View Single Post
Old 04-02-03, 09:42 AM   #87
Uttar
Registered User
 
Uttar's Avatar
 
Join Date: Aug 2002
Posts: 1,354
Send a message via AIM to Uttar Send a message via Yahoo to Uttar
Default

Quote:
Originally posted by mongoled
Please get back to topic I was enjoying the points Uttar was trying to put across and some of the comments received by peeps here.
I already know what card is deemed the best buy right NOW irrespective of programabilty, I want to listen to Uttar's theory, I hope this thread gets back on track.
Eh, thanks

Anyway, to resummarize my points...

- nVidia is NOT violating the DX9 spec by using INT12 for Game 1, Game 2 and Game 3. This is perfectly legal, AFAIK.
- nVidia *is* violating the DX9 spec by forcing FP16. And thus Game 4 on 43.45 is not representative of the performance of legal drivers.
- nVidia's 3DMark 2003 optimizations for Game 1/2/3 are nothing illegal - both ATI & nVidia did such optimizations for 3DMark 2001 in the past and both had WHQL status. I'm sure ATI is likely optimizing for 3DMark 2003 a little too.
- The 3DMark 2003 score is calculated using a known formula.

Conclusion:

It is possible to use the 43.45 drivers to benchmark Game 1, Game 2 and Game 3, while using 43.03 ( or another driver set, that'd have to be determined ) to benchmark Game 4. Then you'd manually compute the score.

That would give a much fairer score, IMO.

As for the whole NV30 vs R300 vs R350 thing...
IMO, the NV30 is superior to both R300 & R350 in texturing and Vertex Shading. The NV30 is howeve badly inferior in Pixel Shading: while it is more flexible, it's also a LOT slower unless you use INT12 ( and you can't even use INT12 in a PS2.0. D3D program )


Uttar
Uttar is offline   Reply With Quote