Goals and Mission
1. Increase in the use of HPC across the campus.
All of the engineering and science disciplines use computation of some form for research and education purposes. Many of these disciplines primarily use desktop or laptop computers and would benefit from moving to use of HPC technologies. Obstacles to doing so include lack of easy access to and unfamiliarity with high performance computers, especially GPU-accelerated systems. The proposed augmented HPC cluster and the accompanying documentation, tutorials, and user support will help to overcome these obstacles.
2. Improvement of student knowledge of and skills in using state-of-the-art HPC technologies.
PI Moore teaches a course on Parallel Computing that is taken by upper-level computer science (CS) undergraduate students as well as graduate students in CS and Computational Science, with the most recent enrollment being more than 60 students. While she usually applies for and receives an ACCESS (previously XSEDE) education allocation with access to GPUs for this course, the number of node hours available per student is limited and can only be used for class assignments. Having a
campus GPU cluster available would allow the students to