Weird issue with nVme ssd due to GPU

[amit_talkin]

This is something I have been struggling with for like an year. I kept checking for solution time to time but I couldnt figure out what exactly was the issue at first. Took me an year to finally find the culprit. well, still havent found the solution though, as I don't understand why such thing is happening.

Now to the point. I have 2x Samsung 970 Evo Plus 1TB SSDs which I am using on my Gigabyte Aorus Pro B550 motherboard ( Fully system spec in my profile ). It was working fine before an year ago, on same system.
Suddenly , since last year I was getting random errors in even viewer. Like "An error was detected on device \Device\Harddisk4\DR4 during a paging operation." and "IO operation at logical block address for disk was retried".
And it lead to system to be unresponsive and eventually hanged. I tried varius fixes shown on internet, nothing worked. Then I re-formatted system like many times. but nothing worked.

At last, I sent SSDs for RMA, they sent it back to me because they didn't find any problem with both of the SSDs. So I sent the motherboard for RMA, same response from Gigabyte noting that nothing is wrong with motherboard.
Now I have secondary system with Ryzen 3400 + A320M motherboard. I plugged my SSD and CPU into it, and it worked flawlessly. So it was now time to pin down the exact hardware which is causing the issue.

After wasting so much time in last few months, I found out that its my GPU which is causing my nVme SSDs to be disappeared. I have 1080Ti. How I found it? well i plugged my secondary 1080 GTX GPU, used it for a while, not a single issue. Both SSDs worked. everything good. Then once I plug my 1080Ti, play some games from SSDs, it hangs in like 15 min. Same errors and problems as above.(
Now, I ran multiple GPU tests on afterburner,no error found. And everything works fine as well if I just remove/disable both nvme SSDs from system. Nothing hangs, nothing crashes. All games work fine on 1080Ti.
Its just my GPU dont like my SSDs..lol.

For last 10 months I am using my system without those SSDs, everything works well. Then if I plug my SSDs, system hangs. But then if I switch to 1080GTX, all is working fine.

Now I am confused as hell, why would my GPU causing such issues to nvme SSDs? Point to be noted that 1080Ti works fine with those SSDs on A320M motherboard. And SSDs also work fine with 1080GTX GPU on my B550 motherboard.

 

[DeathtoGnomes]

Did you do a clean driver install each time you changed gpus?

 

[P4-630]

NVMe are PCIe SSD devices, and they use PCIe lanes, you said you had 2 GPU's running, probably not enough lanes to run them all.

 

[amit_talkin]

Did you do a clean driver install each time you changed gpus?

Yes, as i said, I even reformatted drive many times and reinstalled Windows 10/11 even tried Win7 on it.

NVMe are PCIe SSD devices, and they use PCIe lanes, you said you had 2 GPU's running, probably not enough lanes to run them all.

No, I have only 1 GPU running. 1080GTX is on spare system which I used on main system to find the problem.

 

[DeathtoGnomes]

Yes, as i said, I even reformatted drive many times and reinstalled Windows 10/11 even tried Win7 on it.

Did you use NVCleanstall to install drivers? Try older versions, I seem to recall something about 4.93 or 4.92 (somewhere around there) being ideal for that 1080 card. . Also, have you gone thru the BIOS settings?

 

[amit_talkin]

Did you use NVCleanstall to install drivers? Try older versions, I seem to recall something about 4.93 or 4.92 (somewhere around there) being ideal for that 1080 card. . Also, have you gone thru the BIOS settings?

Yes I used older version drivers. I dont remember how many I tried, but I even tried 1st driver which came out for 10 series for Windows 10 and 11.
I upgraded/downgraded bios, tried different RAM. I eventually swapped whole system parts one by one thats how I came to this conclusion.

Right now, I am on 1080Ti, with both SSDs. And I can already see those errors in event viewer. System runs fine, i just finished playing some games. But if I try to open any of the drives on SSD, system will get unresponsive and eventually hang.
If I just swap to 1080GTX , then without making any changes to system or drivers at all, everything will work. NO errors, even I can play games from one of those nVme ssd.

