Mercurial > sdl-ios-xcode
diff src/video/SDL_blit.h @ 689:5bb080d35049
Date: Tue, 19 Aug 2003 17:57:00 +0200
From: Stephane Marchesin
Subject: Re: [SDL] [patch] MMX alpha blit patches with MMX detection
I think everything is correct now. I've done as much testing as I could,
but some real-world testing wouldn't hurt, I think.
The patch is here : http://icps.u-strasbg.fr/~marchesin/sdl_mmxblit.patch
If you do byte-by-byte comparison of the output between C and MMX
functions, you'll notice that the results for 555 and 565 RGB alpha
blits aren't exactly the same. This is because MMX functions for 555 and
565 RGB have an higher accuracy. If you want the exact same behaviour
that's possible by masking the three lower alpha bits in the MMX
functions. Just ask !
I removed one MMX function because after I fixed it to match its C
equivalent, it revealed to be slower than the C version on a PIII
(although a bit faster on an Athlon XP).
I've also added MMX and PIII replacements for SDL_memcpy. Those provide
some speed up in testvidinfo -benchmark (at least for me, under linux &
X11).
| author | Sam Lantinga <slouken@libsdl.org> |
|---|---|
| date | Fri, 22 Aug 2003 05:51:19 +0000 |
| parents | 417f8709e648 |
| children | b8d311d90021 |
line wrap: on
line diff
--- a/src/video/SDL_blit.h Tue Aug 12 15:17:20 2003 +0000 +++ b/src/video/SDL_blit.h Fri Aug 22 05:51:19 2003 +0000 @@ -410,6 +410,47 @@ } \ } +/* 2 - times unrolled loop */ +#define DUFFS_LOOP_DOUBLE2(pixel_copy_increment, \ + double_pixel_copy_increment, width) \ +{ int n, w = width; \ + if( w & 1 ) { \ + pixel_copy_increment; \ + w--; \ + } \ + if ( w > 0 ) { \ + n = ( w + 2) / 4; \ + switch( w & 2 ) { \ + case 0: do { double_pixel_copy_increment; \ + case 2: double_pixel_copy_increment; \ + } while ( --n > 0 ); \ + } \ + } \ +} + +/* 2 - times unrolled loop 4 pixels */ +#define DUFFS_LOOP_QUATRO2(pixel_copy_increment, \ + double_pixel_copy_increment, \ + quatro_pixel_copy_increment, width) \ +{ int n, w = width; \ + if(w & 1) { \ + pixel_copy_increment; \ + w--; \ + } \ + if(w & 2) { \ + double_pixel_copy_increment; \ + w -= 2; \ + } \ + if ( w > 0 ) { \ + n = ( w + 7 ) / 8; \ + switch( w & 4 ) { \ + case 0: do { quatro_pixel_copy_increment; \ + case 4: quatro_pixel_copy_increment; \ + } while ( --n > 0 ); \ + } \ + } \ +} + /* Use the 8-times version of the loop by default */ #define DUFFS_LOOP(pixel_copy_increment, width) \ DUFFS_LOOP8(pixel_copy_increment, width) @@ -417,6 +458,39 @@ #else /* Don't use Duff's device to unroll loops */ +#define DUFFS_LOOP_DOUBLE2(pixel_copy_increment, \ + double_pixel_copy_increment, width) \ +{ int n = width; \ + if( n & 1 ) { \ + pixel_copy_increment; \ + n--; \ + } \ + n=n>>1; \ + for(; n > 0; --n) { \ + double_pixel_copy_increment; \ + } \ +} + +/* Don't use Duff's device to unroll loops */ +#define DUFFS_LOOP_QUATRO2(pixel_copy_increment, \ + double_pixel_copy_increment, \ + quatro_pixel_copy_increment, width) \ +{ int n = width; \ + if(n & 1) { \ + pixel_copy_increment; \ + n--; \ + } \ + if(n & 2) { \ + double_pixel_copy_increment; \ + n -= 2; \ + } \ + n=n>>2; \ + for(; n > 0; --n) { \ + quatro_pixel_copy_increment; \ + } \ +} + +/* Don't use Duff's device to unroll loops */ #define DUFFS_LOOP(pixel_copy_increment, width) \ { int n; \ for ( n=width; n > 0; --n ) { \