Major overheating issues

Th3M8dH8tt3r

Th3M8dH8tt3r

Th3M8dH8tt3r

Associate
Joined
16 Mar 2013
Posts
396
Hi guys,

I'm turning to you guys for some ideas.

My set up is as follows:
ASUS Crosshair V,
AMD FX6350 @ 4500mhz and 1.5v w/ XSPC water block.
2 x EVGA GTX780 in SLI, 1 w/ XSPC block and 1 w/ EK block.
Corsair 990D case
1 x 480mm roof rad with 4 Corsair fans pulling air through the roof, pushing through the rad
1 x 360mm basement mounted rad with 3 Corsair fans pulling through the side panel and pushing through the rad.
3 x front panel fans pulling air into the case
1 x massive exhaust fan pushing air out the back of the case.

These issues have really become apparent since yesterday when I added the second GTX780 which was second hand and came with an EK water block.

My CPU temps are fine, my GPU temps are through the roof though. I run a triple monitor set up, which I feel is important to share. I have tried various monitor configurations and SLI/physX configurations with no joy.

What I'm seeing is this:
CPU temps are fine, all fine and dandy.
GPU 2 which has two monitors on DVi idles at 30 Celsius.
GPU 1 which has one monitor on HDMI idles at 60 Celsius.

Load a game, we get to 100 Celsius in no time and it all crashes. I have throttled the GPUs to stay at no more than 95 Celsius which is insane as they clock out at 300mhz.

I've tried running the cards with all monitors on one card and physx on another, auto assign physx, disable SLi, tried all monitors on each card, tried splitting monitors and disabling SLI etc etc. You get the point.

I thought that I was just running out of resources in games but I think that it was temp throttling all along and now with 2 cards, it's got worse.

The EK cooled card has a backplate and to touch, the card is red hot. The backplate and cooling plate. Fitting is cool. The XSPC card and block are cool all over. Water doesn't feel particularly hot.

I've ordered up another XSPC block and SLI bridge as at the moment, I can't locate the cards in the right place so will strip both and redo thermal paste.

But, in the meantime, any ideas or thoughts on how bad it is would be appreciated.

I have an option of taking the kit out of an unused rig and run a second DC5 powered loop with a 360mm rad to cool the CPU and GPU independent of one another. Advice on the best twin loop set up based on my available equipment would be great.

Thanks all, I know it's been a long one.
 
It's whichever card has the 2560x1440 HDMI monitor connected that hits 100c and idles at 60. I was running the CPU much faster, it's an old overclock stored in the BIOS that I switched to for testing. Normally it's running 5.1Ghz at that voltage.
 
Plenty of rad space. Sounds like no or bad thermal paste as GPU 2 and cpu is of good temperature. A blockage would mean the whole loop would suffer.

Strip down GPU 1 and replace the thermal paste, get some gelid extreme. Unless you have other paste lying around. Take the block apart and clean the inside with a toothbrush and lemon juice while its out the case just to be safe.
 
Last edited:
I've done some more testing. The issue isn't specific to one or the other GPU. It's whichever GPU is powering the primary display. My three monitors consist of 2 1920x1080 monitors and 1 2560x1440. If I run just the 1920x1080s, then I can load up a game and play for a few minutes before the temps creep over 95C and head for the dreaded 100C. I can implement a 95C heat throttle and that keeps things running and I get 40FPS. Without throttling, I get 100FPS. If the 2560 is primary and the cards throttle, I get 15FPS.

I'm very confused. How can adding one additional card to the loop cause so many issues? I'm doubtful of an air lock. I've ran the system for 12+ hours and don't have the cap on the reservoir so it's well bled. As I mentioned, I do have a new XSPC water block arriving to replace the EK one so I can get the cards in the right slots, so will clear and reapply thermal paste. I'm using Thermalight CFIII HTC.
 
What sort of speed is the pump on?, pump definitely running? Flow to GPU's or cpu first?. Still they shouldn't rise no where near that high. So if you use hdmi with either GPU, either of them get to 100c?

Is there any chance to post photos of your loop

If that's the thing then it's probably not enough flow as your rads will be cold but your blocks will be red hot.

Very odd if it was running fine before the other card. I'd say most likely a blockage in your new GPU reducing the flow to cause high temps. still your cpu would suffer aswell. Unless it's first in your loop. Hmm...

Run a stress on your CPU and if that has a similar effect and temps start rising continuously after a few mins then it's a flow issue.

I have EK blocks and they are very good performers with a max of around 36c on both cards after hours of gaming.
 
Last edited:
A gpu should get nowhere near 95 degrees (mine rarely passes 35 degrees) when it's watercooled so something is clearly wrong. As already asked, can you post some pictures of your loop layout?
 
Are you running the GPUS in series or parallel?

If series then maybe the increased restriction has lowered flow rates below what is optimal for the primary GPU (which puts out more heat), if parallel then the reduced flow rate may be below optimal for the primary GPU (which puts out more heat),
 
tyler_jrb said:
What sort of speed is the pump on?, pump definitely running? Flow to GPU's or cpu first?. Still they shouldn't rise no where near that high. So if you use hdmi with either GPU, either of them get to 100c?

