Originally Posted by Streetdragon
So you get this problem only when you play Battlefield? If the cou alone is stable, i would say the GPU could have a little problem.
I would reinstall the Driver and after that check the PSU. Dont know what PSU you are using
Correct only BFV.
GPU drivers are fresh and GPU is running without an OC currently.
PSU seems to be fine? Corsair RM80i
Originally Posted by spooklexity
What kind of errors is HWinfo throwing, though? Are they CPU related? What type of cooling are you using? It might be that with high load on all components, you are reaching thermal conditions leading to instability in some subsystem (for instance, VRM temperatures increase to a point where the VDroop increases noticably). But without knowing more that's a pretty vague assumption.
Other than that:
To 1) I would recommend to use and old p95, manually disabled p95, or OCCT's Linpack with AVX disabled, combined with everyday tasks. Realbench uses AVX in part, but also has parts without it and loads all subsystems simultaneously, therefore I consider it to be very useful under such conditions.
2) Karhu RAM test or gSAT tie for rank 1, other than that memetest86 or HCI memtest would be my recommendations for testing RAM very specifically
3) Difficult IMO, you can have meta-stable cache configurations that lead to catastrophic errors only sporadically. I would use a three-stage test, starting with the AIDA cache stress test, then Realbench, then p95 in-place large FFTs. Make sure cores and RAM are really stable if you don't want to end up in wild mass guessing here.
4/5) Linpack Xtreme, hands down. You will find nothing that comes even close, and it may punish any slight unstable configuration with an instant bluescreen.
HWinfo throwing L0 Cache
I'm running a Delid chip with a custom loop cooling solution. Temps never above 65*- 68* at 1.3V
Voltage seems to be steady as a rock at 1.312v under all load conditions.
Originally Posted by blodflekk
My go to method is Intel burn test with hyperthreading disabled, 20 passes. If all good move on to p95 26.6 large FFTs for 12 hours. If you can pass both these then the issue has to be RAM, GPU or PSU.
Currently going through suggestions. Only at core frequency and voltage right now. 1.3V @ 5Ghz
Passed 30 Minutes of p95 small
Passed 2 hours of p95 1344 in place
Going to let large run for the rest of the day.
If it passes I'll enable XMP and start the process over?