GPU & SSD causing Windows freezes - Could it be mobo or CPU?

[NobleGamer]

Issues

Windows 10 would freeze during:

• Gameplay
• When I quit a game
• Some read/write operations of my secondary M.2 SSD

 

When this happens, typically the system freezes until I shutdown or restart.

 

The only Application or System event logs that sometimes occur would be like this:

 

Quote

Warning    5/12/2024 8:30:31 PM    disk    51    None    An error was detected on device \Device\Harddisk2\DR2 during a paging operation.
Error    5/11/2024 8:55:47 AM    disk    11    None    The driver detected a controller error on \Device\Harddisk3\DR3.

Or this:

Quote

Level    Date and Time    Source    Event ID    Task Category
Warning    6/1/2024 9:19:33 PM    disk    153    None    The IO operation at logical block address 0x42f3130 for Disk 1 (PDO name: \Device\0000003d) was retried.

 

 

My specs are:

• Windows 10 Home 64 bit
• ASUS ROG STRIX B550-F (BIOS version 3607, latest from 2024/04/03)
  • Has two m.2 slots: pcie 4 x4 and pci 3 x4
  • Has two pcie x16 slots: pcie 4 x16 and pcie 3 x4
  • Note: The second slots for both m.2 and x16 share bandwidth with different slots that I do not use
  • This motherboard is no longer under warranty
• AMD 5600X CPU
  • This was an RMA replacement by AMD in late 2020 due to a memory controller issue where it would not POST in spite of replacing RAM
• 2x8GB Corsair Vengeance RAM (3200MHz DOCP profile)
• EVGA RTX 3080 FTW3 Ultra
• Samsung 970 Evo Plus NVME boot drive (1 TB)
• Samsung 970 Evo Plus NVME storage non-boot drive (2 TB) 
• Sandisk microSDXC Class 10 in USB 3.0 adapter (400 GB)
• ASUS Wifi 5 pcie card + external antenna
• Corsair RM850X (850W 80+ gold)

 

Troubleshooting steps that I have performed, starting with the most recent ones:

• Putting the RTX 3080 in mobo's second x16 slot, and having an empty first x16 s;pt
• Swapping RTX 3080 for an MSI GTX 680 Twin Frozr III
• Putting a Kingston OEM nvme m.2 SSD (512GB) in the same slot as the 2TB
  • This seems to work okay, as I haven't had a freeze when doing various copy, read, and move actions from it
• RMA'd Samsung 2 TB storage drive
  • Samsung said they reflashed it and it tested OK
• Wiped boot SSD drive, and did a clean Windows 10 install+updates with only Steam, NVIDIA, and Hulu apps installed (because my last clean Win 10 install was several years ago)
• Verified SSDs have latest firmware and install latest Samsung NVME driver
• Ran CrystalDiskInfo, which showed no obvious warning signs, and drive life at 98%+
• Ran ASUS BIOS short and extended namespace tests of the SSD, which both succeeded
• Secure Erase of the SSD before sending it to Samsung, who found no issues but did their own reflash & wipe
• Put my 2TB SSD drive with my games + docs + saved games into a nvme m.2 enclosure with USB C
  • This is when the storage errors with crashes appeared to stop
• Moved the games I was playing to a different drive (not the one in the enclosure with games/docs)
• Manual NVIDIA driver install, instead of using Beta app that doesn't require login
• Ran DDU to clean all NVIDIA drivers + settings in safe mode with disabled windows updates & disabled networking until after...
• Clean NVIDIA driver install without GeForce Experience
• Completely removed and reseated the GPU into the pcie 4.0x16 slot
  • This is when the issues stopped for at least a few days, but came back later

 

The workarounds that seem to prevent issues:

• These prevent SSD related freezes that I correlated to the app error messages from earlier:
  • Putting 2TB SSD into external USB C nvme m.2 enclosure
  • Kingston OEM 512GB SSD works fine in the second m.2 slot where I previously put the 2TB drive
• These prevent game related freezes:
  • Swapping RTX 3080 for my spare MSI GTX 680 Twin Frozr III
  • Putting the RTX 3080 in mobo's second x16 slot

 

My questions:

• How likely is it that this is a motherboard issue?
• Any chance that this could be a CPU issue?
  • Would I have to get a different CPU to rule this out?
• Any other troubleshooting I can perform to confirm the root cause?

[191x7]

1 hour ago, NobleGamer said:

Issues

Windows 10 would freeze during:

• ...

You are describing similar issues to mine when the CMOS battery on my Asus X570-P drops below 2.90V (multimeter measured).

Check your CMOS battery voltage or replace the battery outright. Seems that on Asus Am4 boards the second m.2 slots rely on the CMOS battery voltage.

[NobleGamer]

33 minutes ago, 191x7 said:

You are describing similar issues to mine when the CMOS battery on my Asus X570-P drops below 2.90V (multimeter measured).

Check your CMOS battery voltage or replace the battery outright. Seems that on Asus Am4 boards the second m.2 slots rely on the CMOS battery voltage.

Thanks for the tip. I wonder if that could also influence gaming or the GPU too?

