void rgb24toyv12_mmxext(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
int width, int height,
int lumStride, int chromStride, int srcStride,
@@ -2257,9 +2257,9 @@ static av_cold void rgb2rgb_init_mmxext(void)
yuyvtoyuv422 = yuyvtoyuv422_mmxext;
planar2x = planar2x_mmxext;
-#if HAVE_7REGS
+#if ARCH_X86_32 && HAVE_7REGS
ff_rgb24toyv12 = rgb24toyv12_mmxext;
-#endif /* HAVE_7REGS */
+#endif /* ARCH_X86_32 && HAVE_7REGS */
yuyvtoyuv420 = yuyvtoyuv420_mmxext;
uyvytoyuv420 = uyvytoyuv420_mmxext;
More information about the ffmpeg-cvslog
mailing list