nVida O/C tool --- WTF??

[cpt_damage_1]

Hi guys,
First off here are my rig specs:

CPU: AMD 3500 x64
MoBo: AN8SLI - Delux
2 Gig memory
650 watt SLI Psu
x1 XFX 7800GTX (450/1.2)
x1 Inno3D 7800GTX (430/1.2)
Both cards in SLI

Anyway, I installed Coolbits and used it to automatically dectect optimum settings for my GPU's. The settings it came back with were 490/1.3, OK great I though more 'Omph'. Bear in mind that the original clock speeds were 450/1.2 on one card, and 430/1.2 on the other (so quite a big increase).

I ran 3Dmork06 and yeah I got about an 80point increase in my score, hit about 5085 with no artifacts. So then I ran BF2 and it played fine then all of a sudden I got a weird squeeling noise from my headphones and then the sound of my gun firing had dissapeared! Everything else was OK. Restarted the game and it was OK for a few minutes but then WAM it happened again .. no sound for my gun

So then I set my GPUs back to stock, loaded up BF2 and all was fine and dandy again.

Long post I know but does anyone know why coolbits automatically detected that my cards could be run at 490/1.2 when it clearly causes problems? I mean 490, thats an increase of 60mhz on the Inno3d card.

Oh, just incase you are wondering why I have a weired mix of GPU's, its down to a very long RMA procedure and the fact that my GTX broke just as Serious Sam2, COD2, Quake4 and F.E.A.R came out.

 

[Cyber-Mav]

coolbits is not the be all and end all to overclocking. it gives a rough ballpark figure.

also battlefield 2 is not a proper test for the gpu.

you should be using something like ati tool or rthdribl, both will get your gpu and only gpu pumping out max heat and are far better to test for artifacts.

 

[cpt_damage_1]

Hi,
I wasnt using BF2 as a benchmark , just having a few games! I thought that because nVidia design the cards, their software would be the best to determin the optimum clock speeds. Im not really into Ocing but just ran the coolbits to unlock a few bits in the drivers when I cam across the thing that lets nVidia decide the best clock speeds for my cards. And to be honest 3Dmark worked fine ... But BF2 didn't.

 

[Cyber-Mav]

here is some advice, if you really want to test the gfx card for stability then you have to remove the cpu from the equation. the coolbits overclocking is a waste of time sometimes since it does a rough guesstimate on what you can overclock to. some can clock higher than its recommendation and some have to clock lower.

to test the card for artifacts you need to stress the card constantly at 100% for about an hour or so. rthdribl or ati tool and do the job for you since they are the only programs known that can stress a gfx card to 100% and hold that stresspoint till problems appear.

battlefield 2, 3dmark etc are too varied in thier content to provide a constant 100% stress and they usually don;t last as long to give a better idea on reliability.

i say this because i don;t want you to damage your card by pushing it over its limits. i know some people who have overclocked thier cards and tested them in games that are more cpu dependant and not stressing the card to 100% so thier overclocks are much higher.

but those guys were the same ones who came to me when fear came out and put the highest load on their gfx card that they had seen and you can guess what happened then.

 

[Legless]

I find coolbits overclocking (letting it find the fastest speeds) to be very inaccurate, if i let coolbits set the speed, 3dmark artifacts like mad, its much easier to do it manually in rivatuner imho...