Question Motherboard NVME M2 Drive Detection Issues after new GPU Install

[rowleyfan3620]

Hi there,
I have been running an AsRock B550M PRO4 motherboard with a Samsung EVO 970 NVME drive + 1660Ti GPU successfully without issue since Spring 2021.

Just last week, I finally got the chance to upgrade my GPU to a 6700XT, and ever since I installed the GPU I've been having issues with the motherboard's ability to detect the NVME drive in the M2_1 slot.

Sometimes (I'd anecdotally estimate about 30% of the time), when I boot, the motherboard doesn't recognize my NVME. I've confirmed this via booting to the Bios and seeing what drives the BIOS is able to see.

Windows is installed on my NVME, so I can't boot to windows. I'll restart the computer and the Motherboard will usually see the drive on the restart and I can use Windows fine.

I'm able to use Windows fine for some time (~1-2 hours) before my windows completely crashes. My screen goes black, I lose mouse control, and eventually my computer goes into a black state and shuts down. When I finally restart the computer and get to windows, my display driver is bad on the Windows -- I have to uninstall the driver and reinstall it to get it working again. I suspect this is a symptom of the issue, not the actual problem...as the problem is pre-boot to Windows...it's not a problem with Windows itself.

Fixes/Tests I've attempted:

• I've updated the BIOS to the latest version.
• I've updated my NVME Drive's firmware to the latest version.
• I tried unplugging the NVME drive and plugging it back in, thinking it could be a loose connected. I've done this twice.
• I attempted a CMOS reset via battery removal method.
• I normally run an NVME and a SSD...I tried running just the NVME for a while (unplugging the SSD), but was still experiencing the error.
• Per the recommendation of this post, I switched my motherboard PCI gen from "Auto" to "Gen3".

**CURRENT TEST: I have switched my NVME drive to the other M2 slot on the board.

That said, ultimately, I'm confused...How would a new GPU interfere with the motherboard's ability to see the NVME drive? Maybe I have a faulty motherboard, sure -- but why would the issue only be appearing now?

Any thoughts / questions / ideas would be helpful.

Some additional specs/notes:

• Running a brand new 1440p monitor via DisplayPort + 1080p monitor via HDMI to the graphics card
• I remember when I first build this computer in Spring 2021, there was an issue with the motherboard that I solved via CMOS reset battery-removal. Again, I haven't had issues since then.
• I was running a NVidia 1660Ti before this. I ran DDU on my machine before shutting it down and installing the AMD 6700XT card.
• I am running the NVME drive the M2_1 slot and the GPU in the PCIE Slot 1.

HARDWARE SPECIFICATIONS

Operating System	Windows 10 Home 64-bit
Motherboard	ASRock B550M PRO4 (Manual here for reference)
RAM	16.0GB Dual-Channel DDR4 @ 1796MHz (16-16-16-36) = G.SKILL TridentZ Series 16GB (2 x 8GB) DDR4 3600
CPU	AMD Ryzen 5600X
CPU Cooler	Noctua NH-D15
GPU	XFX Speedster AMD Radeon RX 6700XT
PSU	EVGA 210-GQ-0650-V1 650 GQ, 80+ GOLD 650W
Storage 1 (OS, problem Drive) - M2	Samsung SSD 970 EVO 1TB NVME
Storage 2 (Games, separate, no OS) - SATA	Samsung SSD 860 EVO 1TB SSD
Case	Fractal Design Meshify C Mini Black Micro ATX

 

[Darkbreeze]

What are your FULL hardware specifications including EXACT power supply model and approximate age of that unit. Knowing the specific models or at least series of other components would be helpful as well. For example, there are a lot of different "6700 xt" cards out there and many of them have distinctly different power requirements and configurations. And that's just one item.

 

[rowleyfan3620]

What are your FULL hardware specifications including EXACT power supply model and approximate age of that unit. Knowing the specific models or at least series of other components would be helpful as well. For example, there are a lot of different "6700 xt" cards out there and many of them have distinctly different power requirements and configurations. And that's just one item.

