SLI Problem: GPU 2 is Way hotter

[Poochymama]

I recently built a new computer with 2 Zotac GTX 460 in SLI. Up until a few weeks ago I hadn't really done much gaming so I really never noticed a problem. However, now that I am gaming, it's clear that there is some kind of problem going on.

If I'm running in SLI after about 10-20 minutes of gaming the computer locks up and starts stuttering and eventually I just have to hold the power button to shut it down. What I noticed is that when this was happening, the GPU fans were running a million miles per hour. This led me to believe that it might be an overheating problem, so I downloaded afterburner to try and monitor what was going on which led me to the source of the problem.

GPU 2 is considerably hotter than GPU 1.
At Idle
GPU 1 : 37C - fan speed 20%
GPU 2 : 44C - fan speed 20%


Heavy Load
GPU 1: 62C - fan speed 23%
GPU 2: 85C - fan speed 65%

Also, GPU 1's temp rise is much more gradual and looks like a smooth hill where as GPU 2's graph has huge peaks. Here is a picture

Any idea what's going on?

 

[Bo$$]

try cleaning out the fans, how much room/cable mess in the case?

 

[_JP_]

Yeah, I'd put my money on the almost non-existing air-flow for the 2nd card to cool down, for now. You can prove me wrong by telling us what is your chassis, motherboard model and if the PSU is bottom mounted with the fan facing upwards (sucking air from the GPU area). You could also tell us if there are any other expansion cards installed. Fill your system specs here. Don't forget to select 'Yes' on the drop-down box in the beginning of the page.

 

[Bo$$]

Yeah, I'd put my money on the almost non-existing air-flow for the 2nd card to cool down, for now. You can prove me wrong by telling us what is your chassis, motherboard model and if the PSU is bottom mounted with the fan facing upwards (sucking air from the GPU area). You could also tell us if there are any other expansion cards installed. Fill your system specs here. Don't forget to select 'Yes' on the drop-down box in the beginning of the page.

even a photo of the case would do

 

[Poochymama]

Yeah, I'd put my money on the almost non-existing air-flow for the 2nd card to cool down, for now. You can prove me wrong by telling us what is your chassis, motherboard model and if the PSU is bottom mounted with the fan facing upwards (sucking air from the GPU area). You could also tell us if there are any other expansion cards installed. Fill your system specs here. Don't forget to select 'Yes' on the drop-down box in the beginning of the page.

First of all, thanks for the quick replies guys

It's interesting you mention the PSU cause that's what I originally thought it was(and it still might be), I haven't really had time to star messing with things. My PSU is bottom mounted and the fan is facing upwards. The only thing that leads me to believe that's not the problem is that it's the top card that is getting hot(I think), wouldn't the bottom card be the hot card if it were the PSU?

Cable clutter is pretty good, I made sure that wouldn't be a problem when I built it. I eventually will add more fans, but for now stock cooling is all I have.

 

[Bo$$]

is anytihng below your second card?

 

[douglatins]

Also are they the same cards? Or do they have different coolers

 

[sneekypeet]

Could be that MSI is reporting the cards backwards

 

[Poochymama]

is anytihng below your second card?

The PSU

 

[Poochymama]

Also are they the same cards? Or do they have different coolers

Same cards

 

[wolf]

9 times out of 10 in this situation it's that the 'top' card can hardly breathe because the 'bottom' card is so very close to it, theres hardly any gap at all.

looking at a pic of your board tho you should have at least 1 pci slots worth of space between them...

easyest way for us to help I think is a picture of them both sitting in the case.

 

[jsfitz54]

I have SLI 460 setup and the same issue but not to your extreme. I originally had them next to one another and then I separated them for minor difference. F@H appears to use card 2 heavily...96% or so and card 2 gets in the 60c range.

I have 3 120mm side fans, 1 120mm HD fan in, 1 120mm fan out, PSU fan, 1 140mm fan blowing on back of mobo, 130mm fan on cpu cooler.

I don't think it is soley a fan issue but they help.

My mobo allows adjusting in bios of optional fans and they are set to 90%.
CPU fan at 60%.

I am eager to get some other responses here as well. Currently at stock but I bump to 850 when I need it.

EVGA Precision 2.0.2 reports them the same way as MSI Afterburner 2.0, I've used both utilities.

 

[Poochymama]

9 times out of 10 in this situation it's that the 'top' card can hardly breathe because the 'bottom' card is so very close to it, theres hardly any gap at all.

looking at a pic of your board tho you should have at least 1 pci slots worth of space between them...

easyest way for us to help I think is a picture of them both sitting in the case.

Can't get a pic atm because I don't have a camera. I'll get one up as soon as I can. As for the spacing, there is about an inch and a half between the top card's front and the bottom cards back.

 

[Poochymama]

Could be that MSI is reporting the cards backwards

What do you mean by this?

