Distributed¶

import dask
from dask.distributed import Client

client = Client() #processes=True, [n_workers=n_cores, threads_per_worker=1]
client = Client(processes=False) #[n_workers=1, threads_per_worker=n_cores]
client = Client(n_workers=2, threads_per_worker=1) #explicitely set the workers and threads

#client.close()
client.restart()

if __name__ == '__main__':  # This avoids infinite subprocess creation
   from dask.distributed import Client
   client = Client()