Skip to content

DOC: Show very simple usage and benchmark against continuum/miniconda3 #488

@corneliusroemer

Description

@corneliusroemer

I was asked why I prefer micromamba-docker over continuum/miniconda3.

Miniconda shows this very simple command on their readme, i love it:

docker run -i -t -p 8888:8888 continuumio/miniconda3 /bin/bash -c "\
    conda install jupyter -y --quiet && \
    mkdir -p /opt/notebooks && \
    jupyter notebook \
    --notebook-dir=/opt/notebooks --ip='*' --port=8888 \
    --no-browser --allow-root"

No need for dockerfile, just run this snippet.

Among the advantages for micromamba are:

  • much smaller image
  • wide variety of base images (alpine, ubuntu, cuda, ...)
  • no annoying default base environment
  • no annoying default channels

I tested speed, and indeed it is! I think you could add this to the docs to show a) how simple it is to use the container, and b) that it is more performant than miniconda

Micromamba:

docker run -i -t mambaorg/micromamba /bin/bash -c "\
    time micromamba install -yn base -c bioconda -c conda-forge python=3.10 nextclade -y --quiet && bash"

real    0m9.570s
user    0m9.574s
sys     0m3.865s
docker run -i -t -p 8888:8888 continuumio/miniconda3 /bin/bash -c "\
    time conda install -yn base -c bioconda -c conda-forge python=3.10 nextclade -y --quiet && bash"

real    0m22.089s
user    0m14.953s
sys     0m3.726s

In addition this, the pull time is of course much shorter for micromamba (20 seconds for miniconda on 100Mbps, vs 3 seconds for micromamba)

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