4090 Strix OC Edition Recently Stopped Working

[Petroph]

I'm in a little bit of a pickle here. I recently noticed that my pc has been having issues booting past post every once in awhile, but never really thought anything of it, as it happened maby once or twice every 3 or so months. Recently, I updated the graphics card driver and noticed that the pc failed to boot past post (error 97) several times before I was able to get past it to log into windows to revert the driver. So I went from 581.80 to 591.44 then back to 581.80 from within the nvidia app. Reinstalled the older 581.80 drivers without issue. Was able to continue to play games etc. Everything was normal for about a week or so. Then out of the blue I get error 97 again, and I simply cannot get past the GPU initialization phase at all. I can't get into the BIOS because the GPU initialization fails during POST. I can start the PC and get into windows just fine with the GPU unplugged and still connected to the PCI slot, which is indicating to me that the card has failed. This card is only 1.5 years old. I had a 2080 TI last me 6 years, before I upgraded to the 4000 series. 

I am not sure what more I can do now that I have isolated the card as being the issue. Do I RMA this card and have it diagnosed? I've read horror stories here where the RMA process is nothing but a absolute nightmare, especially since the 4000 series cards are no longer in production I doubt the card would be replaced. I know this card still has a warranty on it for up to 3 years and about 1.5 years into it from purchase. But genuinly concerned that if I RMA this card, it'll just end up creating more stress. The lifetime of this card was abysmal compared to my 2080 Ti that was put through much worse, and just considering dropping ASUS for another brand. Any recommendations?

PC Specs: 

Intel Core i9 14900
CORSAIR DOMINATOR TITANIUM RGB DDR5 RAM 64GB (2x32GB) DDR5 6600MHz
x1 SAMSUNG 980 PRO SSD 1TB PCIe 4.0 NVMe Gen 4
x1 SAMSUNG 990 PRO SSD 4TB PCIe 4.0 NVMe Gen 4
x1 SAMSUNG 990 PRO SSD 2TB PCIe 4.0 NVMe Gen4
x1 PHIXERO P5000 SSD 1TB PCie 4.0 NVME Gen4
ASUS ROG STRIX NVIDIA GeForce RTX 4090 Overclock 24GB Graphics Card
ASUS ROG Strix Z790-E Gaming WiFi II
1200W Platinum II Power Supply

[Silent_Scone]

Hi @Petroph 

I wouldn't be so quick to assume it's the GPU if it's otherwise ok after POST.

Q-code 97 is usually PCIe/GPU initialisation, so I’d suspect GPU power / PSU cable before anything else.

Initial steps: Clear CMOS via the button on the motherboard's rear I/O panel.

1. Reseat the GPU, reseat the PCIe power plug fully (12VHPWR), and if your PSU is modular, try a different PCIe cable/PSU port (don’t use a daisy-chain).

2. If your board has another x16-length slot, try the card there just to see if it can POST.

3. If possible, swap your GPU into another system. If it fails in a second system

4. It may also be due to CPU pin-pad contact. This could be a poor mount. Remove the cooler and CPU and inspect the socket and CPU pin-pad for any debris or damage.

5. If using a third-party CPU contact frame, remove it and reinstall the stock retention mechanism. 

6. Reseat the CPU and cooler whilst ensuring consistent mounting pressure.

[Petroph]

Still troubleshooting. But tried a new cable and reseating everything but so far no luck even when resetting BIOS and updating it. Dont have another rig to test this card with unfortunately but signs are continuing to show that I'm the GPU just wont initialize anymore for some reason. Not sure how I can change the power setting of the gpu when I cant plug it in to adjust it since I cant get into bios without unplugging it. Is there a specific setting in the bios I should be looking for to perhaps down volt the GPU before actually hooking it up?

Edit: I did manage to get past post with the GPU installed and plugged in once. The card worked fine played a few matches of battlefield 6. However. After a system restart error 97 reappeared and back to square one....

[Silent_Scone]

If the system is able to POST with the iGPU, enter the UEFI and change the link mode to GEN3, see if this helps.

[Petroph]

I did manage to go down to gen 4 but no luck. I did not try gen 3 though so will give that a try. 

[Silent_Scone]

It's really difficult to diagnose vicariously without other hardware. Have you tried in an alternate PCIE slot yet?

[Petroph]

Unfortunately, the card is too big to put into another PCIe slot with my motherboard layout and case layout. I could get the card in another slot, however, the case would completely stop the fans from spinning as it would be touching the PSU shroud built into my particular case and I def don't want to break anything else. I am working on getting another GPU out though to see if the PCIe slot has been effect itself, however needing to hook up different power connectors for the older GPU since the 16 pin connector for the 4090 is still pretty new. Will keep updating though, but always open to more suggestions. 

[Silent_Scone]

At this point, may be worth "breadboarding" the system outside of the case anyway, to see if it makes any difference.

[Petroph]

Just a quick update. I managed to get the system through post by switching to Gen 3 then reinstalling the gpu. Managed to fresh install Nvidia drivers after a full wipe of nvidia anything prior to gpu reinstall. Opened up a game played a few matches of warthunder no issues. Card was running fine and what appeared to be normal temps. However the computer BSOD after game exit and automatically restarted with 97 error code again. Kinda at a loss for words. 

[Petroph]

Ok got an RMA setup for the card. But now thinking perhaps its something else. I noticed my chances to successfully post tend to happen when I throw the switch on the PSU for a undetermined amount of time. Power cycle the pc to make sure there is no power in it at all, then throw the switch, wait 5 seconds theb turn on my pc, I have been able to successfully post a few times doing this. So perhaps its the PSU going out? Seems odd that everything works fine after post. Maby its a pc standby power issue that is just set outside of normal operating parameters for success full post? Strange if do that this would be happened more often when I nmhadnt changed any power settings or anything.