Cat3.5 scores (9700 pro)

[Roger Squires]

Here are my Cat3.5 3dmark scores. They went up since Cat3.4, 50points on
2003 and over 100points on 2001. I'd like to know whether they removed the
benchmark cheats, as the MotherNature score is even higher than with Cat3.4.
http://service.futuremark.com/compare?2k3=946892
http://service.futuremark.com/compare?2k1=6668071

rms

 

[Roger Squires]

I just noticed a questionable result in 3dmark2001SE.

There is a (231fps-141fps)/141fps = 64% gain in the bumpmapping score from
Cat3.4 to Cat3.5. Looks like a cheat to me, but what do I know Can
anyone else confirm this?

rms

 

[Lithurge]

I just noticed a questionable result in 3dmark2001SE.

There is a (231fps-141fps)/141fps = 64% gain in the
bumpmapping score from Cat3.4 to Cat3.5. Looks like a
cheat to me, but what do I know Can anyone else
confirm this?

rms

No idea on that but they have removed one 'optimization'

"The 3DMark2003 shader optimizations found in previous CATALYST™
releases have been removed"

 

[Roger Squires]

Ok, I hear elsewhere that all nforce2 boards had this low Cat3.4 result,
which apparently has been fixed in Cat3.5.

rms

 

[jat]

I want to know if companies I have or am considering giving money to are
lying to me. Here's a good article with an ATI rep's thoughts on benchmark
cheating:
http://slike.slo-tech.com/masses/index.html

rms

I think you're pointless because first off I don't recall where they
ever advertise any benchmarks. Second off, both ATI and NVidia have
plenty of history of inserting code to fudge the benchmarks. So where
you going to run to on that? Matrox Parsmella?