WSL2 nvidia-smi: Unable to determine the device handle for GPU

Accelerated Computing CUDA CUDA on Windows Subsystem for Linux
1

Hi all,

I’m trying to get CUDA working in WSL 2 (Ubuntu 22.04) with an NVIDIA GeForce RTX 3080 Ti on Windows 11 Education (24H2), but I’m running into issues where the GPU is not being recognized in WSL.
When I run nvidia-smi inside WSL, I get:

$ nvidia-smi
Unable to determine the device handle for GPU0: 0000:C1:00.0: Unknown Error
Unable to determine the device handle for GPU1: 0000:E1:00.0: Unknown Error
No devices were found

On Windows I have successfully installed cuda_12.8.1_572.61_windows. (The driver version is 572.61.) I have also tried the latest driver 576.52 without success.

I would be very grateful for any suggestion on how to further troubleshoot the problem. Below I have already collected a bunch of system information, I hoped might be helpful.

I followed the following installation guide for setup:

The Windows-native nvidia-smi version works as expected:

$ /mnt/c/Windows/system32/nvidia-smi.exe 
Sat May 24 14:55:46 2025 
+-----------------------------------------------------------------------------------------+ 
| NVIDIA-SMI 572.61                 Driver Version: 572.61         CUDA Version: 12.8     | 
|-----------------------------------------+------------------------+----------------------+ 
| GPU  Name                  Driver-Model | Bus-Id          Disp.A | Volatile Uncorr. ECC | 
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. | 
|                                         |                        |               MIG M. | 
|=========================================+========================+======================| 
|   0  NVIDIA T1000                 TCC   |   00000000:C1:00.0 Off |                  N/A | 
| 70%   83C    P0            N/A  /   50W |     571MiB /   4096MiB |    100%   E. Process | 
|                                         |                        |                  N/A | 
+-----------------------------------------+------------------------+----------------------+ 
|   1  NVIDIA GeForce RTX 3080 Ti   WDDM  |   00000000:E1:00.0  On |                  N/A | 
|  0%   47C    P8             23W /  350W |    1975MiB /  12288MiB |      0%      Default | 
|                                         |                        |                  N/A | 
+-----------------------------------------+------------------------+----------------------+ 
+-----------------------------------------------------------------------------------------+ 
| Processes:                                                                              | 
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory | 
|        ID   ID                                                               Usage      | 
|=========================================================================================| 
|    1   N/A  N/A             844    C+G   ...eafile\bin\seafile-applet.exe      N/A      |

Note: What appears as NVIDIA T1000 is my GRAID RAID adapter.

$ wsl --version
WSL version: 2.4.13.0
Kernel version: 5.15.167.4-1
WSLg version: 1.0.65
MSRDC version: 1.2.5716
Direct3D version: 1.611.1-81528511
DXCore version: 10.0.26100.1-240331-1435.ge-release
Windows version: 10.0.26100.4061

WSL Kernel in Ubuntu:

$ uname -r
5.15.167.4-microsoft-standard-WSL2

$ glxinfo -B 
name of display: :0 
display: :0  screen: 0 
direct rendering: Yes 
Extended renderer info (GLX_MESA_query_renderer):    
	Vendor: Microsoft Corporation (0xffffffff)
    Device: D3D12 (NVIDIA GeForce RTX 3080 Ti) (0xffffffff)
    Version: 23.2.1
    Accelerated: yes
    Video memory: 77367MB
    Unified memory: no
    Preferred profile: core (0x1)
    Max core profile version: 4.2
    Max compat profile version: 4.2
    Max GLES1 profile version: 1.1
    Max GLES[23] profile version: 3.1 
OpenGL vendor string: Microsoft Corporation 
OpenGL renderer string: D3D12 (NVIDIA GeForce RTX 3080 Ti) 
OpenGL core profile version string: 4.2 (Core Profile) Mesa 23.2.1-1ubuntu3.1~22.04.3 
OpenGL core profile shading language version string: 4.20 
OpenGL core profile context flags: (none) 
OpenGL core profile profile mask: core profile 

OpenGL version string: 4.2 (Compatibility Profile) Mesa 23.2.1-1ubuntu3.1~22.04.3 
OpenGL shading language version string: 4.20 
OpenGL context flags: (none) 
OpenGL profile mask: compatibility profile 

OpenGL ES profile version string: OpenGL ES 3.1 Mesa 23.2.1-1ubuntu3.1~22.04.3 
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.10

$ ls /usr/lib/wsl/lib/
libcuda.so            libnvcuvid.so          libnvidia-opticalflow.so
libcuda.so.1          libnvcuvid.so.1        libnvidia-opticalflow.so.1
libcuda.so.1.1        libnvdxdlkernels.so    libnvoptix.so.1
libcudadebugger.so.1  libnvidia-encode.so    libnvoptix_loader.so.1
libd3d12.so           libnvidia-encode.so.1  libnvwgf2umx.so
libd3d12core.so       libnvidia-ml.so.1      nvidia-ngx-updater
libdxcore.so          libnvidia-ngx.so.1     nvidia-smi

Docker on Windows also fails to access the GPU:

PS C:\Users\XXX> docker run --gpus all -it --rm nvidia/cuda:12.3.1-base-ubuntu20.04 nvidia-smi
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running prestart hook #0: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: detection error: nvml error: unknown error: unknown
[...]
PS C:\Users\XXX> docker run hello-world
Hello from Docker!
This message shows that your installation appears to be working correctly.
[...]
2

In case someone has a similar issue and stumbles across this topic like i did:

Apparently a mixture of TCC and WDDM driver modes in multi GPU systems does result in GPU detection errors in WSL

SOLUTION: Deactivate TCC card or run both GPUs in WDDM mode.

3

Thanks a lot for sharing this!

Unfortunately, in my case, I can’t change the mode of the TCC card because it’s not really used as a traditional GPU — it’s a GRAID SupremeRAID SR-1000, which relies on the T1000 hardware but uses its own GRAID driver on top of the NVIDIA driver. If I disable it, I essentially lose access to my main storage, so that’s not really an option.

Do you happen to know if this is a known issue with WSL and mixed TCC/WDDM setups, or was this something you figured out through trial and error? Perhaps a bug report/feature request might help resove the problem in the future.
Either way, I really appreciate your help!

4

I’ve reported the issue on GitHub for WSL:

If anyone encounters the same problem and sees it as an issue, please upvote it on GitHub to help bring attention to it.

5

I did not find much on that exact issue, just your thread and mine. We both have a relatively unique setup with a mixture of TCC and WDDM. I know TCC is not supposed to work in WSL and will be blocked by the OS. It took me some time to get there, but from there is was a quick test to get it running on just the WDDM card.

Maybe we get some solution since I also would like to keep my current config if possible. Thanks for reporting it on GitHub.

6

For anybody having a similar issue: User chlimouj posted on GitHub that his setup with mixed TCC and WDDM modes works when the card in WDDM mode is assigned the lower index by the NVIDIA driver. Since the assignment is likely based on the PCI bus ID, it might be worth trying to swap the GPUs in their PCI slots as a temporary fix. Unfortunately, I can’t try this myself due to space constraints in my PC case.