2D alpha blending benchmark for SDL 1.2.6?

I see that SDL 1.2.6 touts “MMX and 3DNow! optimized alpha blending.” Can
anyone recommend a benchmark that does a lot of alpha blending? Like a
game, or some such. Something that can be linked against 1.2.6, i.e. an
open source title. If the program isn’t set up to do formal benchmarking,
then an informal benchmark would do, i.e. a program that clearly does tons
of alpha blending.–
Cheers, www.3DProgrammer.com
Brandon Van Every Seattle, WA

“Troll” - (n.) Anything you don’t like.
Usage: “He’s just a troll.”

Hi,

At Wed, 3 Sep 2003 18:14:05 -0700,
Brandon J. Van Every wrote:

I see that SDL 1.2.6 touts “MMX and 3DNow! optimized alpha blending.” Can
anyone recommend a benchmark that does a lot of alpha blending? Like a
game, or some such. Something that can be linked against 1.2.6, i.e. an
open source title. If the program isn’t set up to do formal benchmarking,
then an informal benchmark would do, i.e. a program that clearly does tons
of alpha blending.

shinichiro.h has some benchmark test for check alpha blitting performance.

I made a quick translation of his work:

http://zinnia.dyndns.org/~hiki/SDLKB/?2D+alpha+blending+benchmark

p.s. I posted this mail from non-member address in list.
Sorry, moderator.–

Tsuyoshi Iguchi (zinnia at risky-safety.org)
http://risky-safety.org/zinnia/
http://zinnia.dyndns.org/

Tsuyoshi Iguchi wrote:

Hi,

At Wed, 3 Sep 2003 18:14:05 -0700,
Brandon J. Van Every wrote:

I see that SDL 1.2.6 touts “MMX and 3DNow! optimized alpha blending.” Can
anyone recommend a benchmark that does a lot of alpha blending? Like a
game, or some such. Something that can be linked against 1.2.6, i.e. an
open source title. If the program isn’t set up to do formal benchmarking,
then an informal benchmark would do, i.e. a program that clearly does tons
of alpha blending.

shinichiro.h has some benchmark test for check alpha blitting performance.

I made a quick translation of his work:

http://zinnia.dyndns.org/~hiki/SDLKB/?2D+alpha+blending+benchmark

p.s. I posted this mail from non-member address in list.
Sorry, moderator.

Cool! Seems like a couple of cases with the mmx need fixing.

I think the bench mark misses the per pixel alpha blits?

Sorry I’m not sure how to use sdl from c so I can’t add to the benchmark.

This silly 48 hour game I made uses all 32 bit per pixel alpha blits.
http://www.ludumdare.com/user/214/__bleten.zip

If someone wants they could take the images from there to add to the
bench mark.

You need python and pygame to run it. That’s one of the games I was
testing with.
Pre was 37fps, post was 50.6 fps. The other game got a bit better increase.
50.6342458682 / 37.2473964489 * 100 - 100 = 35.94% increase

Have fun!

Tsuyoshi Iguchi wrote:

Hi,

Hi,

shinichiro.h has some benchmark test for check alpha blitting performance.

I made a quick translation of his work:

http://zinnia.dyndns.org/~hiki/SDLKB/?2D+alpha+blending+benchmark

p.s. I posted this mail from non-member address in list.
Sorry, moderator.

That’s a great benchmark !

I’ll see what I can do about the performance issues, having a reliable
benchmarking tool is of great help :slight_smile:

Stephane

Hi,

At Sun, 07 Sep 2003 18:38:09 +1000,
Rene Dudfield wrote:

I think the bench mark misses the per pixel alpha blits?

Sure, he (shinichiro.h) made another test for RGBA surface.

http://zinnia.dyndns.org/~hiki/SDLKB/?2D+alpha+blending+benchmark
(updateted)

This silly 48 hour game I made uses all 32 bit per pixel alpha blits.
http://www.ludumdare.com/user/214/__bleten.zip

mm… 404-ed… :frowning:

At Tue, 09 Sep 2003 01:05:47 +0200,
Stephane Marchesin wrote:

That’s a great benchmark !

I’ll see what I can do about the performance issues, having a reliable
benchmarking tool is of great help :slight_smile:

Thanks! (and relieved my poor english is readale :^) )-----
Tsuyoshi Iguchi (zinnia at risky-safety.org)
http://risky-safety.org/zinnia/
http://zinnia.dyndns.org/