Super inconsistent pc crashes, suspect riser or gpu problem

[anemonk]

I've had a few weeks of my pc being super unstable at this point. It was rock solid, with the occasional crash every now and then, but now I have at least a crash a day

 

At this point, I suspect either my riser cable or the gpu itself. Currently trying out the pc without the riser, but wanted to hear if anyone's had similar issues or has anything else I can try. I know this is a long post, tried to include anything I thought might be relevant. Thanks in advance!

 

Symptoms:

 

Varies, but the most common is for the framerate in-game to suddenly chug (from 120 -> 2), then for both monitors to freeze (I can still hear in-game audio, my mic/discord call seem to work for a few seconds), and then the pc eventually restarts (sometimes within 30 seconds or so, sometimes doesn't restart at all). I've noticed it most while gaming, but I've had it happen while watching a youtube video or using text chat

 

Checking event viewer usually shows some nvlddmkm errors, though sometimes there isn't anything, and checking the minidump suggests some daemon tried to reset the video driver, timed out, and triggered reboot

• Have also just had both screens immediately turn black, no audio, pc reboots immediately
• I've had 2 really odd instances where my main display freezes, my secondary monitor stays alive, and un/re plugging the dp cable to my main monitor fixes the issue for a few mins at least (though I believe both times it eventually crashed).
• I've tried keeping an eye on hwinfo. Temps, power look perfectly fine all around. Have about 100w of overhead (I imagine this is sufficient, even for intermittent power spikes)
• I have not had a bsod ever, thought that was weird, don't know if relevant

Attempts to fix:

• Reset gpu and bios settings to stock (used to run with PBO2 at -20 and a mild gpu undervolt), didn't help
• Multiple OS reinstalls
• Tried multiple GPU drivers 531.41 and 531.79, used DDU
• Ran a memtest86, all good
• Swapped to an 850w psu, no luck
• Ran a load test (furmark) for a few hours, totally fine. Given that it's failed outside of games, I feel like its not load related
• Tried swapping to a different DP cable

Parts:

• 5600x
• RTX 3060ti (Asus KO)
• Asrock b550 phantom gaming itx mobo
• 32gb corsair LP RAM
• Corsair SF450 PSU (I know this is less than recommended. It's carried forward from a 2060 super build, and its been fine for months. I did try a beefier power supply and it didn't help)
• SSUPD Meshlicious, with the pcie gen 4 riser

I haven't changed anything about the PC recently, but I did upgrade my monitor recently. Can't imagine what that would affect though

 

At this point, I'm thinking either the GPU or the riser cable are bad. I'm currently running with the pc built outside the case and haven't seen a crash yet. My goal is to either be sure its the cable, and just replace that, or have enough evidence to RMA the GPU

 

Thanks for reading! Any help is appreciated, let me know if I can provide more info