System instability and random BSODs?
System instability and random BSODs?
Hello everyone.
I just got a PC from a friend and have been setting it up myself. All I did was connect my existing drives and reinstall Windows on the SSD already present.
My friend had been experiencing problems with the machine since it arrived—he paid someone to build it for him. It would often crash with bluescreens, and games would freeze or stop unexpectedly. I’m now trying to figure out what’s going on, as the issue hasn’t improved.
Here are the details:
- CPU: Ryzen 5 5600X (with an AIO)
- Motherboard: Gigabyte B550M Gaming (note: earlier thought it was MSI)
- RAM: 32GB DDR4 (Teamgroup T-Force, two sticks)
- Storage: 1TB NVMe SSD (Windows drive), 1TB SATA SSD, 2TB HDDs, 2TB HDDs
- GPU: GIGABYTE GeForce RTX 3060 Ti Gaming OC 8GB
- Power Supply: Gigabyte P750BS (previously P750GM)
No overclocking has been done.
Issues I’ve noticed:
- Random PC BSODs, sometimes during shutdown/restart or when idle (system idling or downloading a game).
- Crash dumps available here: [link]
- Data corruption appears—games like Cyberpunk 2077 get stuck on endless loading screens. Textures often fail mid-game. Sometimes multiple apps (Steam, Discord, Teams) would fail to launch and trigger errors.
My friend has dealt with these problems for years; it seems the machine would shut down completely at times.
What I’ve tried:
- RAM passed four memtest86 checks.
- CrystalDiskInfo reported 89% health on the SATA SSD (100 uncorrectable errors) and 97% on the NVMe drive. Moving a problematic game from SATA to NVMe didn’t fix the problem.
- BIOS reset didn’t help.
- The old PSU had a recall (serial number in range), which sometimes would explode. I replaced it yesterday, but bluescreens remain.
- Ran `sfc /scannow` and `DISM /Online /Cleanup-Image /RestoreHealth`. Results: corruption detected and repaired.
- Dumps: [link]
I also have a spec profile here if that might assist: [profile link]
I’m hoping someone can help diagnose this, as I’m running out of options without replacing parts randomly (I’m in a tough spot).
Welcome to the community, fresh face!
What I did was connect my previous drives into this system and reinstall Windows on the SSD that was already present.
Per your advice, it would have been better to format and reinstall the operating system while disconnecting all drives except the one you wanted to install. Talking about the OS, where did you obtain the installation file? Regarding the reinstallation, did you do it in offline mode?
Share a screenshot of what appears in Disk Management. With Samsung's Magician app, verify if the Samsung SSDs require a firmware update. You may also want to install the NVMe driver for your Samsung 970 Evo. Find the driver from here;
Tool & Software Download | Samsung Semiconductor Global
Explore various tools linked to your device, such as Samsung Magician, tailored for Samsung memory products, and gather related details.
semiconductor.samsung.com
Given you're using Windows 10, everything remains stable.
Are any overclocking settings active?
Is X.M.P. enabled in the BIOS?
CPU: Ryzen 5 5600X (with an AIO)
Could you tell me the model of the AIO cooler you mentioned? What temperatures are displayed in BIOS? Also, what is the make and model of your case?
Mobo: MSI B550M Gaming
Your system info shows a Gigabyte motherboard and BIOS version F11k. Is this the one you're using?
https://www.gigabyte.com/Motherboard/B55...ING-rev-1x
? If so;
https://www.gigabyte.com/Motherboard/B55...pport-Bios
The Gigabyte support page doesn't list BIOS version F11k. It seems it was removed for a newer version or considered unstable/security-related.
If I were you, I'd ensure the platform ran with the latest chipset drivers from AMD's site, install them in elevated mode (Right click installer > Run as Administrator), and then flash the BIOS using the Q-Flash Plus button on the motherboard back panel, following this guide;
View: https://www.youtube.com/watch?v=ABNgLr7YhTc
Even with a different board, the approach stays similar. After confirming the BIOS update to the latest (currently F19), reset the CMOS and test again.
In my opinion, tackling this step by step would be more effective to achieve optimal results. Additionally, consider replacing your PSU—opt for a solid 750W unit, particularly since you're running an RTX 3060Ti and currently have a Tier-C power supply.
I second the comments @Lutfij made about changing PSU. I've just swapped out an old (2021 vintage) Corsair TX850M for a new MSI A1000G to fix a problem on my 7950X rig with an RTX 3060 and then an RTX 4070. With either GPU installed and powered by the TX850M, the system occasionally crashed unceremoniously, then restarted. Now that I've installed a modern ATX 3.x PSU (MSI A1000G) in place of the old ATX 2.x PSU (TX850M) the system has not crashed. I opted for a really good amount of headroom (1000W) despite the PC only drawing 430W max continuous power from the mains. 1000W is definite overkill, but it works. The NVidia 3000 series GPUs are famous for creating short-duration high-current spikes of up to 200% from the +12V rail. These spikes only last a few milliseconds but can upset inadequate PSUs. ATX 2.x PSUs with a 120/130% OCP trip point cannot always cope with high transients. ATX 3.x increases the short term OCP on the +12V rail to 200% to work with NVidia GPUs. I strongly recommend buying a new Tier-A PSU that's ATX 3.x, PCIe 5.x compliant. This won't get rid of other problems, but it should make the RTX 3060Ti more stable. When a computer Blue Screens or crashes, it sometimes corrupts the Windows Registry. If this gets really serious, an OS resinstall might be required. The review for your old P750GM on Tom's states: Disappointing transient response at +12 V and 3.3 V Not a good PSU to pair with an RTX 3060Ti. I haven't found a review for your new P750BS but I did discover this comment which worried me: While Gigabyte's " P xxx B " line is structurally different than their infamous exploding " P xxx GM " line, it's still made by the same obscure OEM (i.e. MEIC) and is pretty mediocre overall. https://www.reddit.com/r/buildapc/commen..._my_build/ For peace of mind, I urge you to return the P750BS and fork out some more cash for a much better ATX 3.x compliant Tier-A PSU. People on this forum recommend "Tier-A" PSUs for important builds, which unfortunately can be a lot more expensive. As you work your way down the list you get to Tier-B which are OK for lesser builds. Then there's Tier-C which I'd only install in a simple office PC with no GPU card. Below that you end up with Tier-F classified as REPLACE IMMEDIATELY. There are loads of PSU Tier Lists; https://cultists.network/140/psu-tier-list/ https://pc-builds.com/psu-tier-list/ https://axeetech.com/psu-tier-list/ Before buying a PSU, I try to find a really thorough review on Tom's web site or Hardware Busters. A PSU is the most important part of a system. Don't cheap out and buy junk, I get this corruption message quite often when checking "perfectly good" systems and I'm no longer worried about it. Perhaps I should investigate further. When I start seeing uncorrectable errors on a spinning hard disk or SSD, i make plans to replace it. Data integrity is more important to me than continuing to trust a "failing" drive. Even if it's not the Windows boot drive, it could still cause the system to crash. With or without XMP enabled in the BIOS? If the RAM was tested at JEDEC default speed and you then overclocked the RAM with XMP, you should run MemTest86 again at the higher memory frequency.