 

[Vario]

This sounds similar to me. I have been having an issue with 2x 970 Evo Plus 2TB and one of the SSD keeps disappearing. It first started with 1x970 Pro 512 + 2x970 Evo Plus 2TB, but I isolated it down to crashing with the 2x970 Evo Plus 2TB. Card is a 3060 12GB. Occasionally my system hangs accessing one of the 970 Evo Plus 2TB. If I run all three SSD, it will crash within minutes. The most recent 970 Evo Plus 2TB that I added which lead to this occuring was RMA'd and returned as 100% working, no fault found, then tested in another system with no issues, so I figure it is the motherboard. Maybe it is the videocard?

Here was my thread

Z370 Taichi - PCI-e issue

I have been running this board since 2017. Since December 2021, I have run it as follows with no issues: PCIE2: Nvidia RTX 3060 12GB M.2_3 Samsung 970 Pro 512GB M.2_2 Old Samsung 970 Evo Plus 2TB M.2_1 Inland Premium NVMe 1TB In June 2022, I swapped the Inland for a second Samsung 970 Evo...

www.techpowerup.com

While I said in my thread that it was at last working with 2x 970 Evo Plus 2TB, it is in fact not.

 

[80251]

If the OP's system was working fine for a year with the exact same hardware (and BIOS presumably) and then suddenly developed problems could it be a windows 10 update is causing these problems?

 

[amit_talkin]

This sounds similar to me. I have been having an issue with 2x 970 Evo Plus 2TB and one of the SSD keeps disappearing. It first started with 1x970 Pro 512 + 2x970 Evo Plus 2TB, but I isolated it down to crashing with the 2x970 Evo Plus 2TB. Card is a 3060 12GB. Occasionally my system hangs accessing one of the 970 Evo Plus 2TB. If I run all three SSD, it will crash within minutes. The most recent 970 Evo Plus 2TB that I added which lead to this occuring was RMA'd and returned as 100% working, no fault found, then tested in another system with no issues, so I figure it is the motherboard. Maybe it is the videocard?

Here was my thread

Z370 Taichi - PCI-e issue

I have been running this board since 2017. Since December 2021, I have run it as follows with no issues: PCIE2: Nvidia RTX 3060 12GB M.2_3 Samsung 970 Pro 512GB M.2_2 Old Samsung 970 Evo Plus 2TB M.2_1 Inland Premium NVMe 1TB In June 2022, I swapped the Inland for a second Samsung 970 Evo...

www.techpowerup.com

While I said in my thread that it was at last working with 2x 970 Evo Plus 2TB, it is in fact not.

Looks like similar issue to mine. I forgot to mention that even with 1 SSD plugged in, it creates such issue. no matter which nvme slot.
I have WD SN550, which I tried as well. Gives same problem with 1080Ti. Technically it should be the motherboard which causing the problem, but then all works fine with 1080GTX GPU with both SSD plugged in. Now, if 1080Ti may be causing the problem, then again all works fine with 1080TI if we remove nvme ssds. Everything is confusing.

Today I have just updated my BIOS to F15d ( AGESA V2 1.2.0.7 ). I will do some testing again. I will check for any ground power issue with chasis and motherboard. Will try running 1080Ti on stock cooler as well.
Will report how it goes.

If the OP's system was working fine for a year with the exact same hardware (and BIOS presumably) and then suddenly developed problems could it be a windows 10 update is causing these problems?

It shouldn't be. Since I am on Win11 now. I have formatted PC many times since then. tried different combination of OS and drivers. Its not drivers or Windows update related issue for sure.

 

[AsRock]

Tried underclocking the 1080 see if it might be a power delivery issue ?, Afterburner to underclock and monitor with HWinfo64.

 

[80251]

I had some bizarre issues with my system and the HDD's and SATA SSD's disappearing and I also have a 1080ti. For me though, using dielectric grease on all the SATA power and data connectors seems to have eliminated the problem, but I don't think this would make any difference with the kind of problems you're having (and my one NVME M.2 SSD has never experienced these issues....yet).

 

[Assimilator]

The fact that it was working fine for months, and then wasn't, means that something changed. Did you update the board/GPU/SSD firmware before this started happening?

 

[Count von Schwalbe]

Check your motherboard manual to ensure lanes are not shared between the GPU slot and any M.2 slots.

A very peculiar issue.

