Dask Setup#

Scheduler functions#

progress(fs)

Perform outstanding calculations similar to distributed.diagnostics.progressbar.progress().

cpu_count_physical()

Return the number of physical cores in the system.

hyperthreading_info()

Detect presence of hyperthreading.

restart_cluster(client)

Restarts cluster.

setup_scheduler(args)

Start a scheduler for use of parallel environment with distributed.Client.

Schedulers#

DistributedLocalClusterScheduler([...])

Scheduler using distributed.LocalCluster for local parallelism.

DummyClient()

Dummy class mimicking distributed.Client without distributed.

ExternalScheduler(scheduler_file[, ...])

Scheduler using an externally started distributed.Cluster.

LocalThreadsScheduler(**kwargs)

Scheduler using dask without distributed.

MPIScheduler(**kwargs)

Scheduler using dask.mpi for cluster setup.

SingleThreadedScheduler(**kwargs)

Scheduler using strictly local, single threaded approach.