RuntimeError: CUDA error encountered, when using cuml

Question

When I use rapids I always meet errors:

Now, I run:

from cuml.datasets.regression import make_regression
data, values = make_regression(n_samples=200000, n_features=50,
                               n_informative=7, bias=-4.2,
                               noise=0.3, random_state=10)

The error message is:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/conda/envs/rapids/lib/python3.9/contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "/opt/conda/envs/rapids/lib/python3.9/site-packages/cuml/internals/api_decorators.py", line 473, in inner
    ret_val = func(*args, **kwargs)
  File "regression.pyx", line 186, in cuml.datasets.regression.make_regression
  File "handle.pyx", line 63, in pylibraft.common.handle.Handle.__cinit__
RuntimeError: CUDA error encountered at: file=/opt/conda/envs/rapids/include/raft/core/handle.hpp line=301:

I use WLS2 Ubuntu20.04 and docker

MAND
docker pull nvcr.io/nvidia/rapidsai/rapidsai-core:22.10-cuda11.5-runtime-ubuntu20.04-py3.9
docker run --gpus all --rm -it \
    --shm-size=1g --ulimit memlock=-1 \
    -p 8888:8888 -p 8787:8787 -p 8786:8786 \
    nvcr.io/nvidia/rapidsai/rapidsai-core:22.10-cuda11.5-runtime-ubuntu20.04-py3.9

My GPU is nvidia GeForece 1080.

I don't know what is the problem.

Answer 1

RAPIDS 22.10 requires a GPU with Compute Capability 7.0 or higher on WSL 2. This requirement is documented in the installation guide at https://rapids.ai/wsl2.html.

The GeForce 1080 has Compute Capability 6.1, and is not supported at this time. See https://developer.nvidia.com/cuda-gpus for additional references on Compute Capability. The RTX 2000 series or newer are supported by RAPIDS on WSL 2. Alternatively, you can use the GTX 1080 with a native installation of Ubuntu instead of Windows via WSL 2.

Thank you for using RAPIDS!