Loading...
Presenters
Spangler, Eric
Publication Date
4-24-2026
Abstract
HPC applications already rely on significant parallelism running across multiple CPUs and across multiple nodes. As GPUs become increasingly central to modern supercomputing, many researchers view them as requiring entirely new programming models. This talk reframes GPUs as a natural extension of the parallel structures HPC users already employ. By examining parallel computing through a simple motivating example we show how these concepts can make direct use of GPUs. The goal is to give researchers in chemistry, material science, engineering and even finance, an intuitive framework for understanding when and why GPUs accelerate existing workloads and how to approach them effectively.
Recommended Citation
Research Computing, "Research Computing Seminar Series - Compare the performance of using GPUs vs CPUs" (2026). Research Computing Seminar Series. 8.
https://digitalcommons.memphis.edu/research-computing-seminar-series/8
