The University of Tulsa’s Tandy School of Computer Science hosted the Community Building for High-Performance Computing Curriculum Development workshop June 10-12, 2019, in Keplinger Hall. The workshop brought together 19 faculty members and professionals in high-performance computing (HPC) and data science from 10 states to discuss how to improve training and build a stronger community in the field.

Also known as supercomputing, high-performance computing is applied to many disciplines such as physics, mechanical engineering, electrical engineering, mathematics and biology. Students with high-performance skills and the ability to operate supercomputers solve multidisciplinary topics and are instrumental engineers skilled in computer programming, hardware and web domains. Examples of high-performance computing include modeling quantum mechanics or searching large medical data sets to determine which protein structures are effective in fighting disease.

Careers in high-performance computing and computational science recently have been ranked as some of the most satisfying in the United States in terms of job satisfaction, compensation/salary and opportunities. Graduates are expected to fuel the next industrial revolution, which will focus on big data, artificial intelligence and machine learning.

The workshop investigated the following three areas:

• How to best provide high-performance computer training for undergraduates in a 4-year degree program.
• How to increase interest in high-performance computing at the undergraduate level.
• How to build and expand the high-performance computing community.

Experts in the high-performance computing field

Career pathways in the field were presented by Henry Neeman, assistant vice president of information technology for the University of Oklahoma, and George Louthan, associate director for research computing strategy at the University of Oklahoma Supercomputer Center for Education and Research. A career in high-performance computing is deeply rewarding and intellectually challenging with professionals learning about a wide variety of subjects through their work with researchers. Strong soft skills are a necessity in this field as one must interact with a wide range of customers from users and researchers to vendors and technical support professionals. Neeman also discussed his experiences on creating and running the annual Oklahoma Supercomputing Symposium, which will be held for the 18th year on Sept. 24 and 25, 2019 in Norman, Oklahoma, on the OU campus. Stephen Harrell, senior research analyst for information technology at Purdue University described the university’s HPC apprenticeship program and its student cluster competition. These programs support Purdue’s efforts to generate home-grown talent for its HPC center and help graduates find jobs in the HPC field, specifically HPC system administration.

Dirk Colby, director of HPC Studies for Michigan State University, discussed the high-performance initiatives at Michigan State University and his experiences teaching HPC using a flipped classroom, a method that involves introducing new material outside of the classroom. Aaron Weeden and Skylar Thompson described efforts by Shodor, a national resource for computational science education, to develop computing and HPC learning modules and their Bootable Cluster CD platform for teaching HPC concepts. Kate Cahill from the Ohio Supercomputer Center spoke about the education and outreach activities facilities by the Extreme Science and Engineering Discovery Environment (XSEDE) program and the HPC knowledge and curriculum repository housed at HPC University, a virtual organization that shares educational and training materials.

Jeremy Evert of Southwestern Oklahoma State University described his experiences teaching HPC, competitive learning and student placement in HPC internships at NASA and the National Weather Service. Evert also described the positive experiences his students gained through the prestigious Blue Waters Student Internship Program.

Encouraging training and participation

The workshop highlighted a variety of training pathways, including TU’s high-performance computing minor, and HPC University’s efforts to collect high-performance computing information. Associate Professor Peter J. Hawrylak presented information on TU’s high-performance computing minor, which provides students in the College of Engineering and Natural Sciences with a roadmap to obtain a solid foundation in high-performance computing during their four-year bachelor’s degree. The annual Oklahoma High-Performance Computing Competition, hosted by TU each spring, also was suggested as an opportunity for competitive learning in the HPC field.

The workshop covered examples of curriculum development and outreach efforts geared toward offering interested students with opportunities in HPC, computational science and data science fields. The venue supported lengthy and in-depth discussions of the topics, resulting in concrete ideas and plans to expand efforts and build collaboration to strengthen talent generation pathways in academia. All workshop attendees left the workshop with a list of ideas on how to improve HPC education and outreach at their home institutions.

The workshop was made possible with support from TU, the National Science Foundation, the Extreme Science and Engineering Discovery Environment, Shodor and the National Computational Science Institute.