View Single Post
Old 11-22-04, 03:06 PM   #43
Registered User
jbirney's Avatar
Join Date: Jul 2002
Posts: 1,430
Default Re: Official HardOCP HL2 benchmarks

Originally Posted by Razor1
What you stated before about the Doom 3 benchs done by [H] both parties were there, both had beta drivers both had equal time to optimize for thier drivers too, these benchmarks are ATi time demos, so they are biased!
I am talking about 2003 (about may) when a few sites got a chance to benchmark Doom3 earliy. NV sposored the event and created custom demos. Catalyst Maker himself stated that ATI new knowthing of the test. Infact there driver was (3.2 or 3.4 forget which) was broke and did not use the 256mb on the 9800Pro's back then.
Anyways.... Doom III.
Interesting little game, even more interesting that reviews would appear on an unreleased game. All I can say at this point is that we have not had that particular benchmark before the review sites started using it. What a shame that people are getting to look at something that we havent had a chance to play around with a bit.
Thus if it was ok for NV to have all the time in the world to work on Doom3 and not ATI last year, turn about is fair play no?

We have differnt timedemos recorded by a few different sites and they all seem to back it up ATIs demo and you still think the are biased?

no I'm not missing your point, I'm trying to figure out why its hurting the gf and not the radeon, Reflect world is your premade cm, Reflect all is real time reflection. It should hurt both cards equally, There really is no extra geomtry calculations just that its a render to texture at every single frame. I might be wrong with the premade CM but even then if the geometry is double it still shouldn't be enough to stress either card, I think its near the million mark when ATI's vertex shader performance will really show through.
Dave over at Beyond3d has a possible reasoning:
R3x0 architectures have a main ALU that contains the DX9 instructions and a secondary ALU that contain PS1.4 modifiers and "some other instructions that we won't say"; it also has a separate texture address precessor so texture instructions/lookups can be interleved with no performance loss (when running numerous other shader instructions). NV40 has "two ALU's" however these do not both contain a full instruction set - its more of a distribution of instructions between the two and there is, IIRC, just one instruction that is in both (MAD or MULL, I forget which); the first ALU also deals with the texture address instructons.
Which seems pausable.
jbirney is offline   Reply With Quote