diff options
| author | aurel <aurel@b3059339-0415-0410-9bf9-f77b7e298cf2> | 2004-10-21 11:55:20 +0000 |
|---|---|---|
| committer | aurel <aurel@b3059339-0415-0410-9bf9-f77b7e298cf2> | 2004-10-21 11:55:20 +0000 |
| commit | 798d2d133780c000166f01fd2d7fbde395196be3 (patch) | |
| tree | 1346451ddb9911c63cd778af21261a6ae6f70346 /libvo/aclib_template.c | |
| parent | 159928a38a3fc65427a1336b2c73aa3f78a82aa7 (diff) | |
| download | mpv-798d2d133780c000166f01fd2d7fbde395196be3.tar.bz2 mpv-798d2d133780c000166f01fd2d7fbde395196be3.tar.xz | |
adapting existing mmx/mmx2/sse/3dnow optimizations so they work on x86_64
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@13721 b3059339-0415-0410-9bf9-f77b7e298cf2
Diffstat (limited to 'libvo/aclib_template.c')
| -rw-r--r-- | libvo/aclib_template.c | 70 |
1 files changed, 35 insertions, 35 deletions
diff --git a/libvo/aclib_template.c b/libvo/aclib_template.c index 54b420eecf..0b50f7ecf3 100644 --- a/libvo/aclib_template.c +++ b/libvo/aclib_template.c @@ -257,62 +257,62 @@ static void * RENAME(fast_memcpy)(void * to, const void * from, size_t len) // Pure Assembly cuz gcc is a bit unpredictable ;) if(i>=BLOCK_SIZE/64) asm volatile( - "xorl %%eax, %%eax \n\t" + "xor %%"REG_a", %%"REG_a" \n\t" ".balign 16 \n\t" "1: \n\t" - "movl (%0, %%eax), %%ebx \n\t" - "movl 32(%0, %%eax), %%ebx \n\t" - "movl 64(%0, %%eax), %%ebx \n\t" - "movl 96(%0, %%eax), %%ebx \n\t" - "addl $128, %%eax \n\t" - "cmpl %3, %%eax \n\t" + "movl (%0, %%"REG_a"), %%ebx \n\t" + "movl 32(%0, %%"REG_a"), %%ebx \n\t" + "movl 64(%0, %%"REG_a"), %%ebx \n\t" + "movl 96(%0, %%"REG_a"), %%ebx \n\t" + "add $128, %%"REG_a" \n\t" + "cmp %3, %%"REG_a" \n\t" " jb 1b \n\t" - "xorl %%eax, %%eax \n\t" + "xor %%"REG_a", %%"REG_a" \n\t" ".balign 16 \n\t" "2: \n\t" - "movq (%0, %%eax), %%mm0\n" - "movq 8(%0, %%eax), %%mm1\n" - "movq 16(%0, %%eax), %%mm2\n" - "movq 24(%0, %%eax), %%mm3\n" - "movq 32(%0, %%eax), %%mm4\n" - "movq 40(%0, %%eax), %%mm5\n" - "movq 48(%0, %%eax), %%mm6\n" - "movq 56(%0, %%eax), %%mm7\n" - MOVNTQ" %%mm0, (%1, %%eax)\n" - MOVNTQ" %%mm1, 8(%1, %%eax)\n" - MOVNTQ" %%mm2, 16(%1, %%eax)\n" - MOVNTQ" %%mm3, 24(%1, %%eax)\n" - MOVNTQ" %%mm4, 32(%1, %%eax)\n" - MOVNTQ" %%mm5, 40(%1, %%eax)\n" - MOVNTQ" %%mm6, 48(%1, %%eax)\n" - MOVNTQ" %%mm7, 56(%1, %%eax)\n" - "addl $64, %%eax \n\t" - "cmpl %3, %%eax \n\t" + "movq (%0, %%"REG_a"), %%mm0\n" + "movq 8(%0, %%"REG_a"), %%mm1\n" + "movq 16(%0, %%"REG_a"), %%mm2\n" + "movq 24(%0, %%"REG_a"), %%mm3\n" + "movq 32(%0, %%"REG_a"), %%mm4\n" + "movq 40(%0, %%"REG_a"), %%mm5\n" + "movq 48(%0, %%"REG_a"), %%mm6\n" + "movq 56(%0, %%"REG_a"), %%mm7\n" + MOVNTQ" %%mm0, (%1, %%"REG_a")\n" + MOVNTQ" %%mm1, 8(%1, %%"REG_a")\n" + MOVNTQ" %%mm2, 16(%1, %%"REG_a")\n" + MOVNTQ" %%mm3, 24(%1, %%"REG_a")\n" + MOVNTQ" %%mm4, 32(%1, %%"REG_a")\n" + MOVNTQ" %%mm5, 40(%1, %%"REG_a")\n" + MOVNTQ" %%mm6, 48(%1, %%"REG_a")\n" + MOVNTQ" %%mm7, 56(%1, %%"REG_a")\n" + "add $64, %%"REG_a" \n\t" + "cmp %3, %%"REG_a" \n\t" "jb 2b \n\t" #if CONFUSION_FACTOR > 0 // a few percent speedup on out of order executing CPUs - "movl %5, %%eax \n\t" + "mov %5, %%"REG_a" \n\t" "2: \n\t" "movl (%0), %%ebx \n\t" "movl (%0), %%ebx \n\t" "movl (%0), %%ebx \n\t" "movl (%0), %%ebx \n\t" - "decl %%eax \n\t" + "dec %%"REG_a" \n\t" " jnz 2b \n\t" #endif - "xorl %%eax, %%eax \n\t" - "addl %3, %0 \n\t" - "addl %3, %1 \n\t" - "subl %4, %2 \n\t" - "cmpl %4, %2 \n\t" + "xor %%"REG_a", %%"REG_a" \n\t" + "add %3, %0 \n\t" + "add %3, %1 \n\t" + "sub %4, %2 \n\t" + "cmp %4, %2 \n\t" " jae 1b \n\t" : "+r" (from), "+r" (to), "+r" (i) - : "r" (BLOCK_SIZE), "i" (BLOCK_SIZE/64), "i" (CONFUSION_FACTOR) - : "%eax", "%ebx" + : "r" ((long)BLOCK_SIZE), "i" (BLOCK_SIZE/64), "i" ((long)CONFUSION_FACTOR) + : "%"REG_a, "%ebx" ); for(; i>0; i--) |