About a year ago, I added a 9800GT (1 GB version) and a Corsair CX500 PSU to an HP M8000N computer. A few weeks ago, the HDD overheated and I decided to transfer the GPU & PSU to a new build, which consists of:

Once I had Win7 up and running, I installed all the essential drivers that came with the Gigabyte Mobo CD. However, whenever I tried installing the Graphics Media Accelerator driver, the computer would crash and enter an endless boot sequence on the next startup. I skipped installing this driver and installed the CD driver for the 9800GT, which by now is a year old.

Everything was working fine, WEI rated my GPU at 6.6 graphics & aero performance. However, after updating my Nvidia drivers to the latest, the WEI dropped my rating to 3.3 for Aero, and 4.7 for graphics performance. Just to make sure that everything was ok, I ran Bad Company 2 on medium settings. The first few minutes ran just fine at a smooth framerate, so I dismissed this as Windows being Windows.

About 6 hours later, I ran BC2 again. This time I averaged anywhere from 2-5 FPS. I checked the GPU temperature through GPU-Z, and it came back as 120C. The problem with this, is that the computer was on for six hours up to that point. Wouldn't the card have experienced a reactor core meltdown a lot sooner than that? Granted, the computer was "sleeping" some of the time, but still...

The next day I took out a temperature gun and ran some tests. I would point the laser at a very specific area on the reverse side of the card (not the fan or "front"), and compare the temp reading with GPU-Z. After leaving the system on idle on idle for a few minutes, I ran BC2 twice. Here are the results:

  • GPU-Z Reading / Temp Gun Reading / Time
  • Null / 22.3°C / Comp is Off
  • 53°C / 33.5°C / 1:49
  • 78°C / 46°C / 1:53 - (First BC2 run; good framerate)
  • 102°C / 64.6°C / 2:01 - (System is again on idle)
  • 113°C / 64.8°C / 2:10
  • 119°C / 71.8°C / 2:17 - (Second BC2 run; poor framerate)

I should also mention that I also took a temp recording of another part of the GPU from 2:01-2:17. The temp in this area jumped from 75°C to 82.9°C in that time frame.

This pretty much confirms that GPU-Z is reporting the temperature accurately, and the card is overheating. But I'd like to know why; the cars is doing nothing and still the temperature climbs at a steady rate. I thoroughly cleaned the GPU and PSU when I salvaged them from the old HP M8000N computer with a can of compressed air, dust cant be the issue. Similarly, the rest of the computer is brand new. I installed various Nvidia drivers, but no luck. It seems strange to me that a year-old card is suddenly failing on me; aren't they supposed to last at least two years?

Could this be a driver issue? Is the motherboard faulty? Could the PSU be overfeeding the card on voltage? Neither case seems likely, as the CPU, RAM and otherwise the rest of the comp has worked flawlessly and has stayed well within respectable temp ranges (the i3 lingers around 50C, the HDD stays at 30C, so does the PSU). How can I pinpoint the issue?

  • What is the case temp and ambient temp (in the room)?
    – Paul
    Nov 26 '12 at 2:39
  • I aimed the gun at various parts of the cases when it was off, came back as 20C or so. The case isnt shoved into any enclosed space as well, its out in the open with plenty of room to breathe. Room temp is 70 Fahrenheit. Not particularly warm, plus Im in cold, cold Washington at the moment.
    – JAD
    Nov 26 '12 at 2:43
  • 1
    tl,dr. but it seems cooling on the card is the issue. either at the heatsink to gpu contact, or fan speed. Either way, climbing temp when the card is idle can be only that (barring really obscure issues). Generally, your temps are sky high though. I think my card hits 60-70 under heavy load. Yours is pulling that at boot.
    – Sirex
    Nov 26 '12 at 3:02
  • I agree with Sirex, if you can see the fan spinning, then perhaps the thermal paste has degraded. You can maybe dismantle and fix this yourself, but the best bet is to use this as an excuse to upgrade the gfx card :)
    – Paul
    Nov 26 '12 at 3:30
  • Youre both right, in that case its most likely the thermal paste drying up. That, and the 9800GT runs pretty hot to begin with. The one thing annoying me is that this was not an issue with the HP computer, just this particular build. Any card recommendations regarding the 650 vs 7700?
    – JAD
    Nov 26 '12 at 3:44

Such extreme temperature is probably the failure of the video card's cooling system. Possibilities are:

  • fan is failing and rpm has dropped to a non-effective level.
  • dust has built up too much
  • gpu chip contact is broken off with the heat sink
  • room temperature is too high

It could be one or a combination of these issues. Since 9800GT is an almost 5 year old video card, if you had it running all the time, the card is probably at the end of its useful life, unless you are good with replacing video card cooling systems. It's probably not worth it to get it repaired by a professional place.

  • If the room temperature was "too high", wouldn't we expect to see much higher temperatures with the system turned off than about +22°C? +22°C is only on the slightly warm side of what is usually considered normal room temperature.
    – user
    Nov 23 '15 at 21:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.