Solved! Need help troubleshooting a new PC that's crashing

[In2Photos]

Apologies as this might get long winded, but my son built a new PC last month and we've had some issues with it crashing while gaming. Here are the specs:

Ryzen 5600x
MSI B550 Tomahawk
32 GB Corsair 3600 C18 (2x16GB)
WD SN750SE 500GB M.2 (OS drive)
Crucial MX500 500GB SATA SSD (Games Drive)
MSI GTX1660
Corsair RM650x
Enermax 240mm AIO

We built the PC with a temporary PSU as the Corsair was late in arriving and everything was working fine. I stress tested everything using Cinebench and Heaven and it all looked great. I enabled XMP and PBO and set a -25mV undervolt on the CPU through BIOS and re-tested, again everything was great. The Corsair PSU arrived and we installed it. A few days later the PSU died. You could hear an audible click when you pressed the power button, but it wouldn't even attempt to turn on the system. So we put the temporary PSU back in until a new Corsair arrived. During this time my son experienced a couple of crashes while playing Minecraft and Fortnite. We dialed back the undervolt to -20 and everything was again working fine. The replacement Corsair arrived and we installed it. Everything again was working fine. Then about a week ago he tried to play a new game, the Division 2. It will not run. He has only been able to play the game for maybe 10 minutes once. All other times it crashes shortly after the Easy Anti-Cheat window appears. I removed the undervolt and disabled PBO, it still crashed. I went on the Ubisoft forums and there appears to be a lot of complaints about crashing so I chalked it up to that. Well last night he bought Rust. It also crashes. He also told me that Modern Warfare had also crashed.

In Division 2 I noticed that as the Anti Cheat window popped up the RGB on my son's RAM would change from a static blue color to a moving white and orange effect. After about 5 seconds the screens go completely black, the RGB on everything turns off and the system reboots. In Rust, he is able to play the game for a few minutes before the screens go completely black, the RGB turns off and the system reboots. Event Viewer shows a hardware error of some sort but it didn't point me in any direction. Both of these games appear to have some ability to control RGB lighting. Not sure if Modern Warfare has that as well. I don't believe Minecraft and Fortnite do that and they aren't crashing. Is there something going on here perhaps? I would think a software issue wouldn't shut the system down like this though.

Here is what I have tried so far:

Removed the undervolt and turned off PBO. Division 2 and Rust still crash. He uninstalled Modern Warfare due to its size (200+GB!!!). Minecraft, Fortnite, and Subnautica run fine. He also runs Discord and Youtube at the same time with no issue.
Run additional stress tests. Cinebench and Prime 95 on the CPU with and without the undervolt and PBO. It runs fine, temps are mid 70s after 10-15 minutes at 4.6GHz all core. I also ran Heaven while Prime 95 was running. The system ran for 20 minutes with no issues whatsoever. Memtest is still running but after 2 passes it has found no errors.
Ran chkdsk on the SATA SSD since that is where all the games are. It found no errors. SMART reports that the drive is fine.
Tried a single stick of RAM. System crashed both times in Rust.
Windows is up to date, drivers are up to date. We did a "clean installation" on video drivers (I have not used DDU yet).

I'm not sure what to do next. I thought about putting the temporary PSU back in, but if the system doesn't crash while running Prime 95 and Heaven at the same time I don't think it is a power delivery issue as that is more stressful than games. Clear CMOS maybe?

I will also mention that this was not a new OS install. I used Macrium to clone the OS from the Crucial drive that was installed in his previous system to the new M.2. This OS install was only a few months old. It wasn't as smooth to transition it over the the new system as I'd hoped but it seemed to work fine. I would think if this was an OS problem he would be getting BSODs or freezing and not just a complete shut down.

Again, sorry for the long read, but I figured having as much info as possible could lead to the culprit.

 

[In2Photos]

So did you run 20 cycles of TM5? Your previous tests only showed 3 cycles for 9 some minutes

View attachment 57336


Here's where you edit the file.

View attachment 57337

But if you think 3 cycles and 9 min is enough, that's fine...

EDIT: btw, if we are using the same cfg file, 32GB with TM5, 1 cycle should take around 5-6m. If you've done 3 in 9 that looks wrong.

Here's a cheat sheet for errors that show up:

View attachment 57341

OK, I started TM5 again. It took about 8 minutes per cycle. I let it run for over an hour (almost 8 cycles) and there were no errors. We were also successful in playing the Division 2 for 2 hours tonight with no crashes/rebooting. So I'd say the BIOS update fixed whatever was happening. Thanks for all your help!

 

[BTRY B 529th FA BN]

I'd say do a fresh install.

You can also run in elevated cmd prompt or administrative level power shell: sfc /scannow

Also run TM5 to test the RAM

Run Y cruncher also: 1-7-0, 4 iterations

 

[In2Photos]

I'd say do a fresh install.

You can also run in elevated cmd prompt or administrative level power shell: sfc /scannow

Also run TM5 to test the RAM

Run Y cruncher also: 1-7-0, 4 iterations

All tests passed.

