You might be surprised to learn that the summer months are the busiest time of year for the Center for Research Computing! This year, they included collaborations on a number of outreach and learning opportunities. In addition to mentoring several summer interns, the CRC also oversaw three separate academic programs.

In early June, the CRC hosted high school students for the Research Computing track as part of the campus Summer Scholars program. The goal of this course is to introduce students to advanced computational resources and show how research computing can be used to support many different career paths. The first week of the class covered both computer hardware and common software tools from a high level. The students spent a day at our Union Station data center putting into practice what they've learned by building a small computer cluster using real CRC servers. The second week focused more on applied Research Computing as we brought in guest speakers from across campus to showcase current research projects from a wide variety of fields. The second week also included a trip to Argonne National Lab where the students were able to visit the Argonne data center and get pictures in front of Aurora—currently the second fastest supercomputer in the world.

The CRC also participated in Notre Dame Global’s International Summer: Technology, Engineering, and Sciences program that was held on Notre Dame’s campus during June. This program was developed for STEM undergraduate students in Latin America and aims to deepen the students’ understanding of several scientific and technical disciplines. This is the second year in which the CRC has worked with Notre Dame Global in this capacity. To support this program, the CRC Software Team staff members provided instruction and demonstration to the 40 visiting students, focusing on the domain areas of Wireless Communication, Generative AI, and Bioinformatics. The students also had the opportunity to hear more about graduate programs at Notre Dame during their time on campus

A student presents her work at the Three Minute Thesis competition.

Across campus, the Research Experience for Undergraduates (REU) program brought in close to 400 students this summer with the CRC supervising 16 of them. Within the CRC, students took on a diverse array of projects, including explorations into Digital Democracy, Blockchain applications, and dark web analysis. The team also developed robots utilizing large language models, engaged in audiovisual collaborations, and conducted innovative research on social media dayparting.

The pinnacle of the REU program was the Summer Undergraduate Research Symposium and the Three Minute Thesis Competition. This event was the perfect platform for students to present their research findings and hone their communication skills. The symposium featured a total of 136 research posters, with 8 of them presented by the CRC students. Each poster represented the culmination of weeks of dedicated research and hard work. Topics ranged across various scientific and technological disciplines, showcasing the breadth and depth of the students' investigations.

Steve Bogol leads a tour of the Union Station data center.

Outside of the classroom, there were many social and recreational activities for the students as well. A campus fun fair, Latin dance lessons, trivia night, cornhole tournaments, a visit to the Union Station data center, and a trip to see the South Bend Cubs are just a few of the fun events. The CRC is a proud supporter of the university's commitment to fostering the next generation of researchers and scholars in a supportive and enriching environment.

Apptainer is a software application that empowers users to take full control of their operating systems without needing super-user privileges. This is achieved through the use of containers or images, terms that can be used interchangeably. A container functions like an empty room that users can configure and customize according to their specific needs. Designed for portability, Apptainer can run on any Linux distribution, making it easy to share containers across different systems.

Benefits of Apptainer

Users sometimes require software packages that are not pre-installed on our systems, typically necessitating a software installation request. With Apptainer, users can create and manage their own environments within containers, running any installed software on CRC machines. This approach offers greater convenience and flexibility, allowing users to utilize the software they need without waiting for system-wide installations.

Installation and Setup

There are multiple methods to download and install Apptainer on local machines, as detailed on Apptainer’s website. Apptainer is already installed on CRC frontends and compute nodes. To get started, simply run apptainer --help.

Obtaining Containers

Once Apptainer is installed, users can either use pre-built containers from sources like Docker Hub (Docker Hub Container Image Library | App Containerization) or create their own customized containers. Building a container requires root or sudo access on your own machine, but prebuilt containers can be pulled without root access when using CRC resources.

Using Pre-Built Containers

To obtain and use a pre-built container, use the pull command. For example, to obtain a basic AlmaLinux 9.4 container from Docker Hub:

apptainer pull almalinux.sif docker://almalinux:9.4

Once pulled, the container is stored locally and can be used as a base for further customization or as-is.

Creating Custom Containers

To create a custom container on a local machine with sudo access, use the build command along with a recipe file. A recipe file defines the container's configuration. Here is an example recipe file:

Bootstrap: docker
From: ubuntu:20.04

%post
    apt-get -y update
    apt-get -y install cowsay lolcat

%environment
    export LC_ALL=C
    export PATH=/usr/games:$PATH

%runscript
    date | cowsay | lolcat

Save the recipe file as ubuntuBuild and build the container with:

sudo apptainer build myOwn.img ubuntuBuild

Accessing and Using Containers

Apptainer offers several commands to interact with containers:

apptainer shell myContainer.img

apptainer run myContainer.img

apptainer exec myContainer.sif /usr/bin/myCommand

Using Apptainer in Batch Jobs

Apptainer can be used in our batch system. Here is an example job submission file:

#!/bin/bash

#$ -N example-job

apptainer exec path/to/container/myContainer.img runme

Where runme is the command being executed within the container. This assumes the container is executable. The script is now ready to be submitted with the qsub command.

Apptainer on GPU Nodes

For GPU usage, include the --nv flag to set up CUDA within the container. An example script for a GPU job is:

#!/bin/bash

#$ -q gpu

#$ -l gpu=1

#$ -N example-job

export APPTAINERENV_CUDA_VISIBLE_DEVICES=${CUDA_VISIBLE_DEVICES}

apptainer exec --nv tensorflow_latest-gpu.sif my_script.py

Conclusion

Apptainer provides a flexible and powerful way to run customized software environments on any Linux distribution, simplifying the process of software deployment and usage in diverse computing environments.

Friday, September 6, 2024
9:00 a.m. – 12:00 p.m.
B01 McCourtney Hall (map)

The CRC is excited to invite you to our upcoming in-person workshop, designed to enhance your skills in utilizing CRC compute resources. This session will cover the following topics: 

Please register by August 30, 2024. If you have any questions, please contact CRCSupport@nd.edu.

Every Wednesday and Thursday in July
2:00 – 3:00 p.m.
via Zoom (In-Person Training Sessions Available by Request)

This training is available to new users and current users interested in a refresher course on how to use CRC resources. Attendees learn the basics of accessing CRC resources and submitting jobs on the CRC clusters. Learn more.

Aerospace & Mechanical Engineering
852,659 CPU hours

Chemical & Biomolecular Engineering
527,873 CPU hours

Chemical & Biomolecular Engineering
490,246 CPU hours

Applied & Computational Mathematics & Statistics
463,190 CPU hours

Civil & Environmental Engineering & Earth Sciences
445,586 CPU hours

Aerospace & Mechanical Engineering
349,568 CPU hours

Aerospace & Mechanical Engineering
267,308 CPU hours

Chemical & Biomolecular Engineering
248,774 CPU hours

Chemistry & Biochemistry
247,597 CPU hours

Chemical & Biomolecular Engineering
237,955 CPU hours