HD2900XT Crysis Artifacting

Ibanezrg570

Distinguished
Aug 23, 2007
94
0
18,630
Well this is a strange one that I would like some input on,

It seems that Crysis will not let me OC my HD2900XT without it artifacting when the card reaches above 75C. That being said, 75C is not a hot temp for this card and I easily run other games well into the 80-82C with no such artifacts what so ever.

If I return my card to stock speeds, the artifacting ceases. I did recieve some silly Microsft .Net Framework error at one point though.

So my question to you is, how can one game artifact my card yet everything else (such as repeated 3DMark06 runs, and hours of play in HL2-EP2 and TF2) runs perfectly smooth with the OC? Could this be some silly driver issue at this point?

Any input would be appreciated,

Cheers
 

NightlySputnik

Distinguished
Mar 3, 2006
638
0
18,980
Doom 3 did exactly that to a lot of games when it first came out. VPU that would work overclock without any problems would start having artefact showing up. I know, it happenned to my (back then) Radeon 9800 Pro.

I can't explain why with certainty tough. It seems that this happen with new high performances game engine. Maybe it has to do with part of the VPU previously unused now being force into submission if you see what I mean.

My 2 cents!
 

Sengoku

Distinguished
Feb 13, 2007
329
0
18,780
If I were to take a wild ass guess at this, I would guess that the artifactiting might be the result of a race condition. Some elements of the crysis code you're looking at are not well optimized.

Lets see what happens with more mature code before we jump to any conclusions though.
 

Ibanezrg570

Distinguished
Aug 23, 2007
94
0
18,630
Just an update upon further testing,

This artifacting is definitely not heat dependent. I incrementally changed the clock speeds on the card until I found a setting at which no artifacts were present.

Anything above 808 MHz GPU clock will cause artifacts immediately. Memory clock seems to not really effect the artifacting in any way but I find it most stable at 888 MHz.

I'm going to chalk this up to some incompatible code with my system at this point. Will see what the next catalysts will do.
 

TRENDING THREADS