diff options
| author | James Almer <jamrial@gmail.com> | 2024-07-10 13:00:20 -0300 |
|---|---|---|
| committer | James Almer <jamrial@gmail.com> | 2024-07-10 13:25:44 -0300 |
| commit | bd1bcb07e0f29c135103a402d71b343a09ad1690 (patch) | |
| tree | f675104705fb8c16b1f6c2dc8833a64d32bf035c /libavutil/x86 | |
| parent | 4a04cca69af807ccf831da977a94350611967c4c (diff) | |
| download | ffmpeg-bd1bcb07e0f29c135103a402d71b343a09ad1690.tar.gz | |
x86/intreadwrite: use intrinsics instead of inline asm for AV_COPY128
This has the benefit of removing any SSE -> AVX penalty that may happen when
the compiler emits VEX encoded instructions.
Signed-off-by: James Almer <jamrial@gmail.com>
Diffstat (limited to 'libavutil/x86')
| -rw-r--r-- | libavutil/x86/intreadwrite.h | 20 |
1 files changed, 7 insertions, 13 deletions
diff --git a/libavutil/x86/intreadwrite.h b/libavutil/x86/intreadwrite.h index 9bbef00dba..6546eb016c 100644 --- a/libavutil/x86/intreadwrite.h +++ b/libavutil/x86/intreadwrite.h @@ -22,29 +22,25 @@ #define AVUTIL_X86_INTREADWRITE_H #include <stdint.h> +#if HAVE_INTRINSICS_SSE +#include <immintrin.h> +#endif #if HAVE_INTRINSICS_SSE2 #include <emmintrin.h> #endif #include "config.h" #include "libavutil/attributes.h" -#if HAVE_MMX - -#ifdef __SSE__ +#if HAVE_INTRINSICS_SSE #define AV_COPY128 AV_COPY128 static av_always_inline void AV_COPY128(void *d, const void *s) { - struct v {uint64_t v[2];}; - - __asm__("movaps %1, %%xmm0 \n\t" - "movaps %%xmm0, %0 \n\t" - : "=m"(*(struct v*)d) - : "m" (*(const struct v*)s) - : "xmm0"); + __m128 tmp = _mm_load_ps(s); + _mm_store_ps(d, tmp); } -#endif /* __SSE__ */ +#endif /* HAVE_INTRINSICS_SSE */ #if HAVE_INTRINSICS_SSE2 @@ -57,6 +53,4 @@ static av_always_inline void AV_ZERO128(void *d) #endif /* HAVE_INTRINSICS_SSE2 */ -#endif /* HAVE_MMX */ - #endif /* AVUTIL_X86_INTREADWRITE_H */ |