Problem with Graphics Card or PSU?

Toggle sidebar Toggle sidebar
T

thomaswp

Commendable
Oct 6, 2016
12
0
1,510
My PC crashes whenever it renders advanced graphics - the screen goes black and the keyboard/power buttons do nothing until a hard reset. For example, I can get through the Windows 10 loading screen, but as soon as I hit the login page it crashes. Safe mode works fine, though. I know it's not (just) a driver issue because the same behavior occurs on Windows 7, 10 and Ubuntu (on separate hard drives). More details on this thread.

I assume the problem is with the GPU. It's a Sapphire Radeon HD 7870 and about 2.5 years old. I've noticed the rear fan stops immediately after spin up on startup, as described here.

But before I replaced it, I wanted to make sure the problem is the GPU and not another component, like the PSU or motherboard, perhaps failing to give it sufficient power to render advanced graphics. Or if perhaps the fan is a cause and not a symptom, and fixing it would solve my problems. The PSU is only a few month old, replaced after my old one (same model, Raidmax RX-530SS - 530W) burned out in a power surge (which may be responsible for the GPU problems as well).

Is there any way to confirm that the GPU is to blame? Unfortunately, I don't have another GPU/PSU to test with. Any help is appreciated!
 
Sort by date Sort by votes
T

thomaswp

Commendable
Oct 6, 2016
12
0
1,510


I'm not sure how I can check the GPU temps when I can't boot to anything but safemode. When I run HWMonitor in safe mode, I get the processor and HD temps (27-35C), which seem reasonable. But the video driver isn't engaged (which is why it doesn't crash), and HWMonitor doesn't detect the card. Any suggestions for how to get around that?

What's so bizarre to me is that everything is fine in safe mode. I'm actually writing this reply on the affected PC. As soon as the video driver engages, though, instant black-screen. Which is why I'm skeptical that it's an overheating issue, since it's always instant, not caused by time. Could have been originally caused by overheating though.
 
Upvote 0 Downvote
V

veldrane2

Reputable
Dec 8, 2015
89
2
4,665
Hmmm

It could be the temps but I don't think so. Open your case, put a table top fan on your computer and run it open, see if it makes a difference. If you can stay on login screen longer or even get into desktop. The computer will still crash but that should be your indication for the temp issue.

Other then that, open your PC, remove everything, and I do mean everything. Mind your RAM sticks, make sure you do not put them back into wrong slots, so don't mix them up. Clean up all ports, fans, their respective power connectors, check for anything out of ordinary, spin the fans by hand, are they all absolutely free to move ?

Check for bent pins. Sometimes its a good idea to clean and re-apply thermal paste on CPU, it DOES make a difference, though I do not think in this case.

Also, try a different cable and make sure it is not bent or looped.

When your PC shuts down, if its the mobo itself that completely goes dark, check that you have adequate and properly connected PSU, if its just the display that goes dark but the PC itself is still running, try a different monitor, if you don't have one try it out on your TV via HDMI. It could be the monitor as well.

I am assuming this system has been built a long while ago, but if it hasn't, then you may have some sort of hardware incompatibility.

From the screenshot in your otehr post, that is indicative of VRAM gone bad, sorry, but thats what it looks like at its beginning stages. As it continues, even if you do get it to work, you will see more frequent and other types of artifacts and you may take out your mobo with the use of the GPU under these conditions.

Also, try to completely uninstall the AMD drivers, use the AMD clena up utility + manually delete the respective install directories + CCleaner for registry and anything leftover. Do not use the guru3d utility under Win 10. There have been problems with it as it took out a bunch of files it wasn't supposed to.

After that, try using microsoft drivers, the generic GPU ones, go to device manager and switch them manually, then try reboot into regular windows. If the hardware functions fine on generic drivers, then you know you just had a bad driver.

 
Upvote 0 Downvote
T

thomaswp

Commendable
Oct 6, 2016
12
0
1,510
Thanks for your detailed reply, veldrane2.



This did work to get the computer to boot fully, which is good to know. The processor takes over to render graphics and the computer runs fine (if slowly). However, when I reinstall the drivers (and the GPU kicks in), it crashes again. Again, I'm 99.9% certain it's not a driver problem, given that a pendrive Linux distro has the same problem.

I'll start following your advice on dismantling the PC, but is there anything that will tell me for sure which piece of hardware is failing? Again, my assumption is GPU or PSU, and given that the GPU has a dead fan and the PSU is quite new, I'm assuming GPU. But of course the PSU could be causing the fan problem as well, I assume.

My plan moving forward, barring any surprises when I clean the machine, is to replace the GPU. Does that sound like a reasonable conclusion?
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom