Windows 11 Data Loss, Random File Corruption

I have an AMD setup that had been fine for the last 2 years or so but a few months back I upgraded to Win11 on a work boot drive and a boot drive for gaming. Both are NVMe SSDs.

About a month ago I had a few large CD ISO files go bad and had to download them again. Then a few weeks ago a number of VM files went missing and others were corrupted or 0 bytes. Out of 6 VMs only one showed up in VMware Workstation and had all the files but still would not boot. Luckily I'd just backed them up to the NAS.

I don't know if the RAM or CPU are having issues. I ran Prime95 for a few days with no errors. I ran the Windows Memory Diagnostic tool but not the extended test yet. It looks like I'd have to run it for a few days?

I'd not looked at motherboard firmware for some time and sure enough there was one for the Windows 11 and the AMD fTPM issue. After installing it the system seems a little more responsive. Maybe?

The second data loss and corruption was with the VMs files. On a Friday I backed them all up and then when I went to use it again on Monday files where missing and others corrupt.

About a month ago I upgraded from a 1.8TB SN850 to a 3.7TB SN850X. I've been using that Game boot drive for extra VM drive space with the 1.8TB and now the 3.7TB version.
It seems fine but the system seemed fine for the last 3 yeas.

On the one hand I feel like it could be the drives. There are a few folks that have had bad WD SN850 and SN850X drives. The SMART stats all look good but the older 2TB SN850 can no longer run the SMART test without failing but I'm no longer using it. The new 4TB SN850X can run both the long and short tests but it's the one with the file issues. Also, that same Monday the older Samsung work-only SSD that I had also upgraded to Win11 stopped booting into Windows. I can't repair it and it says there are no restore points. The Samsung software says there are no errors but they don't let you run a SMART test. I can mount the drive in another OS and read the files.

Random issues stink.

Do you think it could be temperature related? I'd not really thought about it as all the m.2 SSDs have a heatsink that came with the motherboard installed but looking at the SN850 and SN850X they idle at 40-50 °C and when coping lots of files they get up to 70-75 °C.

On this Saturday I got two new Samsung 990 Pro 1.8TB SSDs and installed Win11 from the MS created USB and all seemed ok until last night. I tried making an image to the NAS and the new Gaming-only SSD backed up fine but I'd only used it one day. The Work-only SSD showed an error and the Windows system image tool warns that you need to repair the drive before making the image.

Thinking of just buying new RAM in the hopes it is the issue. The last time I had a RAM issue it only showed up when doing large 60GB+ ISO creation of movies to the NAS. When I'd go to watch one there would be corruption randomly in the and it would fail a diff check when copied. All other RAM tests showed no issue but when I put in a spare single 8GB stick all was well.

If you have prime95, run it with large FFT and all available memory that you can. It sounds like it must be a memory issue.

Any related events in event viewer -> system?

Download and run MemTest86 (free). It will go until it makes 4 passes.

MemTest86

On second thought, if it was memory and corrupting your disk that often, I would think your computer would be crashing quite a bit too. Has it?

It is very unlikely that is a temperature issue as the nvme ssd will throttle performance to lower the temp if it gets too hot.

That and some other things bug me too.

No crashing. There are a few things like some games go to a black screen for second and the back but that only started after getting the RTX4090. But I wonder if whatever hardware issue I may be having is because of the card somehow or is the same issue that is causing random data loss.

Prime95 did not show any errors. I've only run it for about 24hours though. It's hard because I need the PC for work.

GPU stress tests like FurMark are fine.

Before reinstalling, the worst case was when all the VMs were fine and backed up on Friday were then all messed up on Monday morning. I had to restore them from the NAS and they were fine so what would mess them all up after the fact? I may have run a check disk but again if they were good when backed up why did they change? It seems like a disk check (I Fixed It!) kind of thing where some files were gone and others were 0 bytes.

???

There is a lot of drive and RAM activity when running the VMs. The current project has me running like 6 at the same time and I'm using most all of the 64GB of RAM. So that seems like a good time for the RAM or CPU to mess with the files but again, they were good when backup up a few days before.

Starting to think I have a virous but I've never heard of one like this. I installed Win11 from a newly created USB, onto two new SSDs and a few days latter I get the message that the drive has errors when trying to run a Windows backup image. It said it fixed the errors and I've been checking a few times a day with no more errors found.

I've ordered a different RAM kit but it's a hard thing to test for. Can I run some kind of CRC check on the whole drive each night? Is Windows Disk Check doing basicity that. Seems too fast for that.

I also bumped up the RAM voltage a little. Temped to start building a new PC at this point.

geneo said:

On second thought, if it was memory and corrupting your disk that often, I would think your computer would be crashing quite a bit too. Has it?

It is very unlikely that is a temperature issue as the nvme ssd will throttle performance to lower the temp if it gets too hot.

Click to expand...

I did get a different cooler and it dropped the SSDs temps by 10°C. They were like 49°C at idle and had thought maybe that was not good for data integrity but the two new ones seem to have had the same issue so it must be something else.

