[ut2004] ut2004 Benchmark

[Felix Maibaum]

Am Friday 12 March 2004 17:24 schrieb rix:

> Thanks Felix, they were right where you said they would be. Is this the
> standard benchmark command that most gamers are using? I want to know
> because I would like to start comparing how well Linux measures up
> against Windows and I want to make sure I'm comparing apples to apples,
> well except for the OS'es.

I don't know if there is anything you could call the "standard" benchmark for 
2004, yet.
I haven't done much benchmarking myself, but considering what I read on the 
ut2003 list, Windows and linux is always apples and oranges.

First, linux uses the OpenGL renderer, windows the directX one. I've heard the 
directx-version is inherently faster and supports more features. As a result 
of this, you won't have the exact same visual experience in Linux and 
windows, e.g. I think the shadows are missing in Linux.
About using botmatches as a benchmark: I think I remember a thread on the 2003 
list that states differing botmatches for the same command. It seems that 
bots in Linux and Windows do not make the exact same decisions. I don't know 
if that is still true, I don't even know for sure if it ever was, because I 
never tried it.

So if you really want to be on the safe side concerning apples: use flybys, 
and use the openGL renderer in windows (don't ask me how to do that, I don't 
even have windows anymore).
But if you do that you are not comparing real-world performance anymore, so 
you have the choice between a synthetic benchmark that really compares OS and 
driver performance, or you have a real-world comparison between the state of 
native linux and windows gaming, but then you have to note that there are 
visual differences.

Still, both are interesting.

bye

Felix