 

[amit_talkin]

Just updating the thread. Found out that my PSU is the issue ( Corsair RM850 ). Somehow it fails to activate its fan when it runs hot. Now, I am not sure that is the main cause of the problem i am facing, but then it makes both of my nVme SSDs to run very hot even at idle. I am talking about 90-95c temp! I switched to my secondary PSU for like 15 days now and everything is working fine.

I m sending back PSU for RMA citing that FAN doesnt work and it runs hot. I hope they will rectify the issue I am facing and not just fix the fan and send it back.

 

[80251]

If the fan in the PSU is affecting temps of the M.2 SSD's to that degree then I'd be looking at my case airflow.

 

[Count von Schwalbe]

If the fan in the PSU is affecting temps of the M.2 SSD's to that degree then I'd be looking at my case airflow.

A very peculiar issue to be sure, but the power output could have been affected by the high temps. We will see when a new one comes in.

 

[joemama]

Just updating the thread. Found out that my PSU is the issue ( Corsair RM850 ). Somehow it fails to activate its fan when it runs hot. Now, I am not sure that is the main cause of the problem i am facing, but then it makes both of my nVme SSDs to run very hot even at idle. I am talking about 90-95c temp! I switched to my secondary PSU for like 15 days now and everything is working fine.

I m sending back PSU for RMA citing that FAN doesnt work and it runs hot. I hope they will rectify the issue I am facing and not just fix the fan and send it back.

A failing PSU shouldn't make your NVME SSDs hot, usually only intensive writing to the SSD would cause it to heat up to such temperatures. This could be another issue but you should check the task manager if there were any unusual writings done on the SSD during idle.

 

[amit_talkin]

If the fan in the PSU is affecting temps of the M.2 SSD's to that degree then I'd be looking at my case airflow.

A failing PSU shouldn't make your NVME SSDs hot, usually only intensive writing to the SSD would cause it to heat up to such temperatures. This could be another issue but you should check the task manager if there were any unusual writings done on the SSD during idle.

Currently with my new PSU, both nVme stays around 59-69c on full load. Previously with faulty PSU, even when one of the nVme staying idle had max temp. like 90c without any activity.
I guess there must be a connection with PSU and nVme running hot. I did check voltage in hwinfo, never saw anything suspicious. So its all confusing .

 

[80251]

Is the cooling fan on the new PSU always running?

 

[amit_talkin]

Is the cooling fan on the new PSU always running?

No, its my spare PSU ( Antec 750w ). I yet to send my faulty PSU for RMA.

 

[noel_fs]

Huh very interesting

 

[maxfly]

I wonder if the rm850 was overvolting whichever rail your drives were on causing the increase in temps? You would have to have terrible case flow for your PSU fan to have such an effect on only your m.2s. You would surely have noticed increased temps across the board.

 

[amit_talkin]

I wonder if the rm850 was overvolting whichever rail your drives were on causing the increase in temps? You would have to have terrible case flow for your PSU fan to have such an effect on only your m.2s. You would surely have noticed increased temps across the board.

Yea I am thinking the same. Since I m running custom water loop, there are 4x120mm rad fans pulling air from inside. There is 1x 240mm fan in front to pull the air inside the case. However this is not best cooling for an air based setup, it is still enough for air flow for components inside the case. I didn't notice temp. spike on any other devices except the nvme. Now since my PSU was out of warranty, I placed it in my Office system. It works fine in there because it only has 1 nvme SSD and load on PSU is not as much as my primary system. Issue arises only if it gets hot on long gaming session. So in the end, I just keep it on office system and live with it.

 

[dub97gt]

To me it sounds like a power issue.. Late reply I know but I figured I'd chime in as I dealt with something similar and maybe the info could help someone.

I finally decided to spring for a couple m.2 drives after only running 2.5" ssd's for 8 years. For some reason my PC kept killing them. It would detect and format fine, then when I copied all my files over it would die. I would RMA them and Memory Express would confirm they were in fact dead. I think I returned 3 before finding the answer - cheap Amazon sleaved PSU cable extensions.

They were never a problem before because the 2.5" drives weren't pulling power from the motherboard, but I think the bad quality cables were starving the motherboard and GPU of power and somehow the m.2 drives thought it was worth dying over.

PC Specs:
Aorus X570 Master
32gb G.Skill ripjaws
Ryzen 7 5800X3D
MSI Ventus 3x RTX 3090
Corsair AX1200i
Corsair MP600 Pro 2TB