Question I am out of things to try. Gigabyte B550m + Ryzen 3600 stability issues (seemingly random)

Dec 4, 2020
2
0
10
0
Hello guys... I'm really out of options at this point. I've tried troubleshooting this problem for weeks and it's insanely hard because I can't reproduce my issue consistently. I'd really appreciate if anyone has an idea what else to try and fix this problem.
I've submitted a Ticket to AMD on monday in hopes that they'll help me (while I still had some ideas to try left). But they did not answer, yet and I'm officially fully out of ideas now.

specs: Gigabyte b550m-D3SH, Patriot Viper Steel 3200mhz (16gb 2x8 dual Kit), R5 3600, GTX 1660super, 600 Watt be quiet! Pure Power 11 CM Modular 80+ Gold

1. So what is the issue?
My PC will sometimes roughly (0-4times a day at ~12h usage). Sometimes it happens a few times in less than an hour ... but that might be by chance. Or not.
The crash doesn't produce a BSOD, nor can I extract any insightful errorcode anywhere. Windows manager logs only event ID 41 with no additional info. (System probably down too quickly to do anything).
After my PC crashes it will just show "no video signal" on my screen, my keyboard lights go off and I can't hear any sounds any more etc. but my fans are still running. So power seems to have been taken away from my whole board but not the fans.

Of course this might all be because of a faulty PSU, CPU or any part of the mobo... but I sadly do not have replacements with me and I also recently moved into a new town because of a job and know almost no one here. So it's really hard to "swap out a part".
Also I have researched and found a lot of similar cases with similar problems. Some apparently were fixed with some setting or uninstalling monitoring software, etc. some weren't. None of the fixes and suggestions I found worked for me, though.
More in this regard later let's move on to observations first.

2. Patterns that seem to have increased the likeliness of my crashes
It rarely ever happened during gaming, but sometimes straight after gaming and sometimes it happened during idle (I went away for 2-3minutes). Stress-testing never seemed to crash my system and the performance seemed in line with the setup I have.
So the only pattern I could observe had to do with a change of system load from higher to lower.

3. Conclusions drawn and fixes tried
Because of this behavior I assumed it might have to do with my system being unable to deal with the load going down, entering various forms of low power modes or just having issues when certain algorithms at play match perfectly into an unstable condition. This might all be wrong and it might even just be the PSU or something else failing but here is what it led me to try so far:
Next to the first few standard things I tested (multiple 5-10min Stress tests of CPU, GPU and memtestx86 over night) I tried to

- change power settings on Windows and my GPU to prefer performance, to not go into sleep modes and
- change power settings in BIOS e.g. disable global C-States or Power Loading (which adds a dummy load on my system during low load).
- uninstall monitoring software
- disable XMP
And a few more I don't remember. I also did not enable any overclocking or anything.

Note that testing even just one setting can take more than a full day due to randomness of crashes.
I have read up that some systems also have problems with RAM timing especially for the Ryzen processors but I couldn't find anyone with similar issues to mine related to RAM timings.

4. One thing that seems to help but I didn't test it long enough yet (and might be biased because of my assumptions and wishful thinking):
I found this odd "solution" one person had for a problem that sounded exactly like mine. He uses Prime95 to permanently stress-test one of his cores. This apparently sets his system in a stable condition and he didn't crash for 11 days.
I have crashed once since I tried that (2days now) but I think I have chosen a different stress test by mistake so it might work with the "correct one". However this is no real long-term solution I just want to test if this helps to at least minimize the occurences of my crashes. If it does do that to a noticable amount it might help push anyone with more knowledge than me into the right direction of a possible cause (and hopefully fix).

I know I wrote a lot and I tried to be thorough but I am not perfect at getting "directly to the point" so I apologize for the long read. I hope my info provided is sufficient and gives you an idea. Do not hesistate to ask me for more information. I'd be really happy if anyone was able to help!
 

Ralston18

Titan
Moderator
"Windows manager logs "

What did you use: E.g. Reliability History and/or Event Viewer? Or something else?

Reliability History has a timeline format that may help you establish when the problem started and also pick out the "12 hour" pattern.

Start using Resource Monitor and Task Manager to observe system performance - use both but only one at a time.

Look for system and component changes that lead to the system becoming unstable, crashing, etc..

Also:

Power down, unplug, open the case.

Clean out dust and debris.

Check by sight and feel that all connectors, cards, RAM, and jumpers are fully and firmly in place.

Could be something just a bit loose and getting looser over time as the computer's components expand and contract with temperature changes. Vibrations will do the same.
 
Dec 4, 2020
2
0
10
0
!!!UPDATE (solved) !!!

For me the issue got fixed when I enabled "Typical current idle" in my BIOS settings for the PSU. Apparently my PSU thinks my board doesn't need any power when it goes below a threshold in some sort of powersaving/efficiency mode.
 

ASK THE COMMUNITY