Translations:Dask/7/en

In the script Dask-example.py, we launch a Dask cluster with as many worker processes as there are cores in our job. This means each worker will spawn at most one CPU thread. For a complete discussion of how to reason about the number worker processes and the number of threads per worker, see the official Dask documentation. In this example, we split a pandas data frame into 6 chunks, so each worker will process a part of the data frame using one CPU: