Using Containers to Accelerate HPC: Setup

Key requirement

The main requirement for this workshop is a personal computer with a web browser and a command line shell program.

Windows users: MobaXterm is the preferred option. The Portable Edition is best, as it does not require administrative privileges.
In this way you will be able to follow the online materials and to login to a facility with the required software stack.

Training accounts

We will be providing training user accounts to both pre-configured EC2 instances.

This tutorial is supported by the Amazon AWS Machine Learning Research Awards. EC2 images and temporary login credentials will be distributed onsite at the tutorial.

After the tutorial, you can boot our tutorial image yourself on Amazon EC2 to run through the tutorial again. We recommend you use your own EC2 key and change the password.

US-West-Oregon: ami-0fe12765123c6a840

Materials repository

To run the examples yourself, you can download the materials with:

cd ~
git clone https://github.com/supercontainers/sc20-tutorial.git
cd sc20-tutorial/exercises

Prerequisites for self-paced tutorials

Users can also install Docker and Singularity prior to attending the tutorial session. MPICH is required for running the MPI examples, too.
Here, it may be beneficial to create a Docker and Sylabs (Singularity) account in advance at https://cloud.docker.com and https://cloud.sylabs.io. These accounts will be needed to create images on Docker Cloud/DockerHub and Sylabs Cloud.

Install Singularity on Linux | docs

Install Docker on Linux | docs (unofficial)

Install MPICH on Linux | docs

Note: install scripts have been tested on a Ubuntu machine through a user that can run sudo commands without password prompts. There’s no warranty they will work in your Linux box, you should consider them as templates.

macOS or Windows machine

For Singularity, you will need to setup a Linux virtual machine, and then follow the same instructions as above.
It’s not as bad as it sounds… the main two options are:

For Docker, you can download and run installers for macOS and Windows.