 

[BTRY B 529th FA BN]

The TM5 test should run 16 tests. Run as Administrator:



Be aware there's different TM5 config files that vary the testing process. Some are more extreme than others. I use the default one by 1usmus_v3. I have mine set to run 20 cycles. You can access it in the folder and it's a .cfg file.

 

[In2Photos]

OK, I ran it again using the 1usmus_v3 config file. No errors.

 

[In2Photos]

OK, I ran it again using the 1usmus_v3 config file. No errors.

View attachment 57296

 

[BTRY B 529th FA BN]

Change the config file to at-least 20 cycles. I've had an error on the 24th cycle but that's up to you if you want to run 25 or more cycles. That should take maybe 2hrs or more.

 

[Tech Junky]

Everything to me points to the PSU or MOBO

The fact that you already swapped the PSU 2-3 times after needing to RMA the original. non-specific HW reboots.... I had what I thought was a bad PSU for my laptop that I thought was spitting kernel-power errors but, it turned out to be some shotty Intel drivers that needed to rollback.

A few years ago when I was building another system I went with a Corsair based on reviews and it arrived DOA and I swapped it with a temp Thermaltake and the final was an EVGA that's still running 4 years later. I put another EVGA into my new setup and it's been rock solid since Nov. PSU's can cause odd symptoms to appear as other issues though.

 

[In2Photos]

It's not the PSU. We swapped it with the temporary PSU, which is an EVGA 600W unit and it crashed in Rust after about 10 minutes. We also tried going directly to the receptacle rather than through the UPS to rule that out as well. Still crashed. In regards to PSU manufacturers I have 3 Corsair PSU in systems right now, 2 of which run 24/7. One of them is 13 years old and the other is 7 or 8 I think. The unit that died after a few days was the first PSU I have had fail personally. I've had a PSU die on a work PC. But I know EVGA makes good ones as well which is why I picked up a unit to have around for troubleshooting and/or as a spare.

While reading some other posts from people that have had similar experiences I decided to check the BIOS version to see if it was the latest. It was not so I updated the BIOS and he is playing right now to see that helped. I'll report back with our findings.

 

[OlyAR15]

You really should do a clean install of the OS, then install just the basic drivers to get the machine functional, then the games. Forget about the RGB crap for now, and any other drivers for things like the mouse and keyboard. If all that works, then start installing those programs one by one and see if any of them create instability.

 

[BTRY B 529th FA BN]

Any luck?

 

[In2Photos]

My son had baseball tryouts tonight so he only got to play for 30 minutes but it didn't crash during that time so fingers crossed! He should get some time to play tomorrow and maybe try more than 1 game.

 

[In2Photos]

After the BIOS update yesterday he was able to play Rust for 30 minutes with no crashing/rebooting. This morning he played for an hour and a half with no rebooting. The game crashed once, but the PC didn't reboot and all other apps stayed open. The person he plays with said that is common for this game. So perhaps the BIOS update was the fix.

You really should do a clean install of the OS, then install just the basic drivers to get the machine functional, then the games. Forget about the RGB crap for now, and any other drivers for things like the mouse and keyboard. If all that works, then start installing those programs one by one and see if any of them create instability.

So I've thought about this post since I read it last night and while we probably should have done this from the start I don't think this would have helped in our situation. I'm not 100% opposed to doing the fresh install. Most games and software are easy to reinstall. But Minecraft SUCKS in this regard due to all the mod packs and stuff. As I said in my initial post the PC operated perfectly fine under stress testing after it was built. No iCUE software, no MSI Mystic Light (the RGB crap as you referred to it). It continued to operate perfectly fine after those were installed as well, UNTIL the new games were installed. I would have done the exact same thing with a new OS.

A typical build might look like this
Build PC
Install Windows
Install motherboard drivers, graphics drivers
Stress test the system
Set the XMP profile and stress test some more
Set up any OC/voltage settings and stress test some more
Install software, games, etc.
Troubleshoot any issues
Enjoy the system
Add new software, games, etc.
Troubleshoot any issues
Enjoy the system
Repeat as necessary

On this build it looked like this
Build PC
Install Cloned Windows OS
Install motherboard drivers, graphics drivers
Stress test the system
Set the XMP profile and stress test some more
Set up any OC/voltage settings and stress test some more
Install software, games, etc.
Troubleshoot any issues
Enjoy the system
Replace faulty PSU
Add new software, games, etc.
Troubleshoot any issues
Enjoy the system
Repeat as necessary

So the only difference was a previous OS and a faulty PSU. If he was experiencing crashes without the rebooting I certainly could see how the cloned OS could be problematic. But this situation really seemed to point to a firmware/hardware issue.

 

[BTRY B 529th FA BN]

So did you run 20 cycles of TM5? Your previous tests only showed 3 cycles for 9 some minutes




Here's where you edit the file.



But if you think 3 cycles and 9 min is enough, that's fine...

EDIT: btw, if we are using the same cfg file, 32GB with TM5, 1 cycle should take around 5-6m. If you've done 3 in 9 that looks wrong.

Here's a cheat sheet for errors that show up: