easy5 pts · first answer countsgpu-computingdistributed-trainingpytorch

DataParallel or DistributedDataParallel

PyTorch has nn.DataParallel and DistributedDataParallel (DDP). Why is DDP recommended even on one machine?

A.DataParallel cannot run on more than one GPUB.DDP uses one process per GPU, avoiding a bottleneckC.DDP needs no communication between GPUsD.DataParallel is only for CPUs

Discussion 0

Join free to post in the discussion.

DataParallel or DistributedDataParallel

Discussion 0

Related questions