Sure thing, thanks for responding Darkbreeze. I've updated my original post with full specs. Let me know if I'm missing anything helpful -- first time posting to TH!

 

[boju]

I've been looking around quite a bit to see if your motherboard has the ability to change pcie gen. This is closest thing i got to kind of acknowledging it does - https://forum.asrock.com/forum_posts.asp?TID=19422&OB=DESC - but will just have to ask you. If you can try change it to 3.

 

[Darkbreeze]

The spec is supposed to be automatically backwards compatible isn't it? That's what I was led to believe by some of the reviews of the newer standard I looked at but maybe I'm remembering wrong? Seems to me everything 3.0 through 5.0 are supposed to be inter-compatible.

 

[boju]

That's right. All should be compatible and quite frankly, these suspected issues relating to pcie 4, including usb, or whichever else this problem may have extended to, should have been addressed with bios updates long ago. Maybe not entirely. Running pcie 4 graphics card could be causing conflicts with his pcie 3 nvme.

If can't change pcie gen to at least try see what happens, im not sure what else can be done.

 

[rowleyfan3620]

Thank you all for the post & replies!

I've changed my motherboard to pcie gen3. The tricky thing about this issue is that it's seemingly intermittent and random...so testing isn't exactly an easy thing to do.

I switched to gen3 and I'll post on here with my findings.

If this doesn't work, planned upcoming tests:

• Assuming the m2_2 slot works on the board, I will try switching the nvme drive to the m2_2 slot and see what we got.

 

[rowleyfan3620]

Well, that was fast! After switching to pci gen3, I got the crash in about 15 minutes.

I've switched my motherboard gen back to "auto" and switched my nvme drive to the m2_2 slot. I'm able to boot into windows okay via the second slot.

Based on my motherboard spec - Would there be any significant performance loss by switching to the board's M2_2 slot?

I'll update the main post thus far.

For reference, here's the error I'm getting on Windows after I reboot post-crash. Again, I don't think this is an issue with Windows per se -- I think this is a symptom -- but maybe others in the future will see this.

 

[Darkbreeze]

If it is working fine in the secondary M.2 slot run by the chipset but not in the one run by the CPU, I'd pull the CPU and make absolutely certain that your heatsink/cooler aren't cocked or too tight in one area more than another, and that there are no bent pins on the CPU.

Also, you might try laying the PC on it's side so that the weight of the D15 is not pulling down on the CPU and socket. That B550m Pro4 has a somewhat thinner substrate than some other higher end or higher tiered boards and the weight of the cooler COULD potentially be causing it to bend or flex in the socket area which can act in some cases as though there were occasionally bent pins when or if contact between the CPU pins and receivers in the socket bed is broken or not great. Worth a try anyhow and costs nothing to see.

 

[rowleyfan3620]

If it is working fine in the secondary M.2 slot run by the chipset but not in the one run by the CPU, I'd pull the CPU and make absolutely certain that your heatsink/cooler aren't cocked or too tight in one area more than another, and that there are no bent pins on the CPU.

Also, you might try laying the PC on it's side so that the weight of the D15 is not pulling down on the CPU and socket. That B550m Pro4 has a somewhat thinner substrate than some other higher end or higher tiered boards and the weight of the cooler COULD potentially be causing it to bend or flex in the socket area which can act in some cases as though there were occasionally bent pins when or if contact between the CPU pins and receivers in the socket bed is broken or not great. Worth a try anyhow and costs nothing to see.

Thank you for these thoughts!

(Also, just to mention - I know you're talking about the CPU + CPU cooler here -- but I do have a GPU Brace installed as well. So I know the weight of the GPU isn't an issue).

However, I'm coming back to the fact that all this has worked without issue for ~2 years now. The only variable I changed before and after the issue's appearing is the new GPU + new monitor. That's it. Ultimately this is what's stumping me.