[Tridefender]

19 minutes ago, NobleGamer said:

Thanks for the tip. I wonder if that could also influence gaming or the GPU too?

pcie clocks, I guess 

[NobleGamer]

47 minutes ago, Tridefender said:

pcie clocks, I guess 

Perhaps, thanks, though I thought the BIOS battery was limited to storing BIOS config and datetime clock - Neither of which I found issues with. I would think pcie timings would be too small to lean on the battery.

 

Also if true, I don't know why it won't effect all of my m.2 SSDs nor all of my GPUs. The issue is limited to when my RTX 3080 is my pcie 4 x16 slot (it works in my pcie 3 x16 slot running at pcie 3 x4), and my Samsung SSD in my second m.2 slot (pcie 3x4).

 

ChatGPT can't substantiate the theory either:

 

Quote

ChatGPT said:
No, a computer's CMOS battery does not control PCIe clocks or timings. The CMOS battery’s primary role is to provide power to the CMOS chip, which stores BIOS settings and keeps the system clock running when the computer is turned off. It ensures settings like the date, time, and certain BIOS configurations are retained when the computer is off.

 

PCIe clocks and timings are controlled by the motherboard’s chipset and other clock-generation components. These are hardware functions that operate independently of the CMOS battery, meaning that PCIe performance, stability, and timing are not influenced by the battery. However, if the CMOS battery is low or dead, BIOS settings may reset to default values when the computer powers off, which could include default PCIe configuration settings if you had customized them.

 

[191x7]

12 hours ago, NobleGamer said:

Perhaps, thanks, though I thought the BIOS battery was limited to storing BIOS config and datetime clock - Neither of which I found issues with. I would think pcie timings would be too small to lean on the battery.

 

Also if true, I don't know why it won't effect all of my m.2 SSDs nor all of my GPUs. The issue is limited to when my RTX 3080 is my pcie 4 x16 slot (it works in my pcie 3 x16 slot running at pcie 3 x4), and my Samsung SSD in my second m.2 slot (pcie 3x4).

 

ChatGPT can't substantiate the theory either:

 

 

Talking from experience here. Had the issue two times in my system and two other cases here on the forum. All of them Asus boards, same symptoms same solution.

[NobleGamer]

On 11/9/2024 at 7:43 PM, 191x7 said:

You are describing similar issues to mine when the CMOS battery on my Asus X570-P drops below 2.90V (multimeter measured).

Check your CMOS battery voltage or replace the battery outright. Seems that on Asus Am4 boards the second m.2 slots rely on the CMOS battery voltage.

 

On 11/9/2024 at 8:37 PM, Tridefender said:

pcie clocks, I guess

 

My PC had a few more freezes with the RTX 3080.

 

I swapped out my existing CR2032 for another one in my battery drawer, and my PC had a few more freezes.

 

Took a while for me to find my multimeter, otherwise I would've replied sooner - Thanks.

 

Turns out the original battery was registering 3.17V, well above the nominal 3V and the ASUS-required 2.9V, whereas the battery I got from the drawer was a mere <1V.

 

So it would seem that my mobo's original CMOS battery has been good all along, and I disposed of my spare one.

 

I'll do some more testing, but assuming the freezing persists even when using the pcie 3 x16 @x4, I'm thinking this is a mobo issue unless there's some other theory I can test out.

[191x7]

12 minutes ago, NobleGamer said:

 

 

My PC had a few more freezes with the RTX 3080.

 

I swapped out my existing CR2032 for another one in my battery drawer, and my PC had a few more freezes.

 

Took a while for me to find my multimeter, otherwise I would've replied sooner - Thanks.

 

Turns out the original battery was registering 3.17V, well above the nominal 3V and the ASUS-required 2.9V, whereas the battery I got from the drawer was a mere <1V.

 

So it would seem that my mobo's original CMOS battery has been good all along, and I disposed of my spare one.

 

I'll do some more testing, but assuming the freezing persists even when using the pcie 3 x16 @x4, I'm thinking this is a mobo issue unless there's some other theory I can test out.

Have you tried reseating the CPU, checked for bent pins?

Tried reseating the RAM? Maybe done a Memtest86+ check or a Passmark Memtest86 run?

What does Hard Disk Sentinel say about your drives?

Does the same happen when using Windows 11?

Tried using an older bios version? Latest chipset drivers?

SSD temperatures under load? SSD cooling?

[NobleGamer]

1 hour ago, 191x7 said:

Have you tried reseating the CPU, checked for bent pins?

No, because AMD replaced it a few years ago, and this hasn't been an issue until the past couple of months.

 

I guess I have no choice if I'm going to rule out some pin getting bent while in-socket these past few years...

 

1 hour ago, 191x7 said:

Tried reseating the RAM?

I suppose I could try that, though I've received no DRAM mobo error lights (like my CPU RMA) nor event log errors regarding that.

 

1 hour ago, 191x7 said:

Maybe done a Memtest86+ check or a Passmark Memtest86 run?

I couldn't get the regular one to run with a boot USB, but haven't tried passmark's version. So I'll see if that's any better.

 

1 hour ago, 191x7 said:

Does the same happen when using Windows 11?

I haven't upgraded yet since I keep seeing an occasional new update screwing something up - I might take an image of my boot drive and do that.

 

1 hour ago, 191x7 said:

What does Hard Disk Sentinel say about your drives?

What info does it give that CDI doesnt?

When putting non-boot drive in M.2 enclosure, I no longer get Event Log errors about it. So i think Ive ruled that out. To recap some of my prior SSD troubleshooting:

 

On 11/9/2024 at 6:29 PM, NobleGamer said:

RMA'd Samsung 2 TB storage drive - Samsung said they reflashed it and it tested OK

Ran CrystalDiskInfo, which showed no obvious warning signs, and drive life at 98%+.

Ran ASUS BIOS short and extended namespace tests of the SSD, which both succeeded

 

1 hour ago, 191x7 said:

Tried using an older bios version?

I upgraded BIOS for sure - Can't remember if I also previously revereted to older version too.

 

1 hour ago, 191x7 said:

Latest chipset drivers?

I might've checked that a few months ago. Might as well check again...

 

1 hour ago, 191x7 said:

SSD temperatures under load? SSD cooling?

Internal and external SSDs all have some form of aluminum heat sink from mobo or enclosure. IIRC, SSD temps under load are 50-60C.

[NobleGamer]

On 11/11/2024 at 7:51 PM, 191x7 said:

Tried reseating the RAM? Maybe done a Memtest86+ check or a Passmark Memtest86 run?

I've reseated and have only two in (2x8GB) instead of 4, but after some time I still got the occasional freeze, and I even had a thumb drive disconnect itself while I was trying to move stuff off of it to prep it for passmark's bootable Memtest86.

 

So I got memtest86 booted & running now, and it is now 75% done with 0 errors. I'm gonna let it run through, but I'm leaning toward this more being a mobo issue since I now recall a few things that have happened in the 4 years I've had this board:

• I didn't put as many standoff as possible underneath the board
• The CPU might not have a back/pressure plate behind the mobo.
• I likely dropped a small screwdriver on the mobo within the past year. No obvious physical damage, but theres lots of circuits & such that arent obvious-looking either, though I don't recall any system issues happening suddenly thereafter.

On 11/11/2024 at 7:51 PM, 191x7 said:

Latest chipset drivers?

Sometime before or after I installed these, my onboard NIC is no longer detected at all, like there isn't even an unknown device entry in the Hardware list. Manually installing ASUS' NIC driver didn't help either.

 

Over Thanksgiving I may install Windows 11 and if all else fails replace the mobo since there's more going on than any single part or driver.

[191x7]

3 hours ago, NobleGamer said:

I've reseated and have only two in (2x8GB) instead of 4, but after some time I still got the occasional freeze, and I even had a thumb drive disconnect itself while I was trying to move stuff off of it to prep it for passmark's bootable Memtest86.

 

So I got memtest86 booted & running now, and it is now 75% done with 0 errors. I'm gonna let it run through, but I'm leaning toward this more being a mobo issue since I now recall a few things that have happened in the 4 years I've had this board:

• I didn't put as many standoff as possible underneath the board
• The CPU might not have a back/pressure plate behind the mobo.
• I likely dropped a small screwdriver on the mobo within the past year. No obvious physical damage, but theres lots of circuits & such that arent obvious-looking either, though I don't recall any system issues happening suddenly thereafter.

Sometime before or after I installed these, my onboard NIC is no longer detected at all, like there isn't even an unknown device entry in the Hardware list. Manually installing ASUS' NIC driver didn't help either.

 

Over Thanksgiving I may install Windows 11 and if all else fails replace the mobo since there's more going on than any single part or driver.

Try replacing the CMOS battery first.

[NobleGamer]

On 11/9/2024 at 7:43 PM, 191x7 said:

You are describing similar issues to mine when the CMOS battery on my Asus X570-P drops below 2.90V (multimeter measured).

Check your CMOS battery voltage or replace the battery outright. Seems that on Asus Am4 boards the second m.2 slots rely on the CMOS battery voltage.

 

An update: I no longer have freezes or crashes on occasion while playing games and my ethernet port is operational.  The only issue now appears to be that whatever NVME drive I put in the M.2 slot does not get detected.  Putting the same NVME drive in a USB enclosure works perfectly.

 

Troubleshooting steps taken since my prior comment:

• Ran full passmark memtest86 test suite on all 4 RAM DIMMs, and they all passed.
• Replaced the original battery @ 3.17V with a new sealed battery @ 3.3V minimum
  • That is even higher than 3V nominal for CR2032 as well as the ASUS weirdness you mentioned at <=2.9V
• Installed Windows 10 optional update 22H2

So I still cannot rule out a motherboard problem or Windows 10 suddenly hating my motherboard, but at least I can now rule out RAM and GPU.

During my Christmas break, I may begrudgingly upgrade to Windows 11 in spite of them still releasing updates breaking substantial game related functionality (e.g. Ubisoft/Denuvo).

 

Until then, I dub the CMOS battery comment to be the best solution thus far - Thanks.