Question New Build Intermittent Crashing

[wombat84]

I put together a new PC yesterday, and at first everything seemed fine. The BIOS loaded and the windows instillation went smoothly. So this morning I connected it to the internet and started the update process. Everything went fine for probably 90 minutes, as windows did its updates, and I updated the motherboard bios and installed the drivers for it and the GPU without any issues. The CPU fan seemed to be cycling up and down a bit, but nothing to crazy, and CPU temperatures were in the 40-50 degrees celsius range.

After I installed the recommended sound divers, but before rebooting for them, the screen went blank and the system crashed. The peripherals still had power, the CPU and case fans were running at a flat rate, and the motherboard has an error light on for the GPU. After turning off the system and booting it up again, it seemed to cycle through the boot process once or twice before going back to the login screen, but then crashed again.

Now when I try to turn it on now, sometimes it fails in the boot up, sometimes it makes it to the windows login and freezes there, and sometimes it will work for 5-10 minutes before freezing, seemingly related to how long since I last turned it on. It has also crashed in the bootup troubleshooting screen, but seems to work fine in the bios. And it always fails in the same manner.

I first checked all of the connections, and have taken them all out and reinserted them. The only possible issue there was that the reset switch may have been loose on the pins, but fixing that has not improved anything. I have also reset Windows. Nothing crashed when the reset ran, and windows updates ran fine, but the same crash happened after about 20 minutes, before I could even try any driver updates.

Any ideas you have would be greatly appreciated!

System specs: MSI B450m Gaming Plus motherboard; AMD 3600X; RX 5600 XT graphics card; 2x8 gb Corsair Vengeance LPX 3200 ram, 500gb WD Blue SN550 NVE SSD, EVGA BQ 600w power supply.

 

[Torn Mind]

Have you tried a complete reinstall of Windows, not just a reset?

To me, what seems most probable are bad drivers or bad drive, which corrupted the install Have you uninstalled all drivers? AMD drivers for GPU can be flaky.
What would be best would be to have a known good spare drive to install Windows on to see if the issue replicates itself.

 

[wombat84]

Have you tried a complete reinstall of Windows, not just a reset?

To me, what seems most probable are bad drivers or bad drive, which corrupted the install Have you uninstalled all drivers? AMD drivers for GPU can be flaky.
What would be best would be to have a known good spare drive to install Windows on to see if the issue replicates itself.

It was just a reset, but with the complete wipe option. I have done a clean install, and only downloaded the motherboard chipset and GPU drivers from AMD's website, and nothing from MSI or Gigabyte. Fingers crossed - so far it has been stable. I think it has lasted longer than the first install.

As far as a spare drive to test if it crashes again, I only have an ancient (maybe 12 year old?) HDD kicking around.

Thanks!

 

[Shmee]

Are you running that memory at XMP settings? Either way, if it freezes again, I would do a memtest.

 

[wombat84]

Are you running that memory at XMP settings? Either way, if it freezes again, I would do a memtest.

Nope, I haven't turned that on yet.

It is still intermittently crashing, and I cant see a pattern. Only one has happened under load, while yesterday it ran for 9 hours straight without an issue, only to crash while I went away to make dinner.

I have run the windows memtest, memtest86 and hci memtest. I have also run the seatool disc check and a number of cpu and gpu benchmarks and stress tests. They all came back normal, with the exception of a crash the first time I ran prime95.

To reboot it after a crash I need to turn off the power supply and hold the power button, otherwise it will fail during boot up, if that helps narrow it down.

 

[Shmee]

Sounds like it might be GPU or PSU related. I would try a graphics stress test, and maybe a PSU test.

 

[wombat84]

Sounds like it might be GPU or PSU related. I would try a graphics stress test, and maybe a PSU test.

I ran furmark for about 15 minutes, and the temps had stabilized at about 79 degrees. Is there a different test I should run, or should I let it go for longer? I also ran the passmark 2D and 3D gpu benchmarks.

For a PSU test, is there a software test I can run, or is this just testing the electronics?

 

[Steltek]

If you have to swap anything, I'd try the PSU first.

Did you by chance try resetting the CMOS memory? It also might be worth trying to roll the BIOS back to a prior version.

 

[wombat84]

If you have to swap anything, I'd try the PSU first.

Did you by chance try resetting the CMOS memory? It also might be worth trying to roll the BIOS back to a prior version.

I have reset the CMOS by selecting restore factory defaults in the BIOS, but I haven't tried to do it by shorting JBAT1 on the board. Does that make a difference? I rolled the BIOS back yesterday morning. It lasted for about 8 hours of light use, but has crashed on me once.

And nope, unfortunately no backup power supply.

 

[Shmee]

Well, if the GPU works fine in furmark, (I would try benchmarks and games as well) that pretty much leaves motherboard or PSU. I would also look to see if there is anything in the event viewer, or BSOD logs.

 

[wombat84]

I reset the CMOS on the motherboard, and it recognized it, but then crashed in the handover to windows, which happens often after it crashes.

I also checked the event logs, and the only error log or critical logs that happen around the times of the crash are a critical and error notification of incorrect shutdown when I reboot the system, plus a "Cloud files diagnostic event listener" failed to start error.

I tried monitoring the 12+ and 5+ voItages in HWmonitor, but it doesn't show any fluctuation, so I am assuming it doesn't read them correctly. I can borrow a PSU that has the same wattage from my brother, so I am going to try that and see if that solves things.

Thanks!

 

[Steltek]

You might also check the Windows reliability history to see if there are any specific repeating issues at the time of the crashes.

I do agree with replacing the PSU. You don't even have to remove the old one, just set the new one up outside the case and run the cables in to plug up (if they are long enough) as a time saver just for testing.

 

[Flayed]

I would check to see if the system crashes in safe mode