[ut2003io] Is this list archived?

Matthew Arnold marnold at EZ-NET.COM
Sun Sep 15 21:15:46 EDT 2002

Michael wrote:

> Sure, I meant on my system, rather than in general...I suspect given
> your 25fps (I get about 41fps in botmatch in windows) and the  fact
> you're running at 1024x768 show that your configuration probably does
> hit a graphics bottleneck at 1024x768 first, but also that we have very
> different subjective views about what performance is "enough" :o) I
> think my system is too slow for the game at the moment (windows or
> linux), even at 800x600 (that's without taking into account the fps drop
> I expect to get when playing online)

Hehe. Yep, "enough" is always hard to quantify :)

> I get about (botmatch-asbestos)
> 41fps -nosound
> 32fps whatever sound it uses by default
> 34fps with SDL_AUDIODRIVER unset
> 34fps with SDL_AUDIODRIVER=alsa
> 36fps with SDL_AUDIODRIVER=dma
> But the relative closeness of 41 and 32 doesn't do justice to the
> difference in feel, it is subjective, but there's enough jerkiness to
> make it distracting, for me.

32 is a third less than 41, that's pretty significant. Since I had run 
botmatch-citadel before, I decided to try botmatch-asbestos to see how I 
compared to you. The results were very interesting (well, I like 
benchmarks :) ).

38.60 fps with SDL_AUDIODRIVER=dma (that's what I always have it set at)
42.71 fps with -nosound

Now I know our processors are either the same or very close, but your 
GF4-4200 should be quite a bit faster than my GF3-Ti200. Is it just CPU 
bound at that point?

Even more interesting (at least to me) I ran the same botmatch-asbestos 
in Windows. Granted, the demos aren't the same, but you'll get the point.

41.70 fps using Hardware Audio acceleration with EAX
49.32 fps with -nosound

I was very surprised by how close the scores were to my Linux scores 
compared to the ones with botmatch-citadel. I was also surprised at how 
much of a benefit I got from disabling sound in Windows.

Maybe this ties in with what someone posted about having major slowdowns 
on his GF4-4600 when you can see well off into the distance? Asbestos 
doesn't have many wide open spaces. Citadel, on the other hand, is one 
big wide open space. Any thoughts on this, Dan V.?

> Again though, fiddling with graphic-y knobs and dials shows little
> difference, which suggests that they are either all for show ;) or the
> bottleneck when not running flyby isn't purely graphics card / driver
> grunt.

According to ReadMe-Demo.int.txt, "The benchmark will always try to run 
at the highest detail settings." Thus you can tweak the settings until 
your head falls off and it won't change the scores at all.


More information about the ut2003 mailing list