* [PATCH]: More optimization for accel_putcs()
@ 2003-02-22 3:32 Antonino Daplas
2003-02-23 11:07 ` Antonino Daplas
0 siblings, 1 reply; 3+ messages in thread
From: Antonino Daplas @ 2003-02-22 3:32 UTC (permalink / raw)
To: James Simmons, Geert Uytterhoeven; +Cc: Linux Fbdev development list
[-- Attachment #1: Type: text/plain, Size: 1109 bytes --]
Geert, James,
Here's a patch for linux-2.5.61 + James' fbdev.diff so accel_putcs()
will do only 1 fb_imageblit() at the end when fontwidth is not divisible
by 8. Tested on 4x6 and 12x22 fonts. There should be a significant
performance increase even with the generic functions, however, the
greatest gain can be seen on drivers with accelerated fb_imageblit().
Here are some numbers:
no accel
scrollmode: yredraw
font: 12x22
visual: packed pixels
time cat /usr/src/linux/MAINTAINERS
linux-2.5.62 (old -- 1 fb_imageblit/character)
bpp8
----
real 0m5.247s
user 0m0.001s
sys 0m5.245s
bpp16
-----
real 0m9.640s
user 0m0.001s
sys 0m9.591s
bpp24
-----
real 0m15.943s
user 0m0.001s
sys 0m15.944s
bpp32
-----
real 0m19.653s
user 0m0.002s
sys 0m19.651s
linux-2.5.62 (new - 1 fb_imageblit at the end)
bpp8
----
real 0m3.867s
user 0m0.000s
sys 0m3.866s
bpp16
-----
real 0m5.894s
user 0m0.001s
sys 0m5.892s
bpp24
-----
real 0m13.669s
user 0m0.000s
sys 0m13.670s
bpp32
-----
real 0m11.053s
user 0m0.001s
sys 0m11.054s
Tony
[-- Attachment #2: accel_putcs.diff --]
[-- Type: text/x-patch, Size: 3444 bytes --]
diff -Naur linux-2.5.61-fbdev/drivers/video/console/fbcon.c linux-2.5.61/drivers/video/console/fbcon.c
--- linux-2.5.61-fbdev/drivers/video/console/fbcon.c 2003-02-22 02:34:26.000000000 +0000
+++ linux-2.5.61/drivers/video/console/fbcon.c 2003-02-22 03:16:45.000000000 +0000
@@ -332,16 +332,22 @@
}
#define FB_PIXMAPSIZE 8192
+/*
+ * FIXME: Break up this function, it's becoming too long...
+ */
void accel_putcs(struct vc_data *vc, struct display *p,
const unsigned short *s, int count, int yy, int xx)
{
+ static u8 pixmap[FB_PIXMAPSIZE];
+ struct fb_image image;
struct fb_info *info = p->fb_info;
unsigned short charmask = p->charmask;
unsigned int width = ((vc->vc_font.width + 7)/8);
unsigned int cellsize = vc->vc_font.height * width;
- struct fb_image image;
+ unsigned int pitch, cnt, i, j, k;
+ unsigned int maxcnt = FB_PIXMAPSIZE/cellsize;
+ u8 *src, *dst, *dst0;
u16 c = scr_readw(s);
- static u8 pixmap[FB_PIXMAPSIZE];
image.fg_color = attr_fgcol(p, c);
image.bg_color = attr_bgcol(p, c);
@@ -349,13 +355,9 @@
image.dy = yy * vc->vc_font.height;
image.height = vc->vc_font.height;
image.depth = 0;
+ image.data = pixmap;
if (!(vc->vc_font.width & 7)) {
- unsigned int pitch, cnt, i, j, k;
- unsigned int maxcnt = FB_PIXMAPSIZE/cellsize;
- char *src, *dst, *dst0;
-
- image.data = pixmap;
while (count) {
if (count > maxcnt)
cnt = k = maxcnt;
@@ -381,14 +383,48 @@
image.dx += cnt * vc->vc_font.width;
count -= cnt;
}
+
} else {
- image.width = vc->vc_font.width;
- while (count--) {
- image.data = p->fontdata +
- (scr_readw(s++) & charmask) * cellsize;
+ unsigned int shift_low = 0, mod = vc->vc_font.width % 8;
+ unsigned int shift_high = 8;
+ unsigned idx = vc->vc_font.width/8;
+ u8 mask;
+
+ while (count) {
+ if (count > maxcnt)
+ cnt = k = maxcnt;
+ else
+ cnt = k = count;
+
+ dst0 = pixmap;
+ image.width = vc->vc_font.width * cnt;
+ pitch = (image.width + 7)/8;
+ while (k--) {
+ src = p->fontdata + (scr_readw(s++)&charmask)*
+ cellsize;
+ dst = dst0;
+ mask = (u8) (0xfff << shift_high);
+ for (i = image.height; i--; ) {
+ for (j = 0; j < idx; j++) {
+ dst[j] &= mask;
+ dst[j] |= *src >> shift_low;
+ dst[j+1] = *src << shift_high;
+ src++;
+ }
+ dst[idx] &= mask;
+ dst[idx] |= *src++ >> shift_low;
+ dst += pitch;
+ }
+ shift_low += mod;
