Cycling GPU Usage after GPU Code 43 problem/"Force to reinstall graphics driver"

[rich_27]

Has anyone experienced this before: https://i.imgur.com/IFsBtHL.png? GPU usage seems to drop substantially every so often, causing the fans to spool up and down repeatedly. Is this normal, I wonder if it's to do with YouTube buffering and decoding?

Normally something like this wouldn't set me on edge, but the cyclic change in fan noise is annoying and, more importantly, my GPU started playing up last night; after playing with the pump voltage on my CPU AiO trying to reduce the high pitched pump noise, I managed to overheat my CPU to the point where the automatic 'we need to shut down' message came up. I shut everything down and reset the pump fan curve, but when I booted up again the GPU was acting up and in Device Manager it said 'Windows has stopped this device because it has reported problems. (Code 43)' for the 3070. I tried reinstalling device drivers, but Windows kept popping up the notification 'Force to reinstall graphics driver'. After some searching, I stumbled upon some discussion of it potentially being a PCIe issue (https://www.reddit.com/r/techsupport..._to_reinstall/ and https://www.reddit.com/r/buildapc/co...stall_on_3080/), and I tried setting the BIOS PCIe version setting from Auto to PCIe 3, which fixed the issue. The weird thing is this CPU, mobo, RAM, SSD, and GPU had all been working fine together since Friday, and had been working great with the new PSU for a few hours before it happened (Silverstone SX500-L to a Corsair SF600). I find it strange that it appears to be a PCIe compatability issue, because my CPU, mobo, GPU, and riser cable should all work fine with PCIe 4, the DAN Case website specifically mentioning that the riser cable is PCIe 4:

Risercable	3M™ 8KC3-0726-0300
Speed	PCIe® x16 Gen4

from: https://www.dan-cases.com/dana4_spec.html

I wonder if I've damaged my mobo, riser cable, or GPU with the CPU overheat, any advice on how best to figure that out and whether this GPU cycling is normal? I have no other PCIe 4 devices to test with, unfortunately. Thanks!

[cybermarketingservices]

I would highly recommend trying to test the set up without a Riser Cables. From past experience, Risers have been prone to cause malfunctions from the slightest of errors.

[Millennium]

I think if the GPU is failing after overheating, even slightly, it's probably toast - sorry. Are you still in warranty for a watercooled card? Maybe you could test it in another system to rule everything else out?

Bad timing, anyway!