Skip to content

Memory Allocation? #1

Open
Open

Description

@egrace479

This downloader loads each batch into memory before writing the files to disk, since our initial storage plan was to use HDF5.
Since we decided to transfer the dataset into either HDF5 or webdataset format after completing the download to avoid data backup complications and difficulties with parallel data writes, this setup is open to change.

The question is whether it is worth changing to writing image data as it comes down to reduce the memory bottleneck and utilize cores more efficiently.

Activity

  1. added
    documentationImprovements or additions to documentation
    questionFurther information is requested
    on May 17, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentationenhancementNew feature or requestquestionFurther information is requested

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions