Coolbits 'Detect Optimal Settings' not reliable?

[TheMightyTen]

I've got a 7800GT and the "Detect Optimal Settings" option in the Clock frequency settings tab consistently detects 481 core and 1.17 memory. If however I run the Mother Nature (I use this test because I have found it one of the best to show artifacts if you've gone to far) test in 3DMark2003 I get artifacts. The best I can get in this test without artifacts is core 470 but I can get the mem to 1.25??

Anybody else found something like this happening?

 

[Emlyn_Dewar]

I get this coolbits auto detect puts my GS up too high, artifacts don't take long to appear. Better off just taking some time and finding the max yourself.

 

[TheMightyTen]

What do other people use to find artifacts, is there a better test than 3DMark03's Mother Nature?

 

[Emlyn_Dewar]

You can use ATI tool on nivdia cards to check for artifacts, I think the beta one also lets you overclock NV cards aswell but I'm not sure.
I use this and a the 3Dmarks to check for artifacts, seems to work alright.

 

[mmj_uk]

What do other people use to find artifacts, is there a better test than 3DMark03's Mother Nature?

something called rthdribl is good and no coolbits auto detect is not reliable, nor is atitool artifact test.

 

[TheMightyTen]

something called rthdribl is good and no coolbits auto detect is not reliable, nor is atitool artifact test.

Cheers I'll give that a go. I do remember playing around with this some time ago but completely forgot about it

 

[GingaNinja]

Oddly, my autodetect tops out at about 470 on the core, but the card runs fine, without artifacts, at 500.

Ginga

 

[oTTo]

My GTX has to run slightly below the auto-detect settings to be stable. I also find that a good run of ATI tool is a decent test for artifacts and stability

 

[aardvark]

yeah - auto detect has always been a bit dodgy, sometimes its over the best setting, sometimes it a lot lower than it can be - its a good place to start though.