POV GTX680-SLI - 91c Temp GPU#1

[FMTopfan]

I've bought a couple of great GPUs - POV-TGT 2GB GTX680 Ultra Charged - but have a problem with the temps of the first card under 90%+ load where it will reach 91c. Ain't much space between cards because they are 2.5-slot, I'd call them more like 3-slot - despite that the fans are very quiet..

None of the GPUs are overclocked. The second GPU reaches a maximum of 56c under load and I know this is down to the extra space below the card (mobo is P8P67 Deluxe), the first card has minimal space and I'm stuck thinking of a solution without resorting to buying new mobo. I've tried swapping around the cards, placing a 120mm high-speed fan near the GPUs, and opening the side case because the 800D is average at air-cooling.

I had wanted to overclock these a little but in this situation I can't do that. Below is the photo of the 680s with the Asus Phoebus above in PCIe X1 slot.

Mobo slots layout:
PCIe 2.0 x1 [Asus Phoebus]
PCIe 2.0 x16 (x16 mode) [POV GTX680]
PCIe 2.0 x1
PCI
PCIe 2.0 x16 (x8 mode) [POV GTX680]
PCI
PCIe 2.0 x4

 

[bemaniac]

Swap them, if the bottom one is then the hot one RMA it. Actually forget that I see you tried it

 

[FMTopfan]

Swap them, if the bottom one is then the hot one RMA it. Actually forget that I see you tried it

I had swapped them, doesn't matter which is in the first slot they both get hot, there is no issue with the cards.

One GTX680 installed and temps are a nice 50c, soon as another is installed below it the temps rocket. The fans blow the air directly below and as you see in the photo, there is minimal spacing for this air to escape. Not sure what other people's temps are in SLI but mine looks very high.

 

[Rusty0611]

Do you have a side fan on your case? Not familiar with it....
If so set it to suck the hot air out your case. Those cards dump the hot air on top of each other hence the high temperatures. Unless you have something exhausting the air dumped they will just recycle the same hot air.

 

[OCPete]

Its gotta be worth adding an extra fan directed to blow cooler air between the cards

Try this? > http://www.overclockers.co.uk/showproduct.php?prodid=SY-000-AN&groupid=701&catid=57&subcat=399

 

[MichaelM]

I had the same in a 700D (practically the same) and Gigabyte windforce 670's but my top card was reaching 76C-78C with a sound card between them blocking two of the top GPU's fans. With the sound card removed they did 72C and then with the 120.3 radiator on top changed from inlet to exhaust it dropped that to 68C. Bottom card going from 60C to about 55C.

Now I don't know about the POV cooler and how well it will perform in comparison but here is something to try.
With these type of coolers the top GPU will recycle the bottoms exhaust flow so disable SLI so the 2nd card is idling and stress the 1st GPU in the same conditions to see if it reaches the same temperature.

Ideally use Heaven for both setups.

 

[FMTopfan]

Do you have a side fan on your case? Not familiar with it....
If so set it to suck the hot air out your case. Those cards dump the hot air on top of each other hence the high temperatures. Unless you have something exhausting the air dumped they will just recycle the same hot air.

The Corsair 800D has no fan on the side case. Rather than draw or write how it expels hot air I've found this article which explains it well. Only difference being my 3x120mm rad is in the top for the CPU.

Its gotta be worth adding an extra fan directed to blow cooler air between the cards

Try this? > http://www.overclockers.co.uk/showproduct.php?prodid=SY-000-AN&groupid=701&catid=57&subcat=399

Not sure if it will change the temps by much but would give it a go, not sure how I'll fix it in place when the first GPU is 6.5"/16cms above the floor of the chamber.

Now I don't know about the POV cooler and how well it will perform in comparison but here is something to try.
With these type of coolers the top GPU will recycle the bottoms exhaust flow so disable SLI so the 2nd card is idling and stress the 1st GPU in the same conditions to see if it reaches the same temperature.

I will try that right now on a few loops of Heaven 3.0, will be interesting as I didn't think to test that

 

[Rusty0611]

Still feel that card one is recycling hot air. You need to consider how to remove that hot air.

 

[FMTopfan]

I had the same in a 700D (practically the same) and Gigabyte windforce 670's but my top card was reaching 76C-78C with a sound card between them blocking two of the top GPU's fans. With the sound card removed they did 72C and then with the 120.3 radiator on top changed from inlet to exhaust it dropped that to 68C. Bottom card going from 60C to about 55C.

