Skip to content
This repository has been archived by the owner on Feb 11, 2021. It is now read-only.

Commit

Permalink
Browse files Browse the repository at this point in the history
Okay, I figured out the intrinsics for SIMD memcpy
  • Loading branch information
slouken committed Aug 16, 2007
1 parent a89dec0 commit 4ebb935
Showing 1 changed file with 41 additions and 42 deletions.
83 changes: 41 additions & 42 deletions src/video/SDL_blit_copy.c
Expand Up @@ -24,69 +24,68 @@
#include "SDL_video.h"
#include "SDL_blit.h"

/* The MMX/SSE intrinsics don't give access to specific registers for
the most memory parallelism, so we'll use GCC inline assembly here...
*/
#ifndef __GNUC__
#undef __MMX__
#undef __SSE__
#ifdef __MMX__
#include <mmintrin.h>
#endif
#ifdef __SSE__
#include <xmmintrin.h>
#endif

#ifdef __MMX__
static __inline__ void
SDL_memcpyMMX(Uint8 *dst, const Uint8 *src, int len)
SDL_memcpyMMX(Uint8 * dst, const Uint8 * src, int len)
{
int i;

__m64 values[8];
for (i = len / 64; i--;) {
__asm__ __volatile__ (
"prefetchnta (%0)\n"
"movq (%0), %%mm0\n"
"movq 8(%0), %%mm1\n"
"movq 16(%0), %%mm2\n"
"movq 24(%0), %%mm3\n"
"movq 32(%0), %%mm4\n"
"movq 40(%0), %%mm5\n"
"movq 48(%0), %%mm6\n"
"movq 56(%0), %%mm7\n"
"movntq %%mm0, (%1)\n"
"movntq %%mm1, 8(%1)\n"
"movntq %%mm2, 16(%1)\n"
"movntq %%mm3, 24(%1)\n"
"movntq %%mm4, 32(%1)\n"
"movntq %%mm5, 40(%1)\n"
"movntq %%mm6, 48(%1)\n"
"movntq %%mm7, 56(%1)\n"
:: "r" (src), "r" (dst) : "memory");
_mm_prefetch(src, _MM_HINT_NTA);
values[0] = *(__m64 *) (src + 0);
values[1] = *(__m64 *) (src + 8);
values[2] = *(__m64 *) (src + 16);
values[3] = *(__m64 *) (src + 24);
values[4] = *(__m64 *) (src + 32);
values[5] = *(__m64 *) (src + 40);
values[6] = *(__m64 *) (src + 48);
values[7] = *(__m64 *) (src + 56);
_mm_stream_pi((__m64 *) (dst + 0), values[0]);
_mm_stream_pi((__m64 *) (dst + 8), values[1]);
_mm_stream_pi((__m64 *) (dst + 16), values[2]);
_mm_stream_pi((__m64 *) (dst + 24), values[3]);
_mm_stream_pi((__m64 *) (dst + 32), values[4]);
_mm_stream_pi((__m64 *) (dst + 40), values[5]);
_mm_stream_pi((__m64 *) (dst + 48), values[6]);
_mm_stream_pi((__m64 *) (dst + 56), values[7]);
src += 64;
dst += 64;
}

if (len & 63)
SDL_memcpy(dst, src, len & 63);
}
#endif /* __MMX__ */

#ifdef __SSE__
static __inline__ void
SDL_memcpySSE(Uint8 *dst, const Uint8 *src, int len)
SDL_memcpySSE(Uint8 * dst, const Uint8 * src, int len)
{
int i;

__m128 values[4];
for (i = len / 64; i--;) {
__asm__ __volatile__ (
"prefetchnta (%0)\n"
"movaps (%0), %%xmm0\n"
"movaps 16(%0), %%xmm1\n"
"movaps 32(%0), %%xmm2\n"
"movaps 48(%0), %%xmm3\n"
"movntps %%xmm0, (%1)\n"
"movntps %%xmm1, 16(%1)\n"
"movntps %%xmm2, 32(%1)\n"
"movntps %%xmm3, 48(%1)\n"
:: "r" (src), "r" (dst) : "memory");
_mm_prefetch(src, _MM_HINT_NTA);
values[0] = *(__m128 *) (src + 0);
values[1] = *(__m128 *) (src + 16);
values[2] = *(__m128 *) (src + 32);
values[3] = *(__m128 *) (src + 48);
_mm_stream_ps((float *) (dst + 0), values[0]);
_mm_stream_ps((float *) (dst + 16), values[1]);
_mm_stream_ps((float *) (dst + 32), values[2]);
_mm_stream_ps((float *) (dst + 48), values[3]);
src += 64;
dst += 64;
}

if (len & 63)
SDL_memcpy(dst, src, len & 63);
}
Expand All @@ -107,7 +106,7 @@ SDL_BlitCopy(SDL_BlitInfo * info)
dstskip = w + info->d_skip;

#ifdef __SSE__
if (SDL_HasSSE() && !((uintptr_t)src & 15) && !((uintptr_t)dst & 15)) {
if (SDL_HasSSE() && !((uintptr_t) src & 15) && !((uintptr_t) dst & 15)) {
while (h--) {
SDL_memcpySSE(dst, src, w);
src += srcskip;
Expand All @@ -118,13 +117,13 @@ SDL_BlitCopy(SDL_BlitInfo * info)
#endif

#ifdef __MMX__
if (SDL_HasMMX() && !((uintptr_t)src & 7) && !((uintptr_t)dst & 7)) {
if (SDL_HasMMX() && !((uintptr_t) src & 7) && !((uintptr_t) dst & 7)) {
while (h--) {
SDL_memcpyMMX(dst, src, w);
src += srcskip;
dst += dstskip;
}
__asm__ __volatile__(" emms\n"::);
_mm_empty();
return;
}
#endif
Expand All @@ -148,7 +147,7 @@ SDL_BlitCopyOverlap(SDL_BlitInfo * info)
src = info->s_pixels;
dst = info->d_pixels;
skip = w + info->s_skip;
if ((dst < src) || (dst >= (src + h*skip))) {
if ((dst < src) || (dst >= (src + h * skip))) {
SDL_BlitCopy(info);
} else {
src += ((h - 1) * skip);
Expand Down

0 comments on commit 4ebb935

Please sign in to comment.