[SOLVED] Boot NVMe Drive not recognized with 6950XT installed

[efontyn]

I currently occupy both nvme slots on my motherboard with storage, one 1tb Western Digital Drive, and another 2tb Hynix P31 for my OS. I have been using this system without issue for months, however it started booting into automatic repair a few days ago instead of the OS as usual. I have had issues in the past with boot priority, but this was easily fixed and I never had any issues with the drive since. When I tried to boot up my system last, it will boot into automatic repair and fail as there is no OS detected obviously. In BIOS, the drive is not detected.

When troubleshooting, I found that it will be detected with an older card (HD 6570) and will also be detected in another machine just fine, and will boot as normal. However in my current system it does not want to be detected as of now. I will list my specs below. I have searched reddit and alot of the threads just end in someone saying they got their board replaced, however I have not had issues with the board and the drive and slot are not dead as all drives will be detected upon slotting the 6570 in for the 6950XT.

I also read that it can be an issue for PCIe lanes, however 5800x3d supports 24 lanes. 16 for gpu, 8 for 2 nvme drives. 24 all together so all should be working fine unless I have a false understanding. Below I will list my specs, I appreciate any tips you may have. I am running the latest bios on my board, F38a (previously was running F37 and tried an update as troubleshooting step)

The drive was also detected several times after reseating the 6950XT, and upon shutdown and reboot, the drive was once again not recognized.

Im assuming at this point it is some BIOS configuration prohibiting the drive from being recognized.

System Specs
MOBO: Gigabyte X570 Aorus Elite 1.x (Latest BIOS)
PSU: Corsair RM850x
GPU: 6950XT
CPU: 5800X3D
RAM: Corsair Ballistix 2x16gb 3600mhz CL16
Drive in Question: SK Hynix P31 Gold 2TB

 

[hotaru.hino]

A note about PCIe lanes, 16 lanes are dedicated to 2 x16 slots (they will share lanes if both are populated), 4 lanes for one of the NMVe slots, and 4 for the chipset. The other NVMe is getting lanes from the chipset.

In any case, a cursory glance on the internet tells me that the RX 6950XT is prone to massive power spikes, with some places reporting around 600-650W for brief periods of time. While this should only happen when putting a load on the card, in addition to the fact that this problem only happens when you have the card installed, either:

• Upgrade the PSU to at least a 1000W unit
• Find a new video card

 

[efontyn]

A note about PCIe lanes, 16 lanes are dedicated to 2 x16 slots (they will share lanes if both are populated), 4 lanes for one of the NMVe slots, and 4 for the chipset. The other NVMe is getting lanes from the chipset.

In any case, a cursory glance on the internet tells me that the RX 6950XT is prone to massive power spikes, with some places reporting around 600-650W for brief periods of time. While this should only happen when putting a load on the card, in addition to the fact that this problem only happens when you have the card installed, either:

• Upgrade the PSU to at least a 1000W unit
• Find a new video card

I've had no issues with shutoffs, and I've known others using 6950XT without issues using psus with less wattage, not really convinced it's an issue. I've used both drives with the 6950XT for months.

 

[hotaru.hino]

I've had no issues with shutoffs, and I've known others using 6950XT without issues using psus with less wattage, not really convinced it's an issue. I've used both drives with the 6950XT for months.

Did you change anything regarding the UEFI settings or updated it within that time?

Also power supply issues don't always happen immediately. The spike may happen for milliseconds, but the OCP doesn't trip unless it happened for say 20ms. However even that spike may have degraded something in the PSU.

 

[efontyn]

Did you change anything regarding the UEFI settings or updated it within that time?

Also power supply issues don't always happen immediately. The spike may happen for milliseconds, but the OCP doesn't trip unless it happened for say 20ms. However even that spike may have degraded something in the PSU.

Nah I didn't alter any settings. My psu is still under warranty, as a last ditch I could return and put the refund towards a 1000w unit. Would high transients show up on hwinfo, during stress tests the highest it would report was drawing ~300w

 

[hotaru.hino]

Nah I didn't alter any settings. My psu is still under warranty, as a last ditch I could return and put the refund towards a 1000w unit. Would high transients show up on hwinfo, during stress tests the highest it would report was drawing ~300w

No, because HWiNFO doesn't poll the sensors fast enough (if the sensors are even capable of reporting fast enough).

 

[efontyn]

No, because HWiNFO doesn't poll the sensors fast enough (if the sensors are even capable of reporting fast enough).

Gonna do some more testing next few days to see if I can figure out what exactly is happening, gonna try a few different drives.

 

[efontyn]

No, because HWiNFO doesn't poll the sensors fast enough (if the sensors are even capable of reporting fast enough).

Looks like I was able to solve the problem, but not in the way I expected. I believe one of my PCIe cables was loose on my PSU, I was unable to even post (GPU seemed to be undetected) after some troubleshooting today and decided to tear it down. Took out the PSU and one of the cables just fell out, must have come loose from stretching the cables accidentally while swapping so many cards.

Both drives detected with 6950XT and running great after plugging everything back in, I feel silly as hell tbh. Thanks for your tips Hotaru.