gpu? psu? or something else?

[bmx6454]

hello, i have been trying to diagnose an issue I'm having with my new vr setup, a rift s. it frequently crashes in steamvr, although it has crashed in the same way in oculus app once or twice. when it crashes, the image in the display will freeze solid, and the entire pc will lock up. i can usually get the pc to unfreeze by hitting ctrl+alt+del, and waiting for the pc to respond once task manager comes up, sometimes taking longer than a minute. every time it comes back from the lockup, steam says add ons were blocked, and i have to restart and re-enable to use it again. sometimes it will run fine for like 30 minutes or more, usually crashes within minutes. i can't replicate the issue in any benchmark. i am currently running occt for cpu, and furmark for gpu at the same time on max settings, no issues as of 2 hours of running... i need to test this further, but it seems to be fine in steamvr if i undervolt and or underclock the gpu(not sure which yet), although performance dips hard, making me lean towards gpu or psu, also those are the only 2 parts that haven't been replaced, so issue has to be one of them imo. chasing down this issue has cost me a lot of time, and I'm still not really close to a solution, so any and all input is very appreciated.

 

list of things Ive tried:

-new cpu, ram, and mobo(bought for other reasons, hoping it would also fix the issue)

-fresh os installed(win10 pro)

-tried a different ssd

-ddu, and tried older drivers

-tried a powered usb hub for the headset instead of direct mobo input

-repasted gpu, much better temps, not the issue though

-ruled out any overheating

-checked voltages under benchmarks, 12v dips to 11.788, 5v to 4.98v

-likely more I'm forgetting, will add if i think of any others.

 

current specs:

-asus tuf gaming x570 plus (non-wifi)

-3700x

-g.skillz trident neo 3600

-rog strix gtx1070

-h60 aio

-970 evo plus 500gb (game storage)

-240gb sata ssd (boot drive)

-1tb 7200rpm seagate

-500w evga b 80+ bronze

 

the psu is my main suspect atm, i would have replaced already, but due to covid markup, can't get the one i want for a reasonable price. don't have another to test with unfortunately. if i can confirm the psu for sure, i will gladly pay the markup to fix the issue. hoping someone knows of a better way to replicate this issue in benchmark. sorry about the long post, thanks for any and all input in advance.

 

[bmx6454]

edit: psu, encase anyone else ever encounters a crashing issue under similar circumstance.