Now I don't know about the POV cooler and how well it will perform in comparison but here is something to try.
With these type of coolers the top GPU will recycle the bottoms exhaust flow so disable SLI so the 2nd card is idling and stress the 1st GPU in the same conditions to see if it reaches the same temperature.

Ideally use Heaven for both setups.

My radiator is set to exhaust as default and I never feel hot air from their anyway. I had found a spare 120mm fan and placed that on the rear of the outside case to exhaust hot air away from GPU#1 without any change in temps, and I can feel the warm air. Room temps is always 22c.

However, the Heaven benchmark's I ran (maximum settings) resulted in little change, non-SLI/SLI mode with both GTX680s in place.

Imgur is down for now so here is what I grabbed from Afterburner:
GTX680-SLI
GPU1: 91c, 94% (usage), 93% (fan speed)
GPU1: 60c, 93% (usage), 44% (fan speed)

GTX680 (SLI disabled)
GPU1: 86c, 98% (usage), 78% (fan speed)
GPU1: 30c, 0% (usage), 30% (fan speed)

That spare 120mm I had mentioned, I placed it on the rear of the case. Now as an intake fan it had reduced temps by 7c - still need to push them down more if at all possible?
GTX680-SLI
GPU1: 84c, 98% (usage), 72% (fan speed)
GPU1: 60c, 96% (usage), 45% (fan speed)

 

[MichaelM]

Ok, now if you where to remove the 2nd card completely (freeing up the fans intakes from any blockage) and run the top card by itself, would you still not be able to get it near 70C when stressed?

If so I think it's a fault with that cards cooler.

 

[ilovetechno]

Its gotta be worth adding an extra fan directed to blow cooler air between the cards

Try this? > http://www.overclockers.co.uk/showproduct.php?prodid=SY-000-AN&groupid=701&catid=57&subcat=399

This.

Was using 2 of these on my gtx480`s and it help a bit!

 

[tommybhoy]

Your air isn't going anywhere probably due to negative air pressure somewhere.

Something simple like switching off the side HD bay fan could stop it, if not you could try(Crude diagram alert):

All you need to test is some wire/string or even link up rubbers and 2 spare fans.

Fan 1 is self explanatory.

Fan 2 sucks air towards the side panel, not back towards the mb as the air wont shift against negative pressure from fan 1.

It's simple enough to rattle up in ten minutes, if it works(it might not) then you could try and source a 'Zalman FB123 Adjustable Universal Bracket'.

It's maybe hard to find(fleabay) but that way you can use any fan(quieter than a spotcool) you like, also enabling good flexibilty positioning the fan and it's a better bet than a spotcool.

 

[Kaapstad]

had swapped them, doesn't matter which is in the first slot they both get hot, there is no issue with the cards.

One GTX680 installed and temps are a nice 50c, soon as another is installed below it the temps rocket. The fans blow the air directly below and as you see in the photo, there is minimal spacing for this air to escape. Not sure what other people's temps are in SLI but mine looks very high.

I use the same case as you. I have the rad for a corsair H100 in the top of the case. I use GTX690s in SLI. The highest I have seen temps wise when overclocking was 91c on 1 GPU the other 3 GPUs were between 80c - 85c the card fans were at stock.

I think the difference is GTX690s are only 2 slot cards and on a RIVE mobo there is plenty of room around them.

 

[FMTopfan]

Ok, now if you where to remove the 2nd card completely (freeing up the fans intakes from any blockage) and run the top card by itself, would you still not be able to get it near 70C when stressed?

If so I think it's a fault with that cards cooler.

The temps I got was a maximum of 58c running for 15 minutes.

I did notice that one of cards is a different BIOS version from my other GTX680 and what was included in this fudzilla review.

Here's what I see:

Thanks tommybhoy for your suggestion, grabbed some spare fans and faffed on for an hour yesterday and today and this is what I came up with:

Max temps with that setup is 83c GPU#1 (78% fan), 59c GPU#2 (46% fan). Sitting idle at desktop now, it's a pleasant 31c/26c (fans 30%). Motherboard is 26c, CPU 30.

