WSL2 + CUDA
See original GitHub issueDescription I’m trying to run Triton on the recently released WSL2 and a specific CUDA driver for it. The examples provided by NVIDIA work fine.
Details: https://docs.nvidia.com/cuda/wsl-user-guide/index.html
Triton Information What version of Triton are you using? nvcr.io/nvidia/tritonserver:20.03-py3
Are you using the Triton container or did you build it yourself? nvcr.io/nvidia/tritonserver:20.03-py3
To Reproduce
sudo docker run -d --restart always --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 --name triton -p 8000:8000 -p 8001:8001 -p 8002:8002 nvcr.io/nvidia/tritonserver:20.03-py3
OR
sudo nvidia-docker run -d --restart always --gpus all --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 --name triton -p 8000:8000 -p 8001:8001 -p 8002:8002 nvcr.io/nvidia/tritonserver:20.03-py3
=============================
== Triton Inference Server ==
=============================
NVIDIA Release 20.03 (build 11042949)
Copyright (c) 2018-2019, NVIDIA CORPORATION. All rights reserved.
Various files include modifications (c) NVIDIA CORPORATION. All rights reserved.
NVIDIA modifications are covered by the license terms that apply to the underlying
project or file.
WARNING: The NVIDIA Driver was not detected. GPU functionality will not be available.
Use 'nvidia-docker run' to start this container; see
https://github.com/NVIDIA/nvidia-docker/wiki/nvidia-docker ..
Expected behavior I expect Triton to start up?
Issue Analytics
- State:
- Created 3 years ago
- Comments:27 (25 by maintainers)
Top GitHub Comments
@turowicz Bless you for all the follow-up posts as you continued your research. The internet needs more people like you.
The warning message about GPU that is produced by the entrypoint code is something we are working on fixing for WSL. We haven’t spent much time with WSL and Triton so we can’t yet provide much detailed feedback on your other findings. Are you aware of the CUDA/WSL developer forum? You may find some quicker help there: https://forums.developer.nvidia.com/c/accelerated-computing/cuda/cuda-on-windows-subsystem-for-linux