help! is my gtx580 SC dying?

[darckhart]

2 days ago I noticed artifacting and flickering and generally lots of misdrawn stuff in my games. thought it was the new drivers (266.58) i installed, so i reverted back to 266.35. but the problem still persists. the only thing i notice now is that the heatsink is quite hot to the touch, yet the monitoring programs show my gpu temp to be normal (38C idle, maybe 70-79C in game).

See what I mean here:
hot part on the card
ME2 load screen
ME2 in game

[Funkwagen]

Are your games crashing too? I'm getting my 580 back from RMA in a couple days because I had a similar issue. I think that part you pointed out though is always hot during use, that isnt abnormal. Try reverting back to 263.09 and see what happens

[Chickenfeed]

The part you have pointed out in your picture there is the heatspreader. It is normal for this to get quite warm / hot to the touch as it is in contact with all the heat producing components on the pcb appart from the gpu die itself.

As far as the visual anomalies you've shown what other things have they appeared in (other games? benchmarks? etc) Also what kind of PSU is that? It looks like a Corsair TX950 if I am not mistaken which should be fine as its on Nvidia's approved SLI list.

You haven't mentioned if your cards are running at stock or not. If they aren't try reverting to stock clocks, same for your cpu / ram ( again for trouble shooting, it can't hurt to check even if you believe your cpu / ram clocks to be solid )

[darckhart]

@funkwagon: the only game that consistently CTD was crysis. every other game i played (m2033, me2, guild wars) ran just fine. i'm back on 266.35 for the moment.

@Chickenfeed: yes, i figured that was supposed to get hot. is there a way to be certain these artifacts aren't heat related? because it seems like the anomalies disappear when i point a room fan directly at the card. i'll have to do some more tests later, but for sure the anomalies go away after i shutdown and let the card cool down. i wonder if applying some better TIM could help.

the visual anomalies also occur in evga oc scanner (counts artifacts in the thousands) and fluidmark (the spheres have spikes and shafts randomly shooting through them. couldn't get a screenshot yet). i think once it starts artifacting, every 3d app goes wonky.

yep that's a tx950. and everything (cpu included) is at stock speeds at the moment. in fact, i have been testing each card separately and i think ony 1 of them is having the problem.

i also posted in the evga forums and several other people there have been reporting the same problem.

[Funkwagen]

@funkwagon: the only game that consistently CTD was crysis. every other game i played (m2033, me2, guild wars) ran just fine. i'm back on 266.35 for the moment.

@Chickenfeed: yes, i figured that was supposed to get hot. is there a way to be certain these artifacts aren't heat related? because it seems like the anomalies disappear when i point a room fan directly at the card. i'll have to do some more tests later, but for sure the anomalies go away after i shutdown and let the card cool down. i wonder if applying some better TIM could help.

the visual anomalies also occur in evga oc scanner (counts artifacts in the thousands) and fluidmark (the spheres have spikes and shafts randomly shooting through them. couldn't get a screenshot yet). i think once it starts artifacting, every 3d app goes wonky.

yep that's a tx950. and everything (cpu included) is at stock speeds at the moment. in fact, i have been testing each card separately and i think ony 1 of them is having the problem.

i also posted in the evga forums and several other people there have been reporting the same problem.

I'll run through what I did to see if I could fix mine: (Note every step for me didnt work)
ALL system settings were STOCK prior to this testing

1) Tried every recent driver, including the hacked quadro ones (also played a bunch of different games to make sure it wasnt just the game)
2) Tried underclocking gpu to see if issue persisted
3) Tried running fan on maximum to see if the artifacts showed up even at a low temperature while gaming
4) Tried different nvidia control panel settings (pre rendered frames, AA, AF, etc)
5) Ran OC scanner like you did to see if artifacts appeared
6) Tried a different PCI-E slot
7) Tried different power cables (single rail psu...but still)
8) Tried different PCI-E clock settings on the motherboard (99,100,101mhz respectively)
9) Tried different RAM/ Fewer sticks of RAM
10) Tried different voltage settings for the gpu. a raise from default voltage DID help, but not completely and it still shouldnt be happening
11) Tried an entirely different system

If you get to 11 and still nothing, I'd say RMA time

[Neova]

I did a test to confirm my card was going back and not drivers when my 465 was acting up. I did end up RMA it and now it has been repaired.

Goto your windows desktop and preview any animated screensaver. Just let it run. If it starts to artifact or lockup, or hang your system or driver crash, RMA it.

If it is stable, then try to run uniengine heaven's benchmark, but don't benchmark it. Just let it run at max settings and get a cup of coffee. You want to test stability. If it doesn't crash or artifact, then cycle through the 4 APIs to see if you have bad drivers (opengl, dx9, dx10, dx11).

If it crashes, it is more than likely hardware issues.

[darckhart]

yea it's hard to pinpoint when it will occur. i'll try the unigine tonight.

[darckhart]

so i took the cards out and began trying them one by one. each alone passes oc scanner for 15 min and all 4 unigine benchmarks. in sli, however, it'll reach a point where it blank screens with a "unigine fatal error swap"