Question GPU only works properly in bios not elsewhere

[Elliot Alderson]

Hey everyone! First of all please forgive me for this weird formatting. I am writing all this on my phone because it’s hard to use my laptop at the moment (I will explain below).

Brief idea on background:
So I have this Eluktronics THICC 15 That I got a couple years ago. Have no complaints about it except for its fan noise but not a big deal.

As I mentioned about the fan noise, I figured it was a great idea to clean up the inside to help with airflow. I unscrewed everything and wiped only the “safe area” with disinfectant wipe only the plastic and copper. I also had to unplug the power cables for the GPU and CPU fans(I know… it’s probably my first mistake)

I then let it dry and screwed everything back in place. I restarted the laptop and everything worked great for about 15 minutes. At of a sudden I got a windows death screen. It was too fast I couldn’t take a picture of it. Then once I tried to reboot it, my screen started to have artifacts. It happened during booting, windows loading screen and inside windows. Except it works fine in bios.

HDMI and DP ports stopped working as well.

I got this error in device manager and of course I immediately tried to uninstall and reinstall GPU driver. Didn’t work so I reinstalled wondows. Didn’t work so I updated bios even. Nothing works!

Hardware wise. I tried unplugging everything I could: rams, SSD, Cmos…etc
Both CPU and GPU were not hot so I doubt it has to do with overheating. Also, I didn’t see any thermal paste leak on the board.

No problem in bios at all. Still couldn’t use HDMI and DP ports but that’s normal in bios.

I tried updating the bios. Didn’t work.

It is a tough year for a lot of people and I’m like others, I can’t just spend a couple thousand dollars on a new PC. I would really appreciate any help from you guys.

 

[CodeConnoisseur]

My immediate thought is that the GPU was damaged during the cleaning of the fans given the characteristic horizontal squiggly lines and nonfunctional ports, but we can try to rule out everything else regardless. Specifically one of the VRAM chips surrounding it may be damaged (look up Google images for "VRAM artifacts").

Do you have the ability to test it in Ubuntu via a live USB perhaps? Certainly interesting that it seems to only be happening in Windows and on that firmware flashing screen, though it's likely because the BIOS runs a set of drivers that don't take full advantage of the GPU and as such the failure condition doesn't make itself apparent in that limited set. It also uses less resources, so with the bad VRAM theory it's possible that the bad regions are not being used.

Assuming that no changes have been made since you cleaned the fans they seem to be detected and cooling just fine so I doubt that whatever happened was caused by overheating. It's pretty clear to me that it's not LCD damage or the display cable otherwise it'd be constant. Might there be a minidump for the blue screen in C:\Windows\Minidump or a bigger C:\Windows\MEMORY.DMP?

 

[Elliot Alderson]

My immediate thought is that the GPU was damaged during the cleaning of the fans given the characteristic horizontal squiggly lines and nonfunctional ports, but we can try to rule out everything else regardless. Specifically one of the VRAM chips surrounding it may be damaged (look up Google images for "VRAM artifacts").

Do you have the ability to test it in Ubuntu via a live USB perhaps? Certainly interesting that it seems to only be happening in Windows and on that firmware flashing screen, though it's likely because the BIOS runs a set of drivers that doesn't take full advantage of the GPU and as such the failure condition doesn't make itself apparent in that limited set. It also uses less resources, so with the bad VRAM theory it's possible that the bad regions are not being used.

Assuming that no changes have been made since you cleaned the fans they seem to be detected and cooling just fine so I doubt that whatever happened was caused by overheating. It's pretty clear to me that it's not LCD damage or the display cable otherwise it'd be constant. Might there be a minidump for the blue screen in C:\Windows\Minidump or a bigger C:\Windows\MEMORY.DMP?

Thank you for the response! Now that I think about it maybe you’re right. As the artifacts change its “style” based on the action happening on screen. For example I notice if the background is pure white, everything seems to look slightly better and if I move the cursor, artifacts surrounded it would “update” (pictures below)

Also, Once I booted into Ubuntu, my HDMI and DP ports started working again. Even though they still have artifacts.

(Laptop screen)

(DP output)

(HDMI output)

And for blue screen log. Unfortunately I didn’t know log exists in those locations, so I had a clean windows install wiping them out already.

If it’s a VRAM issue, how likely is it possible to be repaired? I know it’s case by case but I don’t think it’s as easy as swapping out the damaged VRAN could solve the problem…… As in I don’t think it is even worth it to bring it to an repair store… any other suggestions would be appreciated!

 

[Eximo]

I know it’s case by case but I don’t think it’s as easy as swapping out the damaged VRAN could solve the problem…… As in I don’t think it is even worth it to bring it to an repair store… any other suggestions would be appreciated!

Pretty much what would have to be done. Repair costs are likely to be the several hundred dollar range, plus shipping. Basically they have to identify the bad memory chip, and replace it, then test it. If the problem is in the circuit board, then they have to troubleshoot that and make a repair. If a circuit board repair is not possible.

For an RTX2070, probably not worth doing, you can likely find a motherboard in working order for about the same amount. Speaking of that, someone would almost certainly be looking for the screen/chassis/battery from that laptop. So you can easily sell it as spares if you aren't interested in swapping in a new board.

 

[Elliot Alderson]

Pretty much what would have to be done. Repair costs are likely to be the several hundred dollar range, plus shipping. Basically they have to identify the bad memory chip, and replace it, then test it. If the problem is in the circuit board, then they have to troubleshoot that and make a repair. If a circuit board repair is not possible.

For an RTX2070, probably not worth doing, you can likely find a motherboard in working order for about the same amount. Speaking of that, someone would almost certainly be looking for the screen/chassis/battery from that laptop. So you can easily sell it as spares if you aren't interested in swapping in a new board.

Thank you for your quick response! I actually saw that coming, so I guess this thread can be a reference for people in the future to troubleshoot similar issue. In this case, it’s likely VRAM issue.

I am not really familiar with laptop hardware and I did a little eBay research. I don’t think many people on there are selling just the board but the whole laptop or other parts. I will probably just keep the CPU, SSD and m.2 while selling the rest in parts.

It was a monster with Ryzen 9 3900 back in the day. I know it’s not really a “laptop” as it wouldn’t survive on battery for more than 10 minutes, but for my use case it was totally worth it (I had to move country for work and didn’t want to ship my tower at the time with me so having a power laptop which you could bring on the plane was the perfect idea.)

Again, thank you everyone for your help!

 

[Eximo]

Yep, I think we are seeing the tail end of the desktop CPUs for mobile. The big mobile processors from AMD and Intel are plenty good for mobile workstation purposes. And since you can get as large as an underpowered RTX4080 desktop equivalent out of a mobile 4090, the graphics power is there too.