Radeon 5870 running hot at idle

[sjankis630]

I just noticed last night that my Radeon 5870 is running hot at idle.
I opened speedfan 4.49 and my gpu temp is 83-89c at idle. There is nothing else running on my computer at the moment. I can hear the fan churning at 49%.
As a result my system temp is 56c when at idle it is usually below 40c. CPU usage is below 1% and there is nothing else running. I don't get this.
I tried reinstalling drivers and that seemed to calm it down last night and I was able to use the computer as normal, but now today after using the system - which has not been changed in any way that I know of in at least 6 months, the temps are back up.
What steps can i take to a) fix this b) check to see if my card is failing c) check to see how and why the drivers seemed to help last night.
The Catalyst control center keeps shutting down when I try and get some information so it essentially is useless at this point.
Again i have played no games nor have I stressed the machine in any way other than surfing the web in the last 48 hours.
Thanks in advance.

 

[RussianSensation]

1. Remove the GPU shroud and clean all the dust with compressed air/paper towel.

2. While at it, you may want to consider replacing the thermal interface with something like Tuniq TX-2, TX-4, Prolimatech PK-1, Thermalright ChillFactor III, Noctua paste, etc.

-Remove screws (circled in red), (4 gpu cooler screws and 8-10 cooler case screws)
-Remove (2) screws from rear exhaust plate
-Lift entire case/cooler off the pcb slowly and be very careful not to remove and/or break the heat tape located on the VRM (voltage regulator modules)(circled in blue).

When you are replacing the TIM, use Isopropyl 90-99% alcohol with a Bounty style towel, not a soft tissue/Kleenex.

This is enough: http://www.mwave.com/mwave/SKUSearch...m#.UbydCfn-HYE

 

[VirtualLarry]

And scan your system for malware. There may be malware (rootkit, etc) running on your rig, running bitcoin for the bad guys.

 

[bononos]

You could run gpu-z, realtemp or some other monitoring utility that will tell you the gpu core/ram clocks and load/utilization. The core should be 157Mhz at idle (eg browsing with hardware acceleration off) for the 5870 I believe.

 

[tweakboy]

You gotta be brave to do this. Especially on a 500 dollar card. Not worth it,

 

[sjankis630]

Well I just finished scanning my system for Malware(malware bytes) and virus (Avast), as well as RogueKiller and nothing. Both are up to date whatever that tells you.
I used GPU-Z and here are the readings. GPU load 99%.
I pulled the card and blew it out but I have not taken the back off because I don't have the screwdrivers tiny enough to do the task.

 

[mikeymikec]

"GPU load: 99%" is highly suggestive. Try using the latest version of "Process Explorer" (available from the MS site), run as admin, check the GPU graph and it should tell you what process is using so much GPU time.

Another tactic I'd consider is upgrading the graphics drivers.

 

[sjankis630]

Well I appreciate all of the help thusfar.
I have downloaded and run Process Explorer.
I selected the GPU graph as instructed and what it shows me is exactly opposite of what I expected.
GPU usage is 1.15-3%
Mem dedicated 84mb out of 1,034,080
GPU sys mem 66mb out of 1,835,008
**********************************************************
**********************************************************
update now my temps have dropped back down again.

Very strange. what would cause behavior like this? I have not rebooted or anything. I am afraid to start a game at this point for fear of "sticking" it back up to the redline again.

 

[T_Yamamoto]

I would just try removing shroud and changing thermal paste, it always works

 

[sjankis630]

As you and RussianSensation have suggested a little operation could prevent future problems .... but what are the odds I could f*** it up?
Regarding taking off the old thermal paste - is it just a general wiping with Isopropyl 90-99% alcohol which gets the old off or does it take some elbow grease?

 

[blackened23]

Don't worry, you'll be fine. It is extremely easy to do. I think alcohol should be fine, although I typically use TIM removal solution.

Key points from the top of my head:

1) Be very careful and gentle when removing the shroud screws - you DO NOT want to strip them.
2) Remember to plug the GPU fan back when re-applying the shroud
3) Remove dust, re-apply TIM, voila - temperatures are like new again.

 

[VirtualLarry]

If the GPU load is showing as 100%, then this is a software problem, not a problem with cooling, heatsinks, or anything else.

I still think that it could be bitcoin malware. Malwarebytes doesn't do rootkits, that I know of. I think that you either need heavy anti-malware artillery, or you need to reformat.

 

[blackened23]

^ Very valid point. Your usage should definitely not look like that during idle, I glossed over that post initially.

 

[T_Yamamoto]

As you and RussianSensation have suggested a little operation could prevent future problems .... but what are the odds I could f*** it up?
Regarding taking off the old thermal paste - is it just a general wiping with Isopropyl 90-99% alcohol which gets the old off or does it take some elbow grease?

Odds are slim, I've done it multiple times, first time being on this 9800gt I'm using. It's very simple.

 

[notty22]

If the GPU load is showing as 100%, then this is a software problem, not a problem with cooling, heatsinks, or anything else.

I still think that it could be bitcoin malware. Malwarebytes doesn't do rootkits, that I know of. I think that you either need heavy anti-malware artillery, or you need to reformat.

Also agree. Something was pegging the gpu usage. That caused the heat. You have to find why that is happening. Just leave gpu-z running

 

[bononos]

As you and RussianSensation have suggested a little operation could prevent future problems .... but what are the odds I could f*** it up?
Regarding taking off the old thermal paste - is it just a general wiping with Isopropyl 90-99% alcohol which gets the old off or does it take some elbow grease?

I wouldn't bother changing the thermal paste before you get the high gpu utilization sorted out. Why mess around with something that isn't the proximate cause of your problem and possibly breaking something even if that chance is slim.
It looks to me the overheating is occuring because the gpu is running at max speed while the fan is only spinning at 45%(odd). vram usagel is low, normal for 2d. My wild guess is that it could be a driver issue and back down to 12.* maybe 12.1 which wouldn't hurt. You probably need to uninstall the driver instead of just installing the new version over the old one.

 

[bononos]

I would just try removing shroud and changing thermal paste, it always works

No it would not, he has a different problem.

 

[sjankis630]

OK that sounds reasonable. It is very late here and right now i am not near that computer. I have to finish up some late stuff tonight and tomorrow.
So the advice seems to be it may be the drivers which are effecting the temp.
Just for the record it has been running cool ever since about 5pm today but I have not run any gpu intensive apps - like games - yet.
When I have time next week I will :
1) uninstall the current drivers
2) find the drivers from say the last series - before current - and install them.
I have always heard that AMD drivers weren't much but i have not had to upgrade the drivers much because they don't put out new ones that often.
I think my next card will be an Nvidia card as although they put out drivers seemingly every week, I have not had as many problems with them when I last used an Nvidia card. (sometime back in 2010).
Thanks again to everyone for all the help, especially on such short notice.

 

[3DVagabond]

Something was using your GPU. It was not idling. That's why you have 3D clocks and 99% GPU usage. I've never heard of drivers causing anything like this. There was some app running. Guys telling you that you have some sort of malware on your computer are likely correct. You need to find out what it is and sort it.

 

[mikeymikec]

I'd keep an eye on the GPU load figure in Process Explorer to see if it goes back up again, and see what process is causing it.