 

[Eternalchaos]

his system spec say he has a Asus P7P55D-E and all of them have a 1pci slot gap for graphic cards

@Poochymama; with MSI afterburner you can enable an onscreen display (like FRAPS) so you can set it up to show your GPU temps in game and you can also try setting it fans to 100%

 

[Poochymama]

his system spec say he has a Asus P7P55D-E and all of them have a 1pci slot gap for graphic cards

@Poochymama; with MSI afterburner you can enable an onscreen display (like FRAPS) so you can set it up to show your GPU temps in game and you can also try setting it fans to 100%

I'm having trouble setting up the on screen display. I set it to "]" to turn it on, but while playing Oblivion I hit that key and nothing happens.

Also, how do you set the user defined fan speeds in afterburner? I got the graph thing to come up once by clicking the user defined button underneath "auto" on fan speeds, but since then I haven't been able to get back to that screen, clicking user defined simply toggles it on or off.

I tried turning the fans up to 70%(that's the mas it will let me do with out the user defined graph), and disabling auto. Here are the results

Under Load

GPU 1: 59C
GPU 2: 87C

 

[Poochymama]

Disregard second question, I figured it out

 

[_JP_]

Ok, another thing popped to my mind. What is the condition of the thermal paste on the 2nd card?
Please check if it has enough, that it is well spread, covering all of the die, or if there's any at all (I doubt this last one can happen...but there's always that chance).

 

[Beertintedgoggles]

Come on everyone, the most obvious test.... swap the two cards and see if the top one is still hotter. If so then you need to increase the airflow to the area. If the same card is still overheating then check the thermal paste. If all that is still good, then maybe it's time to RMA (especially if it is causing your system to crash like you were mentioning).

 

[Eternalchaos]

I'm having trouble setting up the on screen display. I set it to "]" to turn it on, but while playing Oblivion I hit that key and nothing happens.

I dunno about toggling it on/off, I always have it on by default but check it is showing your gpu temps

go to the afterburner setting> click the tab Monitoring > go down to active hardware monitoring graphs > click on GPU Temp > below click on Show in On-Screen Display

 

[jsfitz54]

Come on everyone, the most obvious test.... swap the two cards and see if the top one is still hotter. If so then you need to increase the airflow to the area. If the same card is still overheating then check the thermal paste. If all that is still good, then maybe it's time to RMA (especially if it is causing your system to crash like you were mentioning).

DID THAT first thing same result, bottom card always hotter?????

Always hotter and has better airflow because nothing is stacked under it. In my case, the bottom is perfed metal, so I know it's not starved for air.

I would presume some sort of load balance in SLI but bottom card gets the big hit, per my screenshots.

 

[Melvis]

Those temps would be considered LOW compaered to my m8s SLi GTX460's, ive seen my m8s top card hit 94c often in summer time here, but he has had similar issues to you even when it wasnt summer here. In the end he is getting another Mobo, as he thinks the mobo is the issue because he ran his old cards (9600GT's) to see if the issue would go away and it did not, yes the temps dropped by ALOT but not the black screen of death and lock ups etc.

My old SLi set up shows one card mega hot compared to the other, but i found out that the temps was incorrect for the second card, so now i read the second card - about 10-15c on it.

 

[Poochymama]

Come on everyone, the most obvious test.... swap the two cards and see if the top one is still hotter. If so then you need to increase the airflow to the area. If the same card is still overheating then check the thermal paste. If all that is still good, then maybe it's time to RMA (especially if it is causing your system to crash like you were mentioning).

Gona do this tonight when I finish my work.

Update: I was wrong about it being the top card getting hot. It's is actually the bottom card that's right above the fan up PSU that's getting hot, so maybe that does have something to do with it.

About the thermal paste issue. I've never really done anything related with that before. Where is a good place to read up on that?


Another thing - When I run oblivion in SLI there seems to be no increase in performance really at all. 60 fps outside and 30-40fps in cities regardless of whether or not SLI is enabled, this leads me to believe it's not even working correctly.

When I run Oblivion without SLI

GPU 1 caps at about 60C and GPU 2 stays at idle temps(as expected).

 

[jsfitz54]

Those temps would be considered LOW compaered to my m8s SLi GTX460's, ive seen my m8s top card hit 94c often in summer time here, but he has had similar issues to you even when it wasnt summer here. In the end he is getting another Mobo, as he thinks the mobo is the issue because he ran his old cards (9600GT's) to see if the issue would go away and it did not, yes the temps dropped by ALOT but not the black screen of death and lock ups etc.

My old SLi set up shows one card mega hot compared to the other, but i found out that the temps was incorrect for the second card, so now i read the second card - about 10-15c on it.

Independently, each card runs fine. OP has same brand different model mobo. I don't think it is mobo. I was thinking Nvidia drivers somehow. Or as you state, how the sensors are mis reporting the temp.

 

[Beertintedgoggles]

Well s**t then, could def. be a load balancing problem with the drivers although I would suspect this would be much more widely seen by others.