I get message cuda on WSL is ok now from this: https://docs.nvidia.com/cuda/wsl-user-guide/index.html#unique_1238660826
so for deepstream? wsl2 ubuntu is so convenient for development
I confirm that it works pretty well !
With minor modification in the nvdsgst-infer plugin to work-around lack of EGL symbols in the cuda library (to enable inference debugging/execution and visualization on host).
1 Like
is it possible to get nvv4l2decoder and nvv4l2h264enc working? could anyone from NVIDIA confirm if there might be support for this in the future? currently it’s complaining about /dev/nvidia0 missing. Thanks!
I’m seeing a number of threads/questions around Deepstream & WSL… and the typical ‘solution’ is that it is not supported.
Whilst the answer is technically correct (not officially supported/tested by nVidia), it does not mean it cannot work.
Deepstream can quite easily be used within WSL.
It works with 6.0.1 & now even with 6.2.
Only minor adjustments are required to do so (recompile 1 or 2 plugins + adapt pipeline).
see deepstream_wsl (for DS 6.2)
1 Like