I was following along an example from Mr. Rocklin’s YouTube video on a Windows Server with 12 cores and 128GB RAM, but Dask never used all the processors. It only used one processors.
I thought it might have been completing too quickly, but even when I increased the size of the array, it only ever used one processor.
I was looking at this video: Dask Setup: Introduction - YouTube
and also the example here: Dask Arrays — Dask Examples documentation
@Seeker Welcome to Dask, and thanks for your question!
How are you diagnosing processor use?
but Dask never used all the processors. It only used one processors.
I think this is because the example creates only one worker. And, since we set
processses=False, we also have only one process while uses one core. Ref: Scheduling — Dask documentation
Could you please try with the default values (as shown below) and let us know if you see all cores in use?
from dask.distributed import Client
client = Client()