Is there any chance to post photos of your loop

If that's the thing then it's probably not enough flow as your rads will be cold but your blocks will be red hot.

Very odd if it was running fine before the other card. I'd say most likely a blockage in your new GPU reducing the flow to cause high temps. still your cpu would suffer aswell. Unless it's first in your loop. Hmm...

Run a stress on your CPU and if that has a similar effect and temps start rising continuously after a few mins then it's a flow issue.

I have EK blocks and they are very good performers with a max of around 36c on both cards after hours of gaming.
Click to expand...

Hi Tyler,

The pump is running at full speed and definitely running. The loop goes pump->CPU->480mm roof rad->GPU1(new GPU)->GPU2(old GPU)->360mm basement rad->pump.

If the 1st GPU is blocked, it's going to get sorted with the replacement of the EK block for an XSPC one. There could be an almighty air lock in that 1st GPU's block as well perhaps.

CPU temps aren't doing anything out of the ordinary, it's been stressed. I would expect temps of around 40C as you're seeing. The non-primary GPU is seeing temps of no more than 33C.

I'll post some pictures of the loop in the next post.


Threepwood said:
There must be something wrong with your flow, surely?
Click to expand...

Got good flow as far as I can tell.


pastymuncher said:
A gpu should get nowhere near 95 degrees (mine rarely passes 35 degrees) when it's watercooled so something is clearly wrong. As already asked, can you post some pictures of your loop layout?
Click to expand...

Coming next post

ubersonic said:
Are you running the GPUS in series or parallel?

If series then maybe the increased restriction has lowered flow rates below what is optimal for the primary GPU (which puts out more heat), if parallel then the reduced flow rate may be below optimal for the primary GPU (which puts out more heat),
Click to expand...

GPUs are in series but the hot GPU changes depending on which is primary. If flow was being affected by the first GPU, then making that GPU primary should cure the issue.

jamiehard1989 said:
Remove the old card from the loop for now and test with the new gpu, just to see if its thats card causing the problem.
Click to expand...

I wouldn't say so as both cards cook depending on if they're primary or not. I can't take it all apart again till the new block arrives so want to rule out as much as possible before replacing the EK block and getting it all finalised.
 
The top GPU is plumbed in wrong and has 0 flow.

To clarify, the in/out are on the same side, which means the coolant is bypassing the GPU itself, the only cooling it is receiving is from thermal transfer of heat from the hot liquid over it to the cooler liquid not over it.
 
ubersonic said:
The top GPU is plumbed in wrong and has 0 flow.

To clarify, the in/out are on the same side, which means the coolant is bypassing the GPU itself, the only cooling it is receiving is from thermal transfer of heat from the hot liquid over it to the cooler liquid not over it.
Click to expand...

It did cross my mind. But, would that explain the bottom GPU getting hot when that is the primary GPU?
 
Th3M8dH8tt3r said:
It did cross my mind. But, would that explain the bottom GPU getting hot when that is the primary GPU?
Click to expand...

No, could be a flow issue, but apart from that nothing comes to mind. It's obviously a separate issue to the plumbing, IIRC the primary GPU does more work so will run hotter.

My first thought was a flow issue because the are only two types of flow, enough and not enough, it changes for different blocks/heat loads. I had thought that maybe the CPU/secondary had enough but the primary went over the limit, but the lower card cannot have no worse heat transfer that the top lol.
 
ubersonic said:
No, could be a flow issue, but apart from that nothing comes to mind. It's obviously a separate issue to the plumbing, IIRC the primary GPU does more work so will run hotter.

My first thought was a flow issue because the are only two types of flow, enough and not enough, it changes for different blocks/heat loads. I had thought that maybe the CPU/secondary had enough but the primary went over the limit, but the lower card cannot have no worse heat transfer that the top lol.
Click to expand...

So it's not me being a dumbass or going mad then? lol. If the top card ran hot when primary but the bottom didn't when primary then I'd say there's an issue with the plumbing across top card or issue with top card but as BOTH cards do EXACTLY the same when either one is primary, I'm stumped.

What I might do is get extreme. By the mounting bracket to hang my other 360mm rad off the back of the case and then run the CPU off the roof rad and GPUs off the 2 360 rads. I'll use the big reservoir you can see now for the GPUs and then I've got a dual bay DC5 pump, reservoir combo as well which I can run the CPU with.

Insane plan?
 
russell664 said:
What pump are you using on the system?

It would be worth stripping the 2nd GPU and having a check nevertheless.
Click to expand...

DC5 on the bottom of the reservoir. I understood it to be that they're the best.

Only thing I can think of is that the top GPU is causing issue so the GPUs get hot. If that's the case, then when I replace the EK with XSPC, it'll be solved. Just want to cover bases for now. If everyone is saying the loop looks good and can't see anything obvious, then the problem is the EK and will be resolved when I remove it.
 
You must log in or register to reply here.
Back
Top Bottom