System instability/freezing under load

[dannyboy75]

Specs are as follows:

Asus TUF Gaming B550 Plus
Ryzen 5800X
nVidia RTX 3080 Founders' Edition
32GB Crucial Ballistix 3200MHz DDR4
Aerocool Mod XT 750W PSU
Arctic Freezer 34 cooler (single fan, not Duo or eSports versions)

I've had this system for about 18 months now. It was fine for the first year or so, but I've been noticing the instability/freezing issues occasionally for a few months now. In the last 2 weeks it's gradually got noticeably worse. Essentially the display will freeze or go black, mouse cursor is usually still responsive but Ctrl-Shift-Esc does nothing. I have to force power-off and boot system back up again to recover.

It happens most frequently in the two most intensive games I play - MS Flight Simulator and BeamNG. It's got to the point now where it happens within a few minutes of playing these. This morning MSFS even froze before the game had finished loading (as FS uses quite a bit of CPU/GPU even on the menu/loading screens).

Idle CPU temp is 34c, but gets pretty high under duress - it quickly hits 90c and throttles if running the Small FFT test in Prime95. However, it's always done this and from what I can gather is normal for the 5800X. In the past this never led to any issues so think it's likely a red herring. I have reapplied thermal paste a couple of times in the last few months, but this makes no difference to the CPU temp. My cooler isn't the best, but it's rated for up to 150W TDP which should be fine for the 5800X (105W according to specs).

MemTest shows no errors, so I don't think the issue is RAM. I also tried disabling XMP anyway, just to be sure, but again no difference.

Running Prime95 on its own doesn't replicate the issue, despite Small FFT test pushing CPU temp up to 90. The Large FFT test (which places more stress on RAM) sees CPU temp hover around 85-88C. Again no issues though.

Similarly, running FurMark on its own gives no issues. Also, FWIW I'm not seeing any artifacting at all.

Interestingly though, when I run both Prime95 and FurMark together (which more closely replicates the stress of gaming across the CPU, GPU and RAM), within a minute or two the system starts to freeze up. If I can quit out of P95/FurMark quickly enough, the system sometimes recovers but not always.

One thing I have just noticed is that, when both Prime95 and FurMark are running my +12V rail is only showing as 11.02 Volts, so am guessing this is the same when gaming. From what I've read, the ATX standard suggests a tolerance of +/- 5% on voltages, however dropping 1V out of 12V equates to 8.5%. Could my PSU be the issue?

Sorry for the long post but trying to give as much detail as possible. Appreciate any thoughts.

thanks
Danny

[Jonj1611]

I would suspect the PSU, its a fairly cheap model and in a system like that I would look to a higher quality PSU

[[GSV]Trig]

PSU would be my first shout, anything in event viewer when its rebooted after a lock up?

[dannyboy75]

Thanks guys, have ordered a Corsair 850x from Amazon. I checked when the Aerocool PSU was purchased as wondered if it might still be under warranty. I was shocked to see it's 5 years old now, so I guess that's not a bad innings for a cheap-ish brand.

Nothing of note in the event logs, other than the usual 'previous system shutdown was unexpected'.

At least I can return the Corsair if we are barking up the wrong tree, but I think that's pretty unlikely. That drop from +12V to 11 definitely doesn't seem right.

[Jonj1611]

I am sure a new PSU will fix the issues

[dannyboy75]

Well the Corsair PSU arrived a day late, but I got it installed this evening. Thankfully, as expected, it seems to have fixed the stability issues. Played FS for half an hour or so, and it was rock solid.

Thanks again for the advice chaps. I think I'd almost answered my own question by the end of my post, but it's always helpful to have a few second opinions

[Jonj1611]

I have done that myself. By the time I get to the end of my question I have pretty much worked through it and got my own answer

Glad its working for you

[MrJim]

I would suspect the PSU, its a fairly cheap model and in a system like that I would look to a higher quality PSU

The RTX 3080 & upwards GPUs are notorious for being very 'peaky', meaning that they tend to draw very high current for short periods. As I was reading the OP my first thought was the PSU, so I'm glad the new unit seems to have sorted out the problem.

[dannyboy75]

The RTX 3080 & upwards GPUs are notorious for being very 'peaky', meaning that they tend to draw very high current for short periods. As I was reading the OP my first thought was the PSU, so I'm glad the new unit seems to have sorted out the problem.

Interesting, I wasn't aware of that 'peakiness'. TBH, with hindsight, even when that old PSU was still healthy 750W was probably cutting it fine, especially once I installed the 3080FE last year. I've had the occasional game crash for quite a long time now, but had just put it down to the joys/foibles of PC ownership or buggy games.

That's a beast mode PC you have there, btw