Device stopped unexpectedly without any notice, no visual alerts, no shutdown screen, nothing apparent.
Device stopped unexpectedly without any notice, no visual alerts, no shutdown screen, nothing apparent.
I own this machine for several years (about four), with many components still covered under warranty—like the power supply has a ten-year guarantee. So far everything has worked perfectly. However, over the past month or so it’s started behaving oddly. It occasionally just restarts, the screen goes black instantly, and then it boots again from the start. There’s no BSOD or any error messages, nothing out of the ordinary. This has occurred roughly four to five times in the last month. The computer is usually always on, or stays on for 4 to 10 days before it powers off or restarts for a couple of hours before coming back. These random reboots happen without any clear cause. It happened during gaming, once while using Firefox, and also when I was lying in bed with nothing on the PC—just idle, and the monitors went dark before it restarted. Temperatures seem normal when I’m playing. Drivers are up to date, and updates are applied automatically. I also have an UPS, so power issues shouldn’t be the problem (it didn’t fail with reboots even before adding the UPS).
My system runs Windows 10 22H2, CPU Ryzen 9 3900x, 32GB RAM Corsair Vengeance RGB Pro 3200, 2x16GB, GPU AMD RX 6700 XT, PSU Corsair RM750x, plus an SSD with M.2 NVMe Adata XPG SX8200 Pro 512GB and two SATA drives for data storage.
My question: Are there any logs I can check for this crash in Windows? Since it restarts instantly, I’m unsure if the OS saves any information about the incident. Where should I look? I don’t have spare parts to replace or test, and I’m not sure if a fix will work right away—after three days of intermittent crashes, it’s hard to tell if something is fixed.
My approach would be to stress-test the CPU (Cinebench, Prime95), run a RAM test with Memtest64, and see if NVME works without data loss. I’d avoid reinstalling Windows at all cost, as I’ve spent a lot of time rebuilding settings and software after each reset. The PSU appears to be good quality, and I have no reason to doubt it.
Thanks for the info. Let me know if you need anything else.
Well, Prime95 ran for 24 hours without any issues. Memtest64 completed successfully with no errors detected. Hard Disk Sentinel reported the boot device as NVME at 83% health, indicating a perfect status. No weak or problematic sectors were identified. The SSD’s TRIM functionality is active and enabled for best performance. Health metrics come from the SSD’s specific S.M.A.R.T. attributes: #169 Remain Life Percentage. It’s advisable to keep monitoring the hard drive condition regularly. Power-on time was 715 days, with an estimated lifespan of 764 days and a write capacity of 16TB. This suggests the problem likely lies with the NVME device itself. No GPU testing was necessary since the crash occurred in both gaming and idle mode without GPU load. I searched for Windows logs but couldn’t locate any relevant entries. It seems a software-related issue may be present, possibly tied to Windows, though I’m unsure of the exact cause. The screen displayed no errors—no freezing, fans didn’t blink, and there were no unusual sounds or visuals during the incident.
Have you got an extra GPU? Consider using it instead. You might find a very inexpensive one for around $5 on eBay, probably a motherboard or power supply unit. Think about swapping your PSU from the warranty. I don’t have any other ideas. Perform a GPU stress test.
I have an extra GPU, but it's quite weak. I'm planning to run tests for a week since the weak one limits my work. I also suspect issues with the motherboard, power supply, or Windows settings. It might be safe to replace the PSU after six years without confirming it's broken, just as a guess. I might call to ask before doing it. I dislike replacing components like GPUs or hard drives because it means rebuilding the whole system. I'm still curious if the problem could stem from software or Windows logs, especially since the crash doesn't occur immediately after turning on the PC but rather after a week or 15-30 days of use without restarting or shutting it down.
I faced the same issue with my PC, which was due to a PSU failure. After replacing it, the problem persisted, so I changed several components and still encountered the same situation. Finally, installing a new PSU resolved everything. It seems the original PSU is likely the cause.
I think I need to return and swap it out for a new one. It's covered under warranty (7 out of 10 years gone). But I’m not sure they’ll just fix it, they might try to repair it within the warranty period if they find any problem... I hope they can locate the issue and either fix it or replace it with a new unit if the same model is still available.
You can try various tests to check the stress on your PSU. Besides Prime95 for 24 hours, you can also measure GPU load and combine it with Prime95. These methods are reliable. I haven’t used 3DMark yet, but I’m considering OCCT as well. Using Heaven or Cinebench is another option for stressing CPU/GPU. You mentioned not wanting to risk a crash during idle times, so running them under heavy load might be safer. The recent crashes seem linked to prolonged usage without shutdowns—usually after several days. If you keep the PC running continuously, it’s likely to fail eventually, especially if left unattended for months. I’ve noticed it only fails after a while without intervention. I’m planning to run OCCT’s stability test soon and will share results.
00:00:00 - Info - Test schedule began on 2024-04-14 at 13:28:54 00:00:00 - Info - Power - Initiated (Duration: 1 hour) 01:00:01 - Info - Power - Test concluded 01:00:01 - Info - Schedule Finished Run OCCT power stability test for one hour, everything normal, no crashes as anticipated. Temperatures hovered around CPU 72°C and GPU 80°C (custom fan curve keeps fans at 500-700 RPM max except GPU and cooling pump at full speed). Power draw from wall PSU to UPS was approximately 400W (395-415W), which is ideal for a 750W PSU, representing about 50% usage. I don’t overclock the PSU when running the PC fully loaded; I seldom run CPU+GPU at peak, usually just GPU during rare gaming sessions. About 99% of the time the system operates in idle mode (Word, Chrome, VS Code, etc., minimal CPU and no GPU strain).