3DMark03 Performance Examined (aceshardware)

[KeiFront]

http://www.aceshardware.com/read.jsp?id=60000242

Interesting article.

[Byock]

Very interesting. A P2, keeping pace with the P4? Bah!

[Indiana]

Originally posted by Byock
Very interesting. A P2, keeping pace with the P4? Bah!

Did you really look at the configurations? The P4 system is equipped with a Radeon9600 which is WAY, WAY slower than the 9700. The 3DMark03 is - on purpose - nearly only gfx-card limited.
So those results fit quite well, if you think of the test only regarding gfxcard-speed and mostly ignoring CPU speed.

See the other (still flawed, since lots of details about how testing was done ) tests for REAL results.

The whole testing was done to prove the authors (previously made) standpoint that 3DMark03 does not correlate with gaming performance. Thus he obviously tested only the clearly CPU-limited Botmatches in UT2003 (although he doesn't even state this). The fact that he uses a higher resolution (1280x1024)in this in any case CPU-limited scenarios than in other benches (where he used 800x600 to make sure the thing stays CPU limited ) just shows what he wanted the teest-result to show...


Conclusion: IF you want to fake a test so that it proves your previously made up opinion, then make it well and at least with some knowledge - just like that Lars Weinand character of THG is doing it...

[GNEP]

Hang on a minute. When 3dMk2k3 was released, weren't the accompanying documents specifically stating that it was made to be as GPU-limited as possible? Much more so than 3dMk2k1? Or am I missing something here?

[UtwigMU]

P2 350 pwns.

[thop]

I think the results are actually good. Now they only need to drop those stupid CPU and Sound tests!