AlphaObeisance
Well-Known Member
I've been dealing with an interesting issue. I recently purchased the new ASUS Crosshair X870E Dark Hero motherboard to finally migrate from AM4 to AM5. I tossed in an AMD Ryzen 9 9950x, 64GiB Gskill RAM and my RTX 4090.
I've put my one and only m.2 into the M.2_1 slot, and my RTX 4090 into the top PCIe slot as should be done. It's my understanding that with this particular board, there is absolutely no lane splitting between the M.2_1 and the PCIe slot; yet, oddly enough.
And I cannot for the life of me figure out why in the heck my GPU insists on downgrading to x8 from x16. I even pulled the system out of the rack this morning, removed the GPU, removed the AIO cooler, and re-seated the contact plate (thermal grizz), ensured a snug but not tight cross tightening pattern on the plate, then an equal torque side to side on my AIO Arctic cooler. I cleaned the pins on the GPU and inspected for any damages yet everything seems to check out.
I updated the BIOS which had a recent release on like April, 26th, 2026 so just a few days ago; sadly this did not resolve the issue.
There is absolutely zero sag on the GPU as it's mounted in a Silverstone RM51 chasis and resides horizontally within my server rack; so the GPU sits vertically rather than the traditional horizontal orientation of a standard PC case.
I'm a bit flabbergasted right now as despite being set to Gen4 manually in the BIOS, I still can't get this thing to utilize all x16 lanes as it should.
It's not like the system won't do everything I ask it too, x8 is plenty. But me being me, I don't get the point of having the ability to do x16 and not utilizing it. It would be different if I had this thing loaded down in M.2 drives (it's got like 5 M.2 slots); which would explain the bifurcation, but at this point I'm at a loss.
I've tried everything I can think of. And despite the hate towards Nvidia the Linux community generally has (yeah yeah yea, I don't care friends). I figure I'll toss this out into the ethos and maybe some other hardware junkie out there might have encountered similar issues and know of a legitimate, or potential solution.
I've put my one and only m.2 into the M.2_1 slot, and my RTX 4090 into the top PCIe slot as should be done. It's my understanding that with this particular board, there is absolutely no lane splitting between the M.2_1 and the PCIe slot; yet, oddly enough.
Code:
lspci | grep -i nvidia
01:00.0 VGA compatible controller: NVIDIA Corporation AD102 [GeForce RTX 4090] (rev a1)
01:00.1 Audio device: NVIDIA Corporation AD102 High Definition Audio Controller (rev a1)
Code:
sudo lspci -vvv -s 01:00.0 | grep -E "LnkCap|LnkSta"
LnkCap: Port #0, Speed 16GT/s, Width x16, ASPM L1, Exit Latency L1 <4us
LnkSta: Speed 16GT/s, Width x8 (downgraded)
LnkCap2: Supported Link Speeds: 2.5-16GT/s, Crosslink- Retimer+ 2Retimers+ DRS-
LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+ EqualizationPhase1+
Code:
nvidia-smi
Sun May 3 10:40:17 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 595.71.05 Driver Version: 595.71.05 CUDA Version: 13.2 |
+-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 4090 On | 00000000:01:00.0 On | 0 |
| 0% 41C P0 47W / 500W | 2223MiB / 23028MiB | 17% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 43897 G Hyprland 601MiB |
+-----------------------------------------------------------------------------------------+And I cannot for the life of me figure out why in the heck my GPU insists on downgrading to x8 from x16. I even pulled the system out of the rack this morning, removed the GPU, removed the AIO cooler, and re-seated the contact plate (thermal grizz), ensured a snug but not tight cross tightening pattern on the plate, then an equal torque side to side on my AIO Arctic cooler. I cleaned the pins on the GPU and inspected for any damages yet everything seems to check out.
I updated the BIOS which had a recent release on like April, 26th, 2026 so just a few days ago; sadly this did not resolve the issue.
There is absolutely zero sag on the GPU as it's mounted in a Silverstone RM51 chasis and resides horizontally within my server rack; so the GPU sits vertically rather than the traditional horizontal orientation of a standard PC case.
I'm a bit flabbergasted right now as despite being set to Gen4 manually in the BIOS, I still can't get this thing to utilize all x16 lanes as it should.
It's not like the system won't do everything I ask it too, x8 is plenty. But me being me, I don't get the point of having the ability to do x16 and not utilizing it. It would be different if I had this thing loaded down in M.2 drives (it's got like 5 M.2 slots); which would explain the bifurcation, but at this point I'm at a loss.
I've tried everything I can think of. And despite the hate towards Nvidia the Linux community generally has (yeah yeah yea, I don't care friends). I figure I'll toss this out into the ethos and maybe some other hardware junkie out there might have encountered similar issues and know of a legitimate, or potential solution.
. 
