Publications
2021 Teaching and Learning HPC through Serious Games
Journal Article
- BibTex Key
- Authors Dennis Milechin | Julia Mullen | Lauren Milechin
- Tags
- DOI Number j.jpdc.2021.07.014
- Book Title Journal of Parallel and Distributed Computing
- Issue Title Keeping up with Technology: Teaching Parallel, Distributed and High-Performance Computing
- Publisher Elsevier
2022 Adapting Collaborative Hands-On Activities for Remote Learning
Conference
After two successful years of in-person workshops designed to educate users about supercomputing basics, techniques for scaling code from the desktop to the supercomputer and best practices for using a supercomputer, the Covid19 pandemic forced the redesign of a three day, in-person workshop into a virtual experience. These workshops are held annually in January, providing ample time to convert the 2021 workshop into an entirely virtual learning experience. The expectation for 2022 was a return to in-person workshops but the rise of the Omicron variant necessitated a rapid shift from the in-person design to a fully virtual course. While many of the challenges associated with the conversion were shared between the two course designs, there were some key differences that lead to a broader set of lessons learned and best practices.
The first time the course was offered it was a two-day workshop. The workshop included a number of shared, hands-on, activities designed to promote an intuitive understanding of High Performance Computing (HPC) concepts. These activities included educational games created by the instructors to abstract the HPC concepts from the implementation details in order to create deeper understanding. The games used physical components to provide a tangible connection to the team and activity. For the second year of the course, minor modifications were made to the format and activities. The format was altered to add an additional half day session to discuss student projects, while the games were streamlined to improve game flow and more clearly highlight the key concepts.
In late 2020, the Covid19 pandemic forced a complete refactoring and the January 2021 course was offered in a blended mode with asynchronous online content and live virtual sessions. Using best practices, the lectures were converted to targeted, concept specific videos with knowledge checks interleaved between them to provide immediate feedback and reinforce learning. The schedule was adjusted to avoid Zoom fatigue while retaining time for hands-on activities and student presentations that described their projects and proposed scaling method. Modifying these course components was fairly straight-forward compared to the changes required to successfully incorporate the games used in previous years. In one case the game was not suitable for use in the virtual environment and a substitute was used. The second game was adapted to the zoom environment. In 2022, the Omicron variant required a rapid shift from a planned in-person course to a fully remote course. Drawing from the lessons learned from the 2021 course, the games and course flow were further modified to create a more fluid remote learning experience.
This paper explores the process of converting a primarily informal workshop course from face-to-face format to a hybrid learning experience with a focus on the modifications required to create successful hands-on activities for remote learning. The paper concludes with a discussion of the lessons that were learned and the best practices that were drawn from those lessons.
- BibTex Key Julia Mullen,Lauren Milechin2022
- Authors Julia Mullen | Lauren Milechin
- Tags active learning | Game Based Learning | Gamification | High Performance Computing | Informal learning | Remote Learning | Supercomputing
- ISBN/ISSN 978-84-09-42484-9
- DOI Number 10.21125/edulearn.2022.1644
- Issue Title EDULEARN22 Proceedings
2022 Building Experience and Confidence in HPC Practitioners through the Project-Based, Hands-On Practical HPC Course
Conference
The MIT SuperCloud and Lincoln Laboratory Supercomputing Center have been introducing High Performance Computing (HPC) to a new audience through the ”Practical High Performance Computing: Scaling Beyond your Laptop” class for the past four years. This informal class, open to the entire MIT community, introduces HPC, identifies canonical HPC workflows, and provides hands-on activities to explore the challenges encountered in the HPC environment. The students use their own research applications as project work to apply the class concepts to gain experience and confidence in using an HPC system and throughout the scaling process. Survey data collected before and after each class demonstrate that students feel they gain familiarity and experience in the concepts taught in the course and confidence in their own ability to apply those concepts.
- BibTex Key
- Authors Chris Hill | Julia Mullen | Lauren Milechin
- Tags active learning | flipped classroom | hands-on learning | HPC training and education
- DOI Number 3491418.3535140
- Book Title Practice and Experience in Advanced Research Computing
- Publisher ACM
2022 Graphulo: Linear Algebra Graph Kernels
Book Chapter
The Big Data and the Internet of Things era continue to challenge computational systems. These challenges, often referred to as the 3V’s of big data, are compounded by heterogeneous data structures and sources. While several technologies, such as NoSQL and NewSQL databases, have been developed to address some of these challenges, these databases often support different underlying data representations and are largely designed to perform the same sets of operations. In order to generate meaningful results from large datasets, analysts often use a graph representation which provides an intuitive way to work with the data. In some cases, graph vertices can represent users and events, and edges can represent the relationship between these users and events. Graph algorithms are used to extract meaningful information from these very large graphs. The MIT Graphulo initiative (http://graphulo.mit.edu/) is an effort to perform graph algorithms directly inNoSQL databases such as Apache Accumulo or SciDB, which have an inherently sparse data storage scheme. Sparse matrix operations have a history of efficient implementations, and theGraph Basic Linear Algebra Subprogram (GraphBLAS) community has developed a set of key kernels that can be used to develop efficient linear algebra operations. However, in order to use the GraphBLAS kernels, it is important that common graph algorithms be recast using the linear algebra building blocks. In this chapter, we provide a brief overview of different classes of graph algorithms, recast some of them into linear algebra operations compatible with GraphBLAS building blocks.
- BibTex Key Lauren Milechin,Shana Hutchison,Hayden Jananthan,Jeremy Kepner,Benjamin A. Miller,Andrew Prout,Siddharth Samsi,Chuck Yee,Vijay Gadepally2022
- Authors Andrew Prout | Benjamin A. Miller | Chuck Yee | Hayden Jananthan | Jeremy Kepner | Lauren Milechin | Shana Hutchison | Siddharth Samsi | Vijay Gadepally
- Tags
- ISBN/ISSN 9781003033707
- Book Title Massive Graph Analytics
- Publisher Taylor & Francis Group
2023 A Data Driven Approach to Informal HPC Training Evaluation
Conference
The High Performance Computing (HPC) community has a long history of educating researchers, students, and practitioners to use HPC systems effectively and efficiently, primarily through informal, non-graded workshops and short courses. While nearly all workshop evaluations capture user satisfaction and perform preand post-workshop evaluations, to our knowledge none of them capture changes in user behavior as a means of evaluating learning. To more fully evaluate a student’s skill development and capture improvements in the use of HPC systems, we developed a data centric method for evaluating informal HPC training. The evaluation methodology is described in four steps: design clear learning objectives, map the learning objectives and content to the data collection instruments and queries of interest, collect the data, and explore and analyze the data. This is a process easily adaptable by any HPC center because we focus on data that are available to all HPC trainers, e.g. workshop surveys, questions, and job performance data from the HPC system, primarily from an HPC scheduler.
- BibTex Key Julia Mullen,Lauren Milechin2023
- Authors Julia Mullen | Lauren Milechin
- Tags data driven evaluation | HPC training and education | learning analytics | training evaluation
- DOI Number 3569951.3597551
- Book Title Practice and Experience in Advanced Research Computing (PEARC '23)
- Publisher ACM