GPU Support

Only CUDA-enabled NVIDIA GPU hardware is supported🔗

Cloudera Data Science Workbench only supports CUDA-enabled NVIDIA GPU cards.

Heterogeneous GPU hardware is not supported🔗

CDSW nodes that have GPUs must all use the same GPU make and model.

Known Issues with Tensorflow 2.4 and CUDA 11.2🔗

During GPU set up, the dynamic library libculsolver.so.10 is not read. For more information, see https://github.com/tensorflow/tensorflow/issues/44777.

Workaround: Enter the following commands when starting a session:

!ln -s /usr/local/cuda-11.1/targets/x86_64-linux/lib/libcusolver.so.11.0.1.105 /usr/local/cuda-11.1/targets/x86_64-linux/lib/libcusolver.so.10
!ln -s /usr/lib/x86_64-linux-gnu/libcuda.so.460.73.01 /usr/lib/x86_64-linux-gnu/libcuda.so.1

Multi-Instance GPU (MIG) Support🔗

The NVIDIA Multi-Instance GPU (MIG) feature is not supported.