Four questions, I'd like your (and others) thoughts on:
(1) Is there any significant bad thing to using the second M2 slot? Is the speed loss significant?
(2) Let's say the PCIE slots are bad/defective in the motherboard. Could that have any negative long-term impact on any other component?
(3) I'm wondering if my motherboard has been defective for two years, and I haven't known it. I did have to do a CMOS reset in 2021 with the initial build. Maybe it's been defective, and it's only being exposed because a new, more powerful GPU is pushing it to its limits - limits it hasn't gone to before. Is there any validity to this reasoning?
(4) How can I know for sure that it's a motherboard issue and not something else?


To be honest, I'm leaning towards buying a new motherboard at this point. Seems silly to cut corners and try to make this motherboard work, when I've invested so much energy into other parts of the build. Any recommendations on a new motherboard, if I get that far? I'm running a microATX case.

 

[Darkbreeze]

There is SOME loss of performance, supposedly, and I'd bet that it's mostly due to latency from longer traces/circuit termination length than anything else, but it's almost certainly not enough that you'd ever notice it. And, not being under the graphics card, so not having that additional heat in the immediate vicinity, might actually offset that somewhat in some cases.

IF the problem is related to the M.2 socket, then yes, depending on WHAT the problem is, it MIGHT evolve to affect other parts of the PCIe circuit or board later but it's an unknown really. What is MORE likely is that if you've had this GQ power supply for quite a while, it might be getting weak or developing problems since the only REAL fact here is that the 6700 xt pulls far more power than the 1660 ti, but you'd think that would affect it equally whether it was in the primary M.2 slot or a secondary one. So I'm moderately skeptical on that front as well ALTHOUGH it's still possible since the chipset driven part of PCIe circuit might be getting it's power delivery from a different part of the PSU than those powered through the CPU part of the board. Just a thought, no real fact based evidence there.

The fact that you NEED a GPU brace, and have a HUGELY heavy D15, makes the idea that the thin PCB found on most budget boards like the one you have, a lot more likely to have suffered incremental and probably intermittent, damage and problems, over time. The weight of giant cards and coolers on thinner boards can absolutely result in eventually damaging traces or solder points. We've seen it in the past but the fact is most people aren't putting giant graphics cards AND a very large cooler like the D15, on a board like the Pro4. Some, but probably most doing that are going to have at least mid tiered boards. Just another thought really.

Might not be the worst idea to consider a higher tiered board. Again, just another thought, not necessarily a "this is your solution" type statement.

 

[rowleyfan3620]

There is SOME loss of performance, supposedly, and I'd bet that it's mostly due to latency from longer traces/circuit termination length than anything else, but it's almost certainly not enough that you'd ever notice it. And, not being under the graphics card, so not having that additional heat in the immediate vicinity, might actually offset that somewhat in some cases.

IF the problem is related to the M.2 socket, then yes, depending on WHAT the problem is, it MIGHT evolve to affect other parts of the PCIe circuit or board later but it's an unknown really. What is MORE likely is that if you've had this GQ power supply for quite a while, it might be getting weak or developing problems since the only REAL fact here is that the 6700 xt pulls far more power than the 1660 ti, but you'd think that would affect it equally whether it was in the primary M.2 slot or a secondary one. So I'm moderately skeptical on that front as well ALTHOUGH it's still possible since the chipset driven part of PCIe circuit might be getting it's power delivery from a different part of the PSU than those powered through the CPU part of the board. Just a thought, no real fact based evidence there.

The fact that you NEED a GPU brace, and have a HUGELY heavy D15, makes the idea that the thin PCB found on most budget boards like the one you have, a lot more likely to have suffered incremental and probably intermittent, damage and problems, over time. The weight of giant cards and coolers on thinner boards can absolutely result in eventually damaging traces or solder points. We've seen it in the past but the fact is most people aren't putting giant graphics cards AND a very large cooler like the D15, on a board like the Pro4. Some, but probably most doing that are going to have at least mid tiered boards. Just another thought really.