From the great suggestions so far, that is the best I'm going to get which is better than hitting a worrying 91c initially.
However, I was wondering if it would be worthwhile watercooling these for much lower temps, and if its actually possible since I'm not 100% sure if these are reference PCBs to fit blocks on?

 

[MichaelM]

With that much effort put into cooling and trying so many different things, I would be inclined to think that the cards cooler is faulty. I'm very sure that even reference 680's would not run that hot in those conditions.

The top GPU is hitting the 70C thermal throttle so it's downclocking and undervolting itself, which makes the fact you are still hitting 80+ even more worrying.

I would suggest that you shoot off an email to POV (or the people you bought them from) and clearly explain the issue and what steps you took to solve it including their effects.

Best of luck.

 

[Mysterae_]

Great illustration!

I looked into case fans and pressure when I had issues with high temps of 2 670 gpu's.

You're case probably has negative pressure, judging by the number of extract vs intake fans. You'll know for sure by adding up the CFM of all intake fans and deducting it from the total CFM of extraction fans. Remember to include the psu fan and your gpu's depending on their fan design.

In effect you are removing more air than you are supplying. With the negative pressure, air will get in from all the holes and seals of your case. That air is useless because it'll be doing nothing for your cooling as it'll go out the nearest fan, and it will be unfiltered. If the case pressure was positive the air would go out of those holes and seals (if the pressure was far greater than the extraction).

I think you need more intake fans or like you suggested go water cooling for your gpus. If you go water cooling I would still suggest more fans though.

 

[FMTopfan]

With that much effort put into cooling and trying so many different things, I would be inclined to think that the cards cooler is faulty. I'm very sure that even reference 680's would not run that hot in those conditions.

The top GPU is hitting the 70C thermal throttle so it's downclocking and undervolting itself, which makes the fact you are still hitting 80+ even more worrying.

I would suggest that you shoot off an email to POV (or the people you bought them from) and clearly explain the issue and what steps you took to solve it including their effects.

Best of luck.

I bought those two POVs from Overclockers last Friday and could ask them for advice but I'll ask POV too to see what they say to resolve this worrying issue. Bar the two cards sitting so close, I can't see how it's particular to my setup unless it needs that extra slot for air clearance which wasn't made clear by OcUK or POV website - in that instance you need 4 slots which isn't SLI compatible - and these are rated SLI compatible.

MichaelM, I didn't think to check voltages and have checked this running Heaven again with some notes as follows:

Starting the benching (maximum settings under load) all looks fine but clocks are still not identical as noted in previous post...
GPU#1: 74% power, 1.175v
GPU#2: 75% power, 1.175v

Getting warm a few minutes in...
GPU#1 (76c): 75% power, 1.162v
GPU#2 (56c): 76% power, 1.175v

5 minutes in I hit.. 80-84c!!
GPU#1 (84c): 73% power, 1.150v
GPU#2 (57c): 75% power, 1.175v

Then back to idle after typing this post...
GPU#1 (32c): 8% power, 0.987v
GPU#2 (26c): 8% power, 0.987v

You're case probably has negative pressure, judging by the number of extract vs intake fans. You'll know for sure by adding up the CFM of all intake fans and deducting it from the total CFM of extraction fans. Remember to include the psu fan and your gpu's depending on their fan design.

Another good suggestion, my extraction for the case as a whole is poor where you look at intakes, only #2 fan is supplying fresh air. Not sure on these bundle of mixes of fans what the CFM is at all. I could pop fan #3 next to fan #2 and see what the temps are without a fan as an exhaust - those two fans are the highest RPM fans I have bar the YS Tech fd121238hs hoover fans I have which are far too noisy @45dB 6k rpm/125cfm - then place yet another low rpm-fan in position #3. All will be clear in the end... every day it seems like I'm building a wind-tunnel!

 

[Kaapstad]

I bought those two POVs from Overclockers last Friday and could ask them for advice but I'll ask POV too to see what they say to resolve this worrying issue. Bar the two cards sitting so close, I can't see how it's particular to my setup unless it needs that extra slot for air clearance which wasn't made clear by OcUK or POV website - in that instance you need 4 slots which isn't SLI compatible - and these are rated SLI compatible.

POV along with a lot of reviewers may have tried them in SLI on a board like the RIVE where you put the cards in PCIe slots 1 and 3 giving an extra slot of space between them. They also would not have tried it with the Asus Phoebus above them.