I have 3 computers with M.2 SSDs. None of them have any cooling. My new laptop has two Samsung 980 NVME M.2 SSDs. They idles in the 50's and run much hotter under load.

Is it possible your RTX4090 is causing stability problems with your power supply?

Yeah, I think the SSD temps is a long shot.

I had a RTX3080 in the same system for a few years and never saw these issues until a few months ago.

Maybe the Corsair AX1500i Digital ATX is dying? ... or the CPU ... or something on the MB ... ???

I had a Corsair RM750i P/S. I had a Nvidia GTX1060 6GB GPU. I ordered a Nvidia RTX3060Ti 8GB GPU. Before I swapped GPUs I played a few games and ran the Heaven benchmark. I started getting BSODs. The same thing happened when I installed the new GPU. The problem went away when I subsituted a 10 year old Antec 650watt P/S. I replaced the original P/S under warranty with a Corsair RM750x one. I suspect under load the original P/S was becoming unstable causing the BSODs.

Even though your computer has different symptoms than mine you can't rule out your power supply. The problem is right now you don't know what is causing the data corruption. The only sure way to find out is to swap out components until you find the one that is failing. The problem is that most people don't have spares making that difficult.

BTW, I had a computer fail 12 years ago. I didn't have spare parts so I couldn't figure out what was wrong with it. I ended up building a new one. A few years later with enough spares I was able to figure out what was wrong with the original computer and fixed it. Today both computers are still working.

It is lame. I like it when support is like: Hey, do have an extra CPU to test with? Sure, the one that is $750 and out of stock everywhere?

Been doing this for 30+ years. When are they going to have an app that can find these random issues. :) Something like SMART for drives but works and is for the whole PC.

If I do start building a new PC I could try a new power supply but I'd be getting DDR5 RAM and a MB with a new CPU socket and so on.

Hopefully the new RAM kit turns the tide. I'm not going to be buying the same gen CPU or MB just to see if it helps.

On top of all that it's a lame random issue. I'd kill for a random BSOD with in 5 minutes or less at this point.

I don't know of a good way to test and know if the issues is still there. I just have to wait and see if files I'm actively using go bad.

I have 5 desktop computers but only the newest has an AMD AM4 CPU and DDR4 RAM. I have a spare CPU, RAM, and GPU only because I upgraded them a year ago. If I hadn't done that I wouldn't have any spares. However, some of my older computers have P/Ss and GPUs that could be be pulled for testing if required.

I guess it is a luxury to have more than computer. I remember many years ago when computers were relatively expensive so spare anything was not an option. If you only have one of everything and you have a problem that is hard to pin down then you have two choices:

• Buy spares parts for your own testing
• Take your computer to a computer repair shop

I'm going to run without the RAM XMP and setting the PCIe to 3.0.

Maybe the AMD Chip Set drivers install is not a good idea? I'm guess most just use whatever Windows installs.

My biggest issue now is how to test for random bad files.

I have always installed chipset drivers and their updates from the motherboard manufacturers (i.e. ASUS, Gigabyte, etc.) websites. I have never had a problem when doing this. Even when drivers came from a Windows update they were usually much older versions. That means if you want the latest drivers you have to get them from the manufacturer.

You should be able to run RAM at XMP. If you can't then there is something wrong. It could be incompatible RAM, defective RAM, or several other causes. You have to troubeshoot the problem to isolate the cause.

I have several desktop computers with the oldest 20 years. All the RAM in them runs at their rated speeds. Rated speeds are overclocked speeds which is greater than the default speeds. If you don't set the speeds to Auto, XMP, or Manual the RAM runs at their default speeds which is somewhat less than their rated speeds. For example, DDR4-3200 RAM is rated for 3200MHz but runs at 2133MHz if not set to XMP.

Here are a some screenshots of two of my computer BIOSs and the RAM used. The 1st uses DDR3 while the 2nd uses DDR4.

Note about XMP alternatives:

DOCP is short for Direct Overclock Profile, whereas EOCP is short for Extended Overclock Profiles. Both are XMP alternatives for AMD motherboards, with DOCP coming from ASUS and EOCP coming from Gigabyte, respectively.

Click to expand...

MisterEd said:

I had a Corsair RM750i P/S. I had a Nvidia GTX1060 6GB GPU. I ordered a Nvidia RTX3060Ti 8GB GPU. Before I swapped GPUs I played a few games and ran the Heaven benchmark. I started getting BSODs. The same thing happened when I installed the new GPU. The problem went away when I subsituted a 10 year old Antec 650watt P/S. I replaced the original P/S under warranty with a Corsair RM750x one.

Click to expand...

That PSU very likely had bad caps.

RJARRRPCGP said:

That PSU very likely had bad caps.

Click to expand...

Not good since it was less than 5 years old. At least it had a 10 year warranty. They didn't have a replacement RM750i so they offered a RM750x instead. I accepted that. I knew that RM750i was no longer made so I would be getting a refurbished one. Note the RM750x was a 2021 model so was probably new.