Might not be the worst idea to consider a higher tiered board. Again, just another thought, not necessarily a "this is your solution" type statement.

Hey @Darkbreeze ,

Just want to clarify - I don't NEED a GPU Brace. The GPU Brace was just something I bought for a few bucks at a trade show, so I use it on my rig just for kicks and giggles. I also get the concern on the DH-15 Cooler. The thing is massive! I don't think I'd buy one again this big, but I kinda have a soft spot for it. It's a good thought...but why would I only be now having CPU Cooler issues?

Also a note - I bought the GQ PSU in Spring 2021 when I bought all the other components. Maybe I lack an understanding when it comes to PSU-aging, but that's still pretty young, right? Not even two years old. I had my last PSU for ~9 years.

I haven't had much time on my rig today, so I don't have test results from the second M2_2 slot.

Two question (both for @Darkbreeze and @boju )
(1) What would YOU do here, if you were in my situation? If the M2_2 slot works fine, would you run it ? Or would you buy a new motherboard? Would you be concerned running the NVME out of the M2_2?
(2) Can you recommend a better motherboard than the one I have? Again - I'm microATX.

Thanks!

 

[boju]

Well even if i could only run 2nd m2 slot, that would eat at me. Performance wise would be hardly different but I'd be more annoyed due to cable management (needing sata data & power cables for 2.5/3.5" drives) having to expand storage limited to single m2. If storage isn't an issue and if knew top m2 slot was all that remains wrong with the board then i wouldn't worry about it. Maybe you'll stumble on a fix. Have you contacted Asrock for unlikely but possible insightful answers?

Regarding motherboards maybe GIGABYTE B550I AORUS PRO AX or
ROG STRIX B550-I GAMING. I like Asus so id probably chose that one.

 

[Darkbreeze]

And I distinctly WON'T use ASUS anymore unless it's because of a deal that is too good to pass up and no other options that are similarly attractive on a different product. And that is ENTIRELY because their customer service has gone down the toilet. Simply too many people legitimately complaining of a lack of response, or an inadequate response or flat refusal to honor their warranty terms, on ASUS motherboards over the last five years or so. When I have to replace the BIOS ROM myself on a board that is still under warranty, and when another veteran member here gets a replacement Crosshair Hero that had a failed onboard NIC back with a replacement that has a completely cracked integrated heatsink I/O rear cover, and MANY others too numerous to list, and all of them have the same basic shortcoming in common which is poor CS, then regardless that the quality is generally pretty good I stop giving that company my business. But, that is just me.

Also, that B550I aorus Pro AX is a mini ITX board, not a microATX board. It would probably work in that same case but there are some pretty significant differences. Still, it could be an option as it looks like micro ATX B550 and X570 boards are already becoming scarce especially if you limit your options to ASRock and Gigabyte, as I mostly do. Personally I think the Gigabyte B550m Aorus Elite AX is a good option as well. I know it has a decently thick PCB and since I have worked with about five of the various Aorus Elite AX models from both camps, Intel and AMD, I can say that the boards are generally good quality and in fact my current system is running a Z690 Aorus Elite AX which I feel like is right there on par with the older Gaming 5 and 7 series boards which were always good options if you didn't want to have to take out a loan just for a motherboard.

Yes, PSU is not old really at 2 years old. Just wondered because that model has been around for a while now so it could have been in the five to seven year age group.

As far as why you might not be having issues until now, EVERYTHING that shoulders a load tends to have issues due to it after TIME. Bridges are fine for years, but the load tells after a while and supports begin to fail. Body bushings on your vehicle are fine when new but thousands of miles later they often begin to break down. Roads. Your body. Everything. This is no different. If the board has to support that weight for weeks, months and years on end, there is no reason that it might not take a toll on the support structure which in this case is the PCB itself and any amount of flexion from being moved around due to moving the tower or cleaning, or whatever, and just time hanging there, MIGHT have an effect on the socket, traces, solder points, etc. I'm not saying that IS the problem, only that it DOES happen from time to time especially with very heavy hardware.