Unlocking the Potential of Diversity

Lofstead was the SC22 Inclusivity Committee Chair and currently serves as the SC23 Reproducibility Committee Co-Chair along with many other committee roles at SC23. In this I Am HPC profile, Lofstead shares his experiences in making inclusive workspaces a reality, as well as his plans for implementing change to make SC23 a truly reproducible and inclusive environment for all participants.

Jay Lofstead

Principal Member of Technical Staff, Sandia National Laboratories

Jay on LinkedIN

Q: What single event most made you realize you wanted a career in HPC/computing?

Lofstead: I’ve been seriously involved in computing starting when I was about 10. As a middle-aged man (that was a long time ago), I found that computers were the ultimate tool that could be made to do almost anything. I loved science but didn’t enjoy math once I got into upper-level undergrad classes and focused more strongly on computer science. When I went back to grad school, I saw that I could work to support science while doing computing, entered the HPC field, and never looked back.

Q: What do you consider your biggest contribution to the HPC/computing community?

Lofstead: I do a lot of mentoring. As a first-generation college student, I had no guidance growing up about how to navigate an education. All anyone could tell me was I needed a “good” GPA [grade point average]. What “good” even meant was unknown. I was lucky and managed to make a number of good decisions that eventually helped me make it into a Ph.D. program then through it into my career at Sandia. Since joining Sandia, I have spent a tremendous amount of time and effort helping other students (any level, including grade school) understand what their career options really are, what education made that easiest, and what is necessary to enter these programs. As I’ve hired summer interns, I’ve paid 13 different students, and all 13 have generated an external to Sandia peer-reviewed publication. In most cases, I have taken the students to the conference venue and had them present to the research community. While I have had a few other students I did not pay, not all of them were able to succeed at the same level largely due to conditions I could not control. Building a more diverse and inclusive workforce will make science, computing, and the world a better place. I strive every day to make that a reality.

Q: In the past 35 years, what is the most significant overlooked breakthrough that has impacted the field in your eyes?

Lofstead: For HPC, the Raspberry Pi is used to make “toy” clusters as an educational tool. I don’t think enough people see the power that having such a tool has in shaping minds. Seeing that you can make things work on a pretty cheap machine you can build yourself is empowering. It can get more people into HPC rather than just computing in general.

Q: What would you like to see change about, within, or among the HPC/computing community?

Lofstead: We need to accelerate diversifying the community. Students today expect a diverse and inclusive environment. If we do not offer a place where they feel comfortable, they will find something else to do. Related to this is reproducibility. Incorporating strong reproducibility standards for all computing will raise the quality standard for research and make learning new things from primary sources easier. Textbooks are expensive. Through a research library, research papers are available at no cost to library visitors. Having these other resources can help people explore new areas and help drive building a more diverse computing environment. We need everyone that can do the work to join us in building a better tomorrow.

Community-Sourced

Part of what makes this the premier venue for HPC research is found in the community: in 2022, over 11,000 attendees and 361 institutions representing the largest and most advanced computing centers in the world attended SC. Another ingredient that puts the “super” in the selected papers is the rigor by which research is evaluated.

Each submitted paper is directed into one of ten technical areas: Algorithms; Applications; Architecture and Networks; Clouds and Distributed Computing; Data Analytics, Visualization and Storage; Machine Learning with HPC; Performance Measurement, Modeling and Tools; Post-Moore Computing; Programming Frameworks and System Software; and State of the Practice.

SC23 will use a streamlined, single-round review-and-rebuttal process. All submitted papers will be reviewed by four or five experts in the technical area. All reviews will be double-anonymous so that reviewers do not know the papers’ authors or affiliations. Authors will have an opportunity to provide rebuttals, answering questions from the reviewers before the technical program committee meets to select papers to be accepted. Acceptance decisions will be released in mid-June. See the Papers page for full details on paper submissions.

The SC Conference is a leader in tangible progress towards scientific rigor, through its pioneering practice of enhanced reproducibility of accepted papers. The scientific method is the foundation of scientific research, utilized by aspiring science fair contestants and Nobel laureates alike. The successful application of the scientific method has a key feature: the result, under given conditions, is reproducible. The SC Papers program promotes this end with the Reproducibility Initiative.

Scientific Rigor

Reproducibility Initiative

First, the Artifact Description and Artifact Evaluation (AD/AE) appendices enhance transparency and improve reproducibility. However, achieving reproducibility might require non-trivial effort and time investment. Therefore, to diminish its impact, it is advisable to incorporate reproducibility practices early in the experimental design phase when validating a paper. Additionally, the benefits of reproducibility extend beyond ensuring trustworthy science, as authors can also reproduce their own experiments in the future, thus enhancing their own research outcomes. All paper submitters should review the information on the Reproducibility Initiative page, including the guidelines for AD/AE Appendices & Badges. Ensuring the reproducibility of scientific results is crucial for establishing trust in scientific findings.

Reproducibility challenge

Second, the Reproducibility Challenge takes a previously accepted paper and has students in the Student Cluster Competition attempt to reproduce it. Now this is where the rubber hits the road! The students will produce publish-quality, citable reports that will be made available through an archive.

Submissions Are Open