Create PyOpenCL Based Dockerfiles to use an Anaconda based package manager for Scientific Computing in DeSci

Estimation

Story points: 5 SP
Estimated duration: 3 days

Estimated focus duration: 2 days
Estimated pessimistic duration: 5 days

Description

Click to expand

Who

1. @avimanyu786

What

1. Create a PyOpenCL-based Dockerfile to support a conda environment.
2. The Dockerfile should be able to set up an environment for both ML and Scientific Computing.

How

1. Start from a minimal Ubuntu 20.04 base image.
2. Install Mamba - a drop-in replacement for conda, offering higher speed and more reliable environment solutions.
3. Create a conda environment and install necessary packages.
4. Add conda channels that facilitate retrieval of scientific computing packages.
5. Perform security checks similar to the existing setup.

Why

1. To provide a flexible environment that supports conda.
2. To ease the deployment of ML and Scientific Computing projects.
3. To ensure security by scanning dependencies.

When

1. This should be done in the current sprint or as soon as possible.

Acceptance Criteria

Click to expand

1. The Dockerfile successfully builds an image.
2. The image includes the Mamba distribution with channels defaults, conda-forge, bioconda and many more.
3. Able to create a conda environment.
4. Security checks pass for installed packages.
5. Documentation on how to use the new image through SPD.

Work Breakdown Structure (WBS)

Task	Description	Duration	Status	Start Date	End Date	Comment
A	Research on best practices for Anaconda in Docker	4 Hrs	Done	September 13 2023	September 13 2023
B	Create the NVIDIA GPU Dockerfile with Mamba	4 Days	Done	September 14 2023, September 28 2023	September 15 2023, September 29 2023
C	Test the NVIDIA GPU Dockerfile	2 Hrs	Done	September 29 2023	September 29 2023
D	Create the AMD GPU Dockerfile with Mamba	1 Week	Done	September 29 2023	October 6 2023
E	Test the AMD GPU Dockerfile	1 Week	Done	September 29 2023	October 6 2023	#30 (comment 1591396025)