GPU crashed, cannot activate

[redleg_runner]

I'm running an Alienware Mx17 with a Radeon 7970M card, running Windows 10.

I was tweaking my streaming setup using OBS and Twitch, running Overwatch in borderless windowed mode. I had Overwatch up, as well as OBS, and I alt-tabbed to my Twitch dashboard (which displayed my stream preview), and I guess the three video renders were too much, and the computer locked up with a black screen.

After that, I was unable to boot into Windows. I could boot into Safe Mode, however. Through BIOS I was able to reset to my Intel integrated HD 4000 graphics chip and boot normally into Windows. Device Manager showed both the Intel and the Radeon display drivers, as normal.

Trying to switch to the GPU, my troubles really began. Windows would restart, but trying to launch any game would freeze the computer and the only way to get back into Windows was a restart/BIOS reset. I was able to do a clean reinstall of the Radeon drivers, and as long as I was on the integrated graphics, no problem.

Changing to the Radeon card, however, would drop both the GPU and the integrated graphics from Device Manager. No display devices would be displayed, only "VGA device" under the "other" category in Device Manager...and trying to do anything (e.g., starting just about any program) would freeze the computer and need a restart.

I've reseated the card, cleaned out the computer, updated BIOS... just about every solution I can find online. I'm at my wit's end.

Is there any remedy that I haven't tried?

Is my card fried and I need to buy a new one?

What caused this / how can I avoid it in the future? (I didn't think I was doing anything grossly unusual...)

Thanks for any help you can provide!

 

[TechyInAZ]

There is only one option I can give you that might just work, reinstall windows 10 completely.

You didn't do anything wrong, which is why it's so weird you got this issue. However that is an old laptop by today's standards.

 

[redleg_runner]

There is only one option I can give you that might just work, reinstall windows 10 completely.

You didn't do anything wrong, which is why it's so weird you got this issue. However that is an old laptop by today's standards.

Yeah, it is nearly four years old. Want to squeeze another year out of it before upgrading.

I wasn't looking forward to a complete reinstall, but it seems that's what I'm going to try next. Thanks!

 

[redleg_runner]

So, by way of update, I finally got Win 10 reinstalled and working (had to manually install Intel and AMD graphics drivers through safe mode), so I can operate the computer.

Device manager displays both display adapters, and everything seems fine....except I can't switch to the Radeon card.

Fn-F7 does nothing. I've uninstalled and reinstalled the Alienware On-screen Display multiple times, setting it to run as admin, etc., but still...no Fn-f7 response.

If I can't get the computer to use the GPU, then I can't tell if it's faulty or see if it will lock up.

 

[redleg_runner]

So, final update, I beleive.

After a full Win 10 reinstall, I fought through multiple BSODs, attempting multiple driver installs. The combo that finally worked was using DDU to do a clean driver install of just the base driver (no extra AMD software).

Well, it 'worked' for about a day. It runs fine without a load (i.e., regular computer or non-graphics intensive games like Hearthstone), but once it's stressed, it crashes. For example, Overwatch will start and run menus, but once it's finding and loading into a match, it locks up with gray or colored bars on the screen and horrible buzzing sound. Sometimes the computer will recover without a restart.

So, I'm down to heat likely being the issue. I'm going to reapply some thermal paste and new thermal pads, so we'll see if that does the trick!

 

[TechyInAZ]

I hope so, what temps are you running at?

 

[redleg_runner]

I hope so, what temps are you running at?

45-50 C idling with no load.

50-55 C with a light load (~25% or so... i.e, starting Steam or the Battle.net app)

Launching a game and getting into the 60 C range is when the aforementioned problems start.

So, maybe the card is just shot?

 

[TechyInAZ]

That is a very low temp for an AMD GPU. Something else must be a miss. I'll bring in some help.

I've been thinking, try running a CPU stress tester. If that runs fine then the GPU is defiantly the problem.

 

[redleg_runner]

That is a very low temp for an AMD GPU. Something else must be a miss. I'll bring in some help.

I've been thinking, try running a CPU stress tester. If that runs fine then the GPU is defiantly the problem.

Thanks, the assistance is much appreciated!

Stress test crashed, with the following BSOD report: