• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

SOLVED Sudden spike in temps

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

SolidxSnake

Member
Joined
Dec 26, 2004
Just finished building my new rig/WC loop over the past month or two. Rig all in sig.

Since I got the GTX580 in, I was getting load GPU temps around 50-55C depending on how hot the room was/if i had the case panels on/off, etc.

I had some issues with a GPU load and my HDD dropping (see other thread in general hardware if you're interested). Got it fixed today, and have been folding all day no issue. I decided now that I've got some free time I'd try and OC the card a bit. Closed FAH GPU, opened up OCCT to stress test the GPU and started up MSI Afterburner. At which point I noticed my GPU jump from around 40ish idle to 55->60->71C.

I just remounted the block and no change.

CPU temps have remained the same (loads around 55C with case panel off, 60-65 with panel on).

Perhaps it's clogged? =_=
 
Last edited:
Prolly not. The CPU block clogs first.

Pushing the card too hard probably.

I will say, having the case side off getting CPU temps of 55 and the 65 with the side on, you need better more fans.

What was the CPU/GPU load temps with Furmark AND Prime 95 at the same time for 30 min? Before overclocking the GPU.

Ambient temps?
Case side on?
Case side off?

Same test with GPU overclocking.

Really can't help till you do the above tests. Science needs tests and eliminating variables. Physics DEMANDS it. So, fill in the blanks please.
 
Prolly not. The CPU block clogs first.

Pushing the card too hard probably.

I will say, having the case side off getting CPU temps of 55 and the 65 with the side on, you need better more fans.

What was the CPU/GPU load temps with Furmark AND Prime 95 at the same time for 30 min? Before overclocking the GPU.

Ambient temps?
Case side on?
Case side off?

Same test with GPU overclocking.

Really can't help till you do the above tests. Science needs tests and eliminating variables. Physics DEMANDS it. So, fill in the blanks please.

All temps are with all components stock (up until now, CPU is running at 4001MHz now, been running P95 blend all night and maxed out at 63°C with panel on, lower ambient than normal though).

I haven't had the time to thoroughly test everything for 30mins or so. I'll get to that when I do have time. I will however mention when running FAH SMP and GPU at the same time, CPU temps stayed around 60-65ish and GPU temps maxed around 49C after an indefinite amount of time (several hours, overnight, etc). However, now if I open up Furmark, OCCT GPU, or FAH GPU (and leave the CPU idle) temps jump almost instantly to 50C and keep climbing about a degree or two every couple of seconds and stabilize at around 70C after about 30 seconds from the time I start loading the card. I have not overclocked or overvolted the card in any way.

As far as the side panel on/off thing, yeah I do need to get stronger fans or remove the restriction on them (the side panel mesh is pretty dense). I'm sacrificing temps for silence though, which is okay by me.
 
I agree with Conumdrum, you need better case cooling. No mater what kind of water cooling you use you will still need the proper air flow through the case. Taking the case cover off and putting it back on proves that you need more case air flow.
 
Where is your rad pulling from? Is the air going in rather warm? Is the air coming out rather warm? Is your block hot?
 
I agree with Conumdrum, you need better case cooling. No mater what kind of water cooling you use you will still need the proper air flow through the case. Taking the case cover off and putting it back on proves that you need more case air flow.

It's not particularly case airflow, it's the side panel itself. Look through my build log. I have weak YLs on the rads, and they are right up against the grate in the side panel, which really hinders their ability to exhaust. So the solution would be stronger fans (which I'm not too keen on due to noise concerns) or a less-restrictive grill. However, it probably wouldn't hurt to cut a hole in the bottom for a cool-air intake, since there isn't very much of that.

This is all off topic however, just because my setup runs warm doesn't help me isolate the cause for a major spike in temps with no change. Before I noticed, upon a load temps would instantly jump about 5-7°C above idle, and then slowly rise another degree or two more. Now temps jump instantly about 10°C and increase very quickly (about a degree Celsius every second or two) until they hit about 30°C above idle (before they maxed about 10-15°C over idle). CPU temps have remained constant, and I've remounted the block once (perhaps I could do it again and see).

edit: again, I know my loop isn't running as best it could (either fans are too weak, rads not pulling the coldest of air, etc). However none of this has to do with a seemingly random spike in temps. I'm looking at delta values, not absolute values. At the moment I just want to try and isolate causes as to why out of the blue my card suddenly runs 20°C hotter under the same loads.

edit 2: in case i wasn't clear above, i have not OC'd the card yet. I was about to, then I noticed my temps well above what they were before, with no change in loop.
 
Last edited:
Where is your rad pulling from? Is the air going in rather warm? Is the air coming out rather warm? Is your block hot?

Answer these, if the air coming out is rather hot, it's a problem with the exhaust or your loop being overwhelmed or something. If the block is hot, heat is not getting out of the block, it's probably clogged or something. If it's neither of those, heat is not getting into the block and you need to make sure contact is good. Those are the only scenarios for any sort of change.
 
Answer these, if the air coming out is rather hot, it's a problem with the exhaust or your loop being overwhelmed or something. If the block is hot, heat is not getting out of the block, it's probably clogged or something. If it's neither of those, heat is not getting into the block and you need to make sure contact is good. Those are the only scenarios for any sort of change.

:bang head Of course I don't logically think of the simple troubleshooting? Will do. Thanks for the straightforward help :thup:
 
So, I decided after a third remount that I would pull the blocks off of my loop (<3 quick disconnects), drain them and disassemble the GPU waterblock. Then I realized I was at school, and I don't have all the necessary tools or time to facilitate such a project at the moment. So I test-mounted the block on the card a couple of times, and noticed that the mount I just made looked better than it did when I pulled it originally (and the second time). So I did it a couple of times, and finally mounted it and reinstalled everything into the loop. Idle was already about 5-6C lower (to around 32-33C from 38-40C), and upon starting OCCT GPU, the temps jumped instantly to 40C and slowly climbed into the mid-high 50s. OCCT GPU is running now, has been for about 3-5 minutes, and temps are fluctuating between 56-58C.

So it looks like it was a mount issue. Guess three remounts isn't enough to isolate stupid error.

Edit2: I tried (while it was loaded/running, of course) to modify mounting pressure. I kept OCCT GPU running and tightened each thumbnut about half a turn and instantaneously my temps changed from 56-58 to 61-64. I tried loosening them and quickly my temps dropped back down.. I think board flex is what is killing my temps.
 
Glad you found it. And yea, overtightening can do bad things. I ruined a GTX 280 by overtightening. Into the trash it went, RMA was not an option, I do have Karma.

Stuff happens, glad you got it right.

LOL, maybe moving your hair outta your eyes would help. J/K, it's a neat pic.
 
Glad you found it. And yea, overtightening can do bad things. I ruined a GTX 280 by overtightening. Into the trash it went, RMA was not an option, I do have Karma.

Stuff happens, glad you got it right.

LOL, maybe moving your hair outta your eyes would help. J/K, it's a neat pic.

:D I work by intuition, not sight!

Been running FAH GPU and CPU all night, GPU hasn't cracked 57C and CPU has maxed at 67C (hottest core is 63C right now, room temps have definitely dropped overnight)
 
In case anyone's interested, I flipped the radiator fans so they pull from the side and blow into the case. So far, temps at the moment are marginally lower, but I also have the videocard overclocked.

Shortly after my last post, I replaced the side panels but pulled the mesh off of the side panels (so that there was no restriction on the fans blowing out). That left temps the same regardless of side panels on/off. Then today I decided to flip the fans, and put the mesh back on the case simultaneously.

So the temps under a full FAH SMP/GPU load have remained the same, but with +.1v on the GPU (and +168MHz clock) and the mesh behind the fans (intake side). The "System" temp has increased from about 38-40C load to 50C, but that makes sense (as now the hot air is going from the middle hot-pocket in the bottom of the case and getting sucked up). The top of the case is actually blowing out warm air now xD
 
Back