Nexuiz 2.0 benchmark

[Ed]

This benchmark is nonsense. Don't use it!

Most users are probably aware of the benchamrking capability of Nexuiz. For the purposes of friendly competition, thought I'd start this thread for people to post their scores on. It may be useful for people wanting to see what hardware performs well with Nexuiz.

Benchmarking is best done with the highest settings. That way there is less uncertainty in the results. To run the benchmark, do this from the console:

Code: Select all

exec ultimate.cfg
vid_restart
timedemo demos/demo1.dem

Here's my spec:
Athlon XP 2400+ (mildly OC'd at 135MHz FSB)
Shuttle SN25G (nForce 2 Ultra chipset)
2x512Mb GEiL PC3200 DDR RAM
Galaxy Geforce 6600GT 128Mb AGP (stock clocked 525/1050MHz)
Windows 2000 Pro
Forceware 91.28 beta

And the result (found from Nexuiz/data/benchmark.log)
15.6477856 fps min/avg/max: 1.6611296/15.6477856/76.9230769

This will be surpassed quickly!

Bear in mind that this result depends upon these settings being used so don't post results unless they have been obtained by this method.

[esteel]

Woah i hope you know ultimate.cfg was build to kill PCs It uses HDR effects and reliefmapping which are very very advanced features. Maybe the high.cfg or ultra.cfg would be better for a benchmark as these should still be playable on modern PCs

[Zico]

With my A64 3500+ and my 6800U
I only get around 19FPS average.
Compared to your system Ed, it's very bad

System:
SuSE Linux 10.1
A64 3500+ (running 32Bit system)
PoV GF6800U (256MB)
2x 512MB DDR (Dual, Corsair)

Complete result:
ultimate.cfg
1909 frames 99.8046896 seconds 19.1273577 fps
min/avg/max: 1.9351863/19.1273577/155.8620271

[Ed]

Woah i hope you know ultimate.cfg was build to kill PCs It uses HDR effects and reliefmapping which are very very advanced features.

That's the idea. If you use low settings on a benchmark it becomes CPU limited rather than graphics card limited which doesn't make it a very good gaming benchmark.

Any benchmark should be made to stress comtempory PC's. If someone with an Athlon 64 X2 and a Geforce 7900GT runs it then they will get something fairly playable. Someone with an AThlon 64 FX-62 with 2x7900GTX's will get a lot higher benchmark in comparison using ultimate settings than any of the lower settings.

[Willis]

There is a small flaw in the benchmark method in 2.0.

The first time you run a demo, it compiles shaders as you need them, which is kind of a slowdown, the second time it keeps them precompiled and can run considerably faster.

Example:
Run 1: result 1910 frames 42.0570000 seconds 45.4145564 fps min/avg/max: 1.8348624/45.4145564/333.3333333
Run 2: result 1909 frames 38.5820000 seconds 49.4790317 fps min/avg/max: 15.1515152/49.4790317/333.3333333

Notice the huge difference in minimum fps and that 1 less frame has been drawn. In run 1 there are two pauses, one right away, and one on the first fire of the electro. After that, its good. In the second run, there are no pauses.

And just for reference, on a second run at 1920x1200 & ultimate.cfg, I get around 37.7fps average

AMD Opteron 175 @ 2.625Ghz (Dual Core)
2x XFX 7900GTs @ 630Mhz Core/1800Mhz Memory in SLI (Alternate Frame Rendering)
2GB Ram, WinXP x64
Det 91.31's, vsync & thread optimizations off.

[Ed]

Is there any way of pre-compiling shaders before a timedemo? Doom 3 did a similar thing which was fixed with -usecache. Alternatively, is there a way of flushing all shaders compiled on the previous run?

[KadaverJack]

Is there any way of pre-compiling shaders before a timedemo? Doom 3 did a similar thing which was fixed with -usecache.

The best way is to just run the demo twice or play a bit before running the demo. The problem according to LordHavoc is: the engine can't know which shader will be used and since the number of possible shaders has increased by about 20 times (compared to 1.5), precompiling all shaders would take _really_ long...

Alternatively, is there a way of flushing all shaders compiled on the previous run?

Just restart Nexuiz.

[k0jak]

esteel, i need a new high detail config for the new movie, but i need one that my pc will be able to play demos/capture at :p


the one in 1.5 was fine.


which one would you recommend using?

remember i dont have the best pc

[Ed]

I just worked out that this is not suitable for use as a benchmark at all. The .cfg files do not specify resolution so the users existing resolution is used.

Looks like I'll need to produce a proper benchmark.

[Urmel]

The .cfg files do not specify resolution so the users existing resolution is used.

Better is that. Try to run 1900x1440 on an older 17"...