dask-cuda 21.12.0a211115

Utilities for Dask and CUDA interactions

Dask CUDA

Various utilities to improve deployment and management of Dask workers on CUDA-enabled systems.

This library is experimental, and its API is subject to change at any time without notice.

Example

from dask_cuda import LocalCUDACluster
from dask.distributed import Client

cluster = LocalCUDACluster()
client = Client(cluster)

Documentation is available here.

What this is not

This library does not automatically convert your Dask code to run on GPUs.

It only helps with deployment and management of Dask workers in multi-GPU systems. Parallelizing GPU libraries like RAPIDS and CuPy with Dask is an ongoing effort. You may wish to read about this effort at blog.dask.org for more information. Additional information about Dask-CUDA can also be found in the docs.