[FFmpeg-cvslog] swscale/x86/rgb2rgb: add support for AVX
Michael Niedermayer
git at videolan.org
Tue Jan 21 22:09:03 CET 2014
ffmpeg | branch: master | Michael Niedermayer <michaelni at gmx.at> | Tue Nov 19 14:46:01 2013 +0100| [7597e6efe492cb2449bb771054d64cc7fdf62ff5] | committer: Anton Khirnov
swscale/x86/rgb2rgb: add support for AVX
This does not yet include any actual AVX code
Signed-off-by: Anton Khirnov <anton at khirnov.net>
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=7597e6efe492cb2449bb771054d64cc7fdf62ff5
---
libswscale/x86/rgb2rgb.c | 11 +++++++++++
libswscale/x86/rgb2rgb_template.c | 8 ++++----
2 files changed, 15 insertions(+), 4 deletions(-)
diff --git a/libswscale/x86/rgb2rgb.c b/libswscale/x86/rgb2rgb.c
index 98cf1ff..9cfe831 100644
--- a/libswscale/x86/rgb2rgb.c
+++ b/libswscale/x86/rgb2rgb.c
@@ -89,6 +89,7 @@ DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
#define COMPILE_TEMPLATE_MMXEXT 0
#define COMPILE_TEMPLATE_AMD3DNOW 0
#define COMPILE_TEMPLATE_SSE2 0
+#define COMPILE_TEMPLATE_AVX 0
//MMX versions
#undef RENAME
@@ -109,10 +110,18 @@ DECLARE_ASM_CONST(8, uint64_t, blue_15mask) = 0x0000001f0000001fULL;
#define RENAME(a) a ## _sse2
#include "rgb2rgb_template.c"
+//AVX versions
+#undef RENAME
+#undef COMPILE_TEMPLATE_AVX
+#define COMPILE_TEMPLATE_AVX 1
+#define RENAME(a) a ## _avx
+#include "rgb2rgb_template.c"
+
//3DNOW versions
#undef RENAME
#undef COMPILE_TEMPLATE_MMXEXT
#undef COMPILE_TEMPLATE_SSE2
+#undef COMPILE_TEMPLATE_AVX
#undef COMPILE_TEMPLATE_AMD3DNOW
#define COMPILE_TEMPLATE_MMXEXT 0
#define COMPILE_TEMPLATE_SSE2 0
@@ -142,5 +151,7 @@ av_cold void rgb2rgb_init_x86(void)
rgb2rgb_init_mmxext();
if (INLINE_SSE2(cpu_flags))
rgb2rgb_init_sse2();
+ if (INLINE_AVX(cpu_flags))
+ rgb2rgb_init_avx();
#endif /* HAVE_INLINE_ASM */
}
diff --git a/libswscale/x86/rgb2rgb_template.c b/libswscale/x86/rgb2rgb_template.c
index 5aeef8c..c8bbb04 100644
--- a/libswscale/x86/rgb2rgb_template.c
+++ b/libswscale/x86/rgb2rgb_template.c
@@ -1871,7 +1871,7 @@ static inline void RENAME(rgb24toyv12)(const uint8_t *src, uint8_t *ydst, uint8_
}
#endif /* !COMPILE_TEMPLATE_SSE2 */
-#if !COMPILE_TEMPLATE_AMD3DNOW
+#if !COMPILE_TEMPLATE_AMD3DNOW && !COMPILE_TEMPLATE_AVX
static void RENAME(interleaveBytes)(const uint8_t *src1, const uint8_t *src2, uint8_t *dest,
int width, int height, int src1Stride,
int src2Stride, int dstStride)
@@ -1941,7 +1941,7 @@ static void RENAME(interleaveBytes)(const uint8_t *src1, const uint8_t *src2, ui
::: "memory"
);
}
-#endif /* !COMPILE_TEMPLATE_AMD3DNOW */
+#endif /* !COMPILE_TEMPLATE_AMD3DNOW && !COMPILE_TEMPLATE_AVX */
#if !COMPILE_TEMPLATE_SSE2
#if !COMPILE_TEMPLATE_AMD3DNOW
@@ -2509,7 +2509,7 @@ static av_cold void RENAME(rgb2rgb_init)(void)
uyvytoyuv420 = RENAME(uyvytoyuv420);
#endif /* !COMPILE_TEMPLATE_SSE2 */
-#if !COMPILE_TEMPLATE_AMD3DNOW
+#if !COMPILE_TEMPLATE_AMD3DNOW && !COMPILE_TEMPLATE_AVX
interleaveBytes = RENAME(interleaveBytes);
-#endif /* !COMPILE_TEMPLATE_AMD3DNOW */
+#endif /* !COMPILE_TEMPLATE_AMD3DNOW && !COMPILE_TEMPLATE_AVX */
}
More information about the ffmpeg-cvslog
mailing list