Sign in with
Sign up | Sign in
Your question

Stress test, nature of SLi readings

Last response: in Graphics & Displays
Share
December 23, 2007 2:22:21 AM

Wow, I posted about 3 hours ago on another topic and I already recieved two responses, I dunno why I didn't come to the infamous Tom's HArdware earlier...

I was running two 6800 Nvidias under the control panel Performance test simply to figure out if one of my GPU's is slacking, or a little damaged. This stems from me noticing at times a 25 degree temp differential from GPU 1 to GPU 2 . Anyways, upon monitoring my temps thru Nmonitor, I saw that my GPu 1 was running at a 71 degree (C) steady, while GPU 2 was leaping from 58 degrees or so to 98 degrees (c) or so. I plan to switch my Gpu's around and do the same test, however, My question revolves around the temp differential between GPU's. Is it normal for the GPu's to differ 20 degrees or so in temp? If they do, is damage likely a cause?

Thanks
December 23, 2007 2:49:02 AM

UPDATE: I switched my GPU's and ran the stress test again. Instead of Seeing GPU 1 stable at 70 degrees while GPU 2 does all the fluxuating, NOW GPU 1 remains at a cool 50 degrees stable and GPU 2 still fluxuates about 10 degrees... I am completely confused now because it seems that the difference in stable GPU 1 readings would indicate a damaged card, but then the constant 10 degree flux between 55 and 65 degrees on GPU 2 seems to not indicate a damaged card at all. Anyone with more experience able to shed a bit o light on this research which has me searching for the answer??????
THX
December 23, 2007 3:17:22 AM

do you have good airflow in ur case? also if u havent done so already give them a good shot of air duster.. might help out alittle..
Related resources
December 23, 2007 4:10:36 AM

i know when i had my 7900gtx's the top card would have a higher temp probably from the card underneath it but they never fluctuated that much until one of them died....repeatedly..lol 5 dead in 3 months XFX non-XXX versions. sorry can't be more help. btw when one of my cards was about to die it had the typical random glitches and artifacting on screen.
December 23, 2007 7:11:56 AM

Watch the temperature on the bottom card, especially if it's very close to the bottom (or side) of the case. It's most likely inconsistent air flow for the card in that position.

Try the one card on its own (remove the other) and stress test it. Then try the other. You should not have difference in operating temps of more than 10 degrees C between the two, assuming testing under similar conditions. My cards run at a difference of 4-5.

If that's the case, then one position is hotter than the other.

Hot air is not leaving the GPU quick enough and/or other hot air within the case is passing over the GPU.

Test again using both cards installed without and with the side panel on and compare the temps. Should be a bit hotter with the cover on. If confirmed, get another fan in there for the hot spot, if you can.
December 23, 2007 9:28:20 AM

I think I am having problems similar to Jingles 98, as now, I notice no more than 5 degrees difference between gpu 1 and 2. Given the sketchyness of past logs, I wouldn't be surprised if The temps shoot up again on one of them.


!