Tom heeft een nieuwe methode gevonden om betrouwbaar met Quake3 te kunnen benchen. Daarvoor gebruikt-ie de IHV versie voor hardware developers:
Sometimes you just have to wait and things get better. I'm not really a believer in those words, but after the nasty Q3test-disaster that struck me some two weeks ago, I may not have been able to do anything against it right away, but I felt that finally my chance would come to rectify things. Eventually last week I somehow got hold of a Q3-version that is supposed to be for IHVs or hardware vendors. The interesting thing about this Q3-version is the fact that you can't really play the game with it, but you can beautifully run benchmarks. Id is obviously supplying this software to hardware developers, so that they can test the performance of their products under Quake 3. Now some or even many of you may wonder, why this version or maybe a mixed playable/benchmarking-version has never been made available to the public. I have got my own ideas why Id wouldn't do this, but maybe you want to ask Id yourself. It could be that Q3test was indeed never meant to be used as a benchmark by anyone. This would be a strong reason
- why there was still no new version released that fixed the broken benchmarking option, and
- why I was bashed badly by Mr. Hook, because I was never supposed to do something with Q3test that wasn't meant to be done.
...begint-ie toch weer te zeiken over Brian Hook.
Anyway, op dit moment heeft Tom allen nog maar benchmarks met de PIII-550, die vind je hier. De TNT2 blijkt duidelijk sneller, waarschijnlijk omdat de Voodoo3 aan z'n fillrate limit begint te komen bij een 550MHz PIII. Op lagere CPU snelheden is de verhouding meestal andersom (V3 iets sneller).