Skip to content

RunCellpose and CUDA out of memory #230

@obruck

Description

@obruck

Description: When trying to run the cellpose plugin with ≥1 core the pipeline fails after 2-50 images. Is there any chance that cellpose would not process the images in batches or not clear processed images from memory? I would assume that the image tile and batch size are the default defined by the cellpose API so these should not consume too much memory (8 tiles = 400 Kb). Or is the model duplicated as suggested in an earlier issue #135?
Data: grayscale images of size 7000 x 10000 px
Plugin: runcellpose.py
Computing instance: 2 NVIDIA GeForce RTX 2080 Ti GPUs à 10.8 Gb memory and 20 cores (128 GB RAM memory)
Error message:
image001

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions