Could my 580s be dying? Temps going bonkers.

[imaheadcase]

Last month or so I been getting random resets on computer. Did not think anything of it, now they started happened more when playing games. Guild wars 2 for example it would just randomly reset every 10-20min, then when back in win7 would have that "check for errors message cause bluescreen".

I ran MSI Afterburner with default fan speed , the 2 GTX 580s are default settings. Temps on one card would reach 85c and the other bottom one 76c. That is about when it resets. Idle they are 65c/59c (that seems high) I manually put fan at %70 and it kept it around 82c for about a hour and no reset. I know they are rated close to 100c.

The reason I suspect temps is I also had case cover off for most of the night playing Gw2 and not a single reset. Only this morning after I put cover back on.

Is it worth replacing the fan unit? Very odd never had a problem until recently. I know original fans on it from purchase about a year ago. They look dirty or anything.

 

[RussianSensation]

Fermi generation is notorious for very poor TIM application. You can just take the plastic shroud off and see if there is a dust build up. While at it, you can replace the TIM with after-market TIM like Tuniq TX2, AC MX-4, etc. (be careful with AS5 as it can drop on the PCB and it's electrically conductive).

Here is MX-2 for $4 with free shipping for 4 grams.

 

[lopri]

Sounds just like my (now-gone) GTX 580. It used to throttle in the middle of gaming. Short pause, Windows Aero gone (with or without error message), and clock speed halved. Only way to fix was to restart the system.

Temperature is the most likely suspect. Clean up the fans and maybe think of an additional fan on the case near the cards, if the case allows? You could try underclock its 3D frequency as a last resort. ^^

 

[SZLiao214]

I always had nice drops in temperatures by removing their thermal paste and adding my own. They seem to use this really thick and dried out paste.

 

[BD231]

I had to replace the tim on my 480, the stock stuff dried up and temps shot through the roof. Just make sure you get the screws back on tight, stock sinks need a ridiculous amount of pressure on the heat plate for best results.

 

[AnandThenMan]

Another vote for redoing the TIM. Heck, I've had to do it on brand new cards, why can't these card makes get the simplest things right?

 

[VulgarDisplay]

If I was going through the trouble of redoing the TIM on my GPU I'd get an aftermarket cooler for it also.

 

[tviceman]

I removed stock TIM on an older laptop and it worked wonders with new TIM regarding the cooling. Give this a shot before giving up.

 

[Dadofamunky]

I think that TIM advice makes total sense. Those are great cards. And there are plenty of aftermarket coolers.

 

[RussianSensation]

Here is a 2-part review that shows how TIM actually has a positive effect on GPU temperatures, even among after-market TIM:

Part 1
Part 2

The heatsink on the 580 is very smooth which should provide clear benefits and makes it fairly easy to apply the TIM onto the GPU:

Just make sure to apply MORE than you would onto a CPU heatspreader.

 

[imaheadcase]

Update: So apparently its not temp problems. I would venture to guess at this point my power supply is dying. When i disable SLI i don't get any reboots into a BSOD. When i enable it, within 5min the PC resets and get the "send info to MS" for BSOD message.

I tried a couple different drivers and same thing. The power supply seems to be doing ok from a stats standpoint. Not voltages going to high/low out of spec. I guess its possible for it to start to die without any symptoms that i can see?

The power supply is suppose to be a good one, about 1 year old now Corsair HX1050. Its plenty fast enough. I guess even good ones have lemons.


I guess i got to get a new ps..

 

[OVerLoRDI]

Update: So apparently its not temp problems. I would venture to guess at this point my power supply is dying. When i disable SLI i don't get any reboots into a BSOD. When i enable it, within 5min the PC resets and get the "send info to MS" for BSOD message.

I tried a couple different drivers and same thing. The power supply seems to be doing ok from a stats standpoint. Not voltages going to high/low out of spec. I guess its possible for it to start to die without any symptoms that i can see?

The power supply is suppose to be a good one, about 1 year old now Corsair HX1050. Its plenty fast enough. I guess even good ones have lemons.


I guess i got to get a new ps..

Hmmm.. HX1050 dying after a year and struggling with 2x 580s?

Try swapping the cards around. Maybe your secondary card is faulty. Disabling SLI disables the secondary card (if a monitor isn't plugged in to it) and thus would "fix" your issue.

 

[imaheadcase]

You know I did not even consider that. lol I will try that.

 

[imaheadcase]

I switched the cards around, i had zero crashes in the 2 hours i played. However, I was getting flashing lines scaling up and down the screen in multiple games, including Heaven benchmark. Like when you film a TV, those lines that go up and down.

When i disabled SLI, everything works ok (well besides the decrease in fps).

So now i'm confused, does that mean the bottom card is bad, or the one i just switched. lol

 

[Keysplayr]

I switched the cards around, i had zero crashes in the 2 hours i played. However, I was getting flashing lines scaling up and down the screen in multiple games, including Heaven benchmark. Like when you film a TV, those lines that go up and down.

When i disabled SLI, everything works ok (well besides the decrease in fps).

So now i'm confused, does that mean the bottom card is bad, or the one i just switched. lol

Try each card individually (one card installed in the system at a time).

If they are fine, and you only have any sort of problem with SLI or both cards taking the load, then replace your power supply or RMA it.

 

[imaheadcase]

Ok i did that. Each card ran perfect, with benchmark programs, games, etc. No crashes at all.

So i was exploring the PS might be the cause, i'm not sure how accurate the program GPU-Z is. But this is what it showed at idle, any other program that is better let me know.
This is the MODULAR power connection to video card . Is this good?

This is the one i was using for 2nd card, it is the DIRECT CONNECTED one to power supply. Is this bad?

Perhaps try modular for both cards? I'm going to assume the VDDC current is a issue?

Sorry for all the questions before i try stuff, i'm just doing trial and error to see what you guys would do.

 

[imaheadcase]

Well i switched out PCI-E connection straight from PS and used a modular cable instead. So far everything is running smoothly, no resets..however something weird happened.

I had case cover off while I was testing all this, when i put case cover back on the back of it pressed against the modular cables at loop that connect to cards, then computer reset? That almost sounds like it could be a grounding issue causing the problem all along? That would explain random resets..very odd. I did a quick "no metal touching electronics" check and did not see anything, it could be a loose adapter on video card itself if its when case door touched it.

But so far no resets since last post this morning so hopefully swapping cards around, and replaying PCI-E cables fixes it.

/cross fingers

 

[Keysplayr]

Absolutely a contributing cause to sudden restarts. Loose power connections could mean a restart even if your foot taps the case. Happened to me a few times and drove me nuts. The particular power supply I had used power connectors that didn't seem to fit snugly on many locations the connectors were supposed to go. (mobo ATX connector, P4 power connector, PCI-e, etc.) What I did, was painstakingly pinch slightly each metal tube on the power supply connectors to they'd attach more snugly to everything. It was more difficult to plug into items, but I knew the contact was mint.
No problems after that.
Another time though, it was just a tired PSU that couldn't provide enough power at load. Would reset almost every time I fired up a game.