Skip to content

Missing documentation on how to use the launcher with GPU programs such that each job runs on a different GPU? #61

@mauriciogtec

Description

@mauriciogtec

Is there any documentation/examples on how to use the launcher with a GPU Tensorflow or Pytorch program correctly? Specifically, is it possible to use the launcher such that each job is assigned a different GPU of a multi-GPU node? Something like --gpus-per-task 1.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions