qf/quakeforge
0
0
Fork 0
mirror of https://git.code.sf.net/p/quake/quakeforge synced 2024-11-26 22:31:05 +00:00
Code Issues Projects Releases Packages Wiki Activity

[sw] Speed up Draw_FadeScreen

Browse source 
It now processes 4 pixels at a time and uses a bit mask instead of a
conditional to set 3 of the 4 pixels to black. On top of the 4:1 pixel
processing and avoiding inner-loop conditional jumps, gcc unrolls the
loop, so Draw_FadeScreen itself is more than 4x as fast as it was. The
end result is about 5% (3fps) speedup to timedemo demo1 on my 900MHz
EEE Pc when nq has been hacked to always draw the fade-screen.
This commit is contained in:
Bill Currie 2021-07-08 14:18:06 +09:00
parent ef6dd422e5
commit 4b1eb0d760
1 changed files with 9 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
libs/video/renderer/sw/draw.c
View file

@ -777,21 +777,22 @@ void
Draw_FadeScreen (void) Draw_FadeScreen (void)
{ {
int x, y; int x, y;
byte *pbuf; int height = vid.conheight;
int width = vid.conwidth / 4;
uint32_t *pbuf;
VID_UnlockBuffer (); VID_UnlockBuffer ();
S_ExtraUpdate (); S_ExtraUpdate ();
VID_LockBuffer (); VID_LockBuffer ();
for (y = 0; y < vid.conheight; y++) { for (y = 0; y < height; y++) {
unsigned int t; uint32_t mask;
pbuf = ((byte *)vid.buffer) + vid.rowbytes * y; pbuf = (uint32_t *) ((byte *)vid.buffer + vid.rowbytes * y);
t = (y & 1) << 1; mask = 0xff << ((y & 1) << 4);
for (x = 0; x < vid.conwidth; x++) { for (x = 0; x < width; x++) {
if ((x & 3) != t) *pbuf++ &= mask;
pbuf[x] = 0;
} }
} }
vr_data.scr_copyeverything = 1; vr_data.scr_copyeverything = 1;