Cuda-gdb report internal error while using under wsl2

Developer Tools CUDA Developer Tools CUDA-GDB
1

I am trying to use cuda-dbg under wsl2, my driver version is 531.41 and cuda version is 12.1. GPU card is RTX 2060S.
when I use cuda-gdb to debug my program, It reports “Error: Internal error reported by CUDA debugger API (error=CUDBG_ERROR_INTERNAL(0xa)). The application cannot be further debugged.” after exctuing the first cuda function,and also after starting new threads.

2

截图
截图958×119 2.53 KB

3

Hello Hiao, could you please check if the registry key HKEY_LOCAL_MACHINE\SOFTWARE\NVIDIA Corporation\GPUDebugger\EnableInterface is present, with a value of (DWORD)1?
If not, create it and try again.

4

Yes，I have checked. The registry key “EnableInterface” is already set to 1(REG_DWORD).

6

Can you please provide the output of nvidia-smi?

7

Sorry, I didn’t reply in time, The nvidia-smi output is as follows

截图
截图711×316 3.23 KB

8

Thank you @hiao, can you run any cuda app outside of the debugger?
Also, can you try with the latest toolkit version 12.1U1?

9

yes, I can run cuda app outside the dubbger. And I have just tried the latest version, the question is still existing.

10

I am having a similar issue. was there ever a resolution or do we just have to wait for cuda 12.2?

11

The issue is under investigation, how did you set up your WSL environment with CUDA?

12

I am also having a similar issue. My GPU is RTX 3050 Laptop. I have tried different driver versions and cuda versions, but it didn’t work.

13

Similar issue but I just installed

  • WSL2
  • Cuda toolkit 12.3
  • Win11 driver 546.33
  • Running the sample VectorAdd in VSCode w NSight plugin
    Cant stop at breakpoint even in CPP code (e.g. main). Is this not working yet in WSL2?
14

Hi, @drstriate

Please make sure registry key HKEY_LOCAL_MACHINE\SOFTWARE\NVIDIA Corporation\GPUDebugger\EnableInterface is present, with a value of (DWORD)1

15

Thanks for this. This was set properly, but I did manage to get the samples to compile (again) after the Toolkit 12.3 Update by adding various environment settings found on this forum.

export PATH=/usr/local/cuda-12.3/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-12.3/lib:$LD_LIBRARY_PATH
export CPATH=/usr/local/cuda-12.3/targets/x86_64-linux/include:$CPATH

16

Thanks for the update !
So you now have sample built/run successfully on WSL, right ?

Does cuda-gdb ./vectorAdd work now ? If not, can you clarify your GPU and paste the output of cuda-gdb ?

17

This topic was automatically closed after 13 days. New replies are no longer allowed.