+ shift_low &= 7;
+ shift_high = 8 - shift_low;
+ dst0 += (shift_low) ? width - 1 : width;
+ }
+
info->fbops->fb_imageblit(info, &image);
- image.dx += vc->vc_font.width;
- }
+ image.dx += cnt * vc->vc_font.width;
+ count -= cnt;
+ }
}
}
diff -Naur linux-2.5.61-fbdev/drivers/video/fbmem.c linux-2.5.61/drivers/video/fbmem.c
--- linux-2.5.61-fbdev/drivers/video/fbmem.c 2003-02-22 02:34:26.000000000 +0000
+++ linux-2.5.61/drivers/video/fbmem.c 2003-02-22 02:42:06.000000000 +0000
@@ -363,14 +363,14 @@
static int ofonly __initdata = 0;
#endif
+#ifdef CONFIG_LOGO
+#include <linux/linux_logo.h>
+
static inline unsigned safe_shift(unsigned d, int n)
{
return n < 0 ? d >> -n : d << n;
}
-#ifdef CONFIG_FB_LOGO
-#include <linux/linux_logo.h>
-
static void __init fb_set_logocmap(struct fb_info *info,
const struct linux_logo *logo)
{
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH]: More optimization for accel_putcs()
2003-02-22 3:32 [PATCH]: More optimization for accel_putcs() Antonino Daplas
@ 2003-02-23 11:07 ` Antonino Daplas
2003-02-26 19:29 ` James Simmons
0 siblings, 1 reply; 3+ messages in thread
From: Antonino Daplas @ 2003-02-23 11:07 UTC (permalink / raw)
To: Antonino Daplas
Cc: James Simmons, Geert Uytterhoeven, Linux Fbdev development list
On Sat, 2003-02-22 at 11:32, Antonino Daplas wrote:
> Geert, James,
>
> Here's a patch for linux-2.5.61 + James' fbdev.diff so accel_putcs()
> will do only 1 fb_imageblit() at the end when fontwidth is not divisible
> by 8. Tested on 4x6 and 12x22 fonts. There should be a significant
> performance increase even with the generic functions, however, the
> greatest gain can be seen on drivers with accelerated fb_imageblit().
> Here are some numbers:
>
The optimization logic is flawed :-( Please apply this patch also.
Tony
diff -Naur linux-2.5.61-fbdev/drivers/video/console/fbcon.c linux-2.5.61/drivers/video/console/fbcon.c
--- linux-2.5.61-fbdev/drivers/video/console/fbcon.c 2003-02-23 18:54:32.000000000 +0800
+++ linux-2.5.61/drivers/video/console/fbcon.c 2003-02-23 18:58:05.000000000 +0800
@@ -400,7 +400,7 @@
image.width = vc->vc_font.width * cnt;
pitch = (image.width + 7)/8;
while (k--) {
- src = p->fontdata + (scr_readw(s++)&charmask)*
+ src = p->fontdata + (scr_readw(s++)&charmask)*
cellsize;
dst = dst0;
mask = (u8) (0xfff << shift_high);
@@ -412,15 +412,18 @@
src++;
}
dst[idx] &= mask;
- dst[idx] |= *src++ >> shift_low;
+ dst[idx] |= *src >> shift_low;
+ if (shift_high < mod)
+ dst[idx+1] = *src << shift_high;
+ src++;
dst += pitch;
}
shift_low += mod;
+ dst0 += (shift_low >= 8) ? width : width - 1;
shift_low &= 7;
shift_high = 8 - shift_low;
- dst0 += (shift_low) ? width - 1 : width;
}
-
+
info->fbops->fb_imageblit(info, &image);
image.dx += cnt * vc->vc_font.width;
count -= cnt;
-------------------------------------------------------
This SF.net email is sponsored by: SlickEdit Inc. Develop an edge.
The most comprehensive and flexible code editor you can use.
Code faster. C/C++, C#, Java, HTML, XML, many more. FREE 30-Day Trial.
www.slickedit.com/sourceforge
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH]: More optimization for accel_putcs()
2003-02-23 11:07 ` Antonino Daplas
@ 2003-02-26 19:29 ` James Simmons
0 siblings, 0 replies; 3+ messages in thread
From: James Simmons @ 2003-02-26 19:29 UTC (permalink / raw)
To: Antonino Daplas; +Cc: Geert Uytterhoeven, Linux Fbdev development list
Applied.
-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-02-26 19:30 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-02-22 3:32 [PATCH]: More optimization for accel_putcs() Antonino Daplas
2003-02-23 11:07 ` Antonino Daplas
2003-02-26 19:29 ` James Simmons
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).