I know this is not very helpful but I think the real problem is the cards are dumping all their heat into the case not out the back like the ref design. They are also too close together with other cards also close.

Where the cards are non ref it maybe worth finding out if they do throttle @70c or can go higher.

Have you thought about changing them for gtx680s that dump all the heat out the back of the case.

I am sorry I have not been much help.

 

[Aurhinius]

Water cool them and all your problems will go away.

Pretty expensive but it will sort the problem as long as you have enough radiator surface area.

The board is likely to be reference but you can check on the EK website or just google it.

Someone will have fitted a waterblock to a POV card.

Notes - I am running two reference EVGA 680's in SLI that exhaust out the back.

The Top card is about 5-10°c hotter than the bottom card with one slot between them.

Ultimately the top card in an SLI setup is partially starved of air due to the small slot gap. My CM HAF 932 has a 200mm side fan that blows cool air in over the cards. This helps but doesn't solve the issue.
Remember also the back of the bottom card radiates heat up towards the intake of the top card! The back of said bottom card is not cooled at all. You end up with a fairly static pocket of heated air in the space between the two cards. Shoving cards in that space just makes things worse.

Also remember no two 680's will hit the same speeds or throttle in exactly the same jumps. One of my cards never hits the same high speed as the other. Luck of the draw and silicon lottery.

I think you have the right location for your soundcard you are just having issues with those cards circulating the hot air in your case and not exhausting it out.

Experiment with having the side fan pull air out of the side of the case it is at odds with what the GPU fans are trying to do but may help clear the static pockets of hot air.

The sure fix is of course to water cool them.

 

[FMTopfan]

POV along with a lot of reviewers may have tried them in SLI on a board like the RIVE where you put the cards in PCIe slots 1 and 3 giving an extra slot of space between them. They also would not have tried it with the Asus Phoebus above them.

....

I tested with the Phoebus above and removed and it made little difference, only 1c difference which is quite good. This is running Heaven each time for 15 minutes to give it enough time to reach thermal maximum. The reviewers are lucky they have four PCIe x16 slots whereas I only have two to use, would have been nice to space them out - even then that means going without a soundcard.

My previous GPU was the 5970 (dual-gpu on a single pcb) and those temps never worried me, reached 73c with a 12% overclock because it exhausted the hot air.

However, the issue remains that on testing them again (installing POV GTX680 on it's own) there is still a discrepancy in the clocks from one another as posted before. I ran Heaven twice and then Furmark to monitor the vitals as follows again (spent a few hours testing these GPUs ):

[Test 1] Single-GPU#1 installed only in first PCI-E 2.0 x16 slot (plenty of room for air)
GPU#1 - 57c max, Core: 1189MHz, Shader: 2378MHz, Fan: 83% (auto), Voltage: 1.175

[Test 2] Single-GPU#2 installed only in first PCI-E 2.0 x16 slot (plenty of room for air)
GPU#1 - 58c max, Core: 1215MHz, Shader: 2430MHz, Fan: 83% (auto), Voltage: 1.175​

No thermal throttling so why the lower clocks on the first GPU? Must be due to BIOS being incorrect and not being tested to verify, POV-TGT fault there maybe. They should be identical because I bought them together. And is it right that the BIOS is different too from one-another?

Notes - I am running two reference EVGA 680's in SLI that exhaust out the back.

The Top card is about 5-10°c hotter than the bottom card with one slot between them....

Thank you, that is the variation I expect too although you didn't say what the top GPU temp was. I had already placed a 140mm fan on the side of the case as an exhaust (see my previous posts) without solving the heat issues. I have just checked a reference Gainwood GTX680 and those temps on it's own was 80c - link to article. Dread to think what SLI would do, maybe 91c like mine.

I would love to try watercooling since the CPU watercooling was done by OcUK but realise that building it will mean I can maintain/repair - however I have no idea exactly what I'll need so maybe it's not such a good idea despite my years of building PCs. This is what I have for CPU cooling currently; Black Ice SR-1 360 rad, XSPC 5.25" single bay reservoir (BR525AL), Supreme HF Nickel Plexi, 6x12mm (1/2") G1/4 high-flow fitting, not sure what the pump is but it's a small 2" block, and 2m 7/16" tubing.