Hard resets playing games

[Night]

Try stressing the memory using TestMem5 v0.12 using Anta777 Extreme config, could as well be faulty RAM.

 

[tabascosauz]

OK! So this beta BIOS is promising! No resets at default settings and not a single WHEA error yet! It's a little disconcerting that P95 still fails (both small FFT and large FFT) so only time will tell if that really is an indicator of an unstable system. Every other stress test I've run has come out clean (but that was true with the old BIOS version as well). I'm going to try getting IF and memory speeds up over the weekend and see how that goes.

Did those "other stress tests" include the aforementioned LinX amongst them? That's the only other CPU stress test I'd say comes close to P95 Small's workload on the cores. If a similarly long period of strenuous testing in LinX yielded no problems, I'd probably feel a bit better. Otherwise, something's really up here because there's zero reason why you should have to disable AVX or do anything else to make Prime95 run as intended. The PB algorithm is beyond mature in Q2 2021 and automatically accounts for the degree of load, type of workload, etc.

Especially since both Small and Large are failing, that's kinda alarming. The two are nothing alike. Large hammers the memory and UMC, but it's pretty light on your cores even though it says 100% utilization. Small tortures the CPU cores but basically doesn't touch memory. That's part of the reason why LinX (well, if you have the right settings) works well, it does both.

When P95 is crashing, what does it look like? Do individual worker threads stop? Does the application crash? Do you get an abrupt reboot without warning?

 

[RJARRRPCGP]

Did those "other stress tests" include the aforementioned LinX amongst them? That's the only other CPU stress test I'd say comes close to P95 Small's workload on the cores.

Linpack, at least in the past, seemed to more consistently give a major core workout, IMX. (But that was with Intel, and Linpack possibly has false errors with AMD, I ended up staying away from Linpack on AMD processors, so far) OTOH, Prime95 in blend, is the only one that I can confirm uncovering bus errors. ("Bus/Interconnect Error" was rearing its ugly head with Prime95 blend)

 

[flashfrenzy]

Especially since both Small and Large are failing, that's kinda alarming.

I was incorrect in my statement. I remembered now it was large and blend that were failing (which makes sense). I re-ran smallest FFT for 4 hours without any errors.

When P95 is crashing, what does it look like? Do individual worker threads stop? Does the application crash? Do you get an abrupt reboot without warning?

It's failed workers. The only hard resets I've experienced are in-game.

Did those "other stress tests" include the aforementioned LinX amongst them?

Yes, I've run OCCT (both memory and CPU), Linpack, AIDA64, and quite a few different memory tests. The only test that has shown errors has been prime95 blend and large FFT. Obviously this has all been to try to track down the source of hard resets. Since updating to the beta BIOS, I've experienced no hard resets. Prior to the update, I could turn on Tomb Raider, leave it paused, and within 2 hours the system would reset. I had it paused for most of the last 2 days and have experienced zero resets. I even re-enabled DOCP at default settings all day yesterday and have seen nothing. So I guess this firmly points to board/firmware being the original issue? I'm really hopeful that this stability is real.

As an aside, something definitely changed in the thermal throttling and/or CPU voltage. Even at full load, it never goes above 80-82C now (even in P95) whereas on 1804 it would always go up to 90C. I'm not complaining too much. It looks like I lost a couple hundred points in Cinebench multicore but that's a small price to pay for system stability.

Also, here's the default DOCP settings for BIOS version 2006. VDDG IOD seems high but I haven't compared this to 1804 to see if anything changed.

 

[RJARRRPCGP]

I was incorrect in my statement. I remembered now it was large and blend that were failing (which makes sense). I re-ran smallest FFT for 4 hours without any errors.


It's failed workers. The only hard resets I've experienced are in-game.

99.9 percent of the time, those reboots mean Windows bugchecked, look for a bugcheck entry in the event log. Sounds like a BSOD without seeing the BSOD on your monitor.
Especially if you got an error when running Prime95. Windows often will just reboot itself, that's the default behavior since NT 5x.

 

[flashfrenzy]

99.9 percent of the time, those reboots mean Windows bugchecked, look for a bugcheck entry in the event log. Sounds like a BSOD without seeing the BSOD on your monitor.

Yeah, that was always the first thing I looked for. There was never anything in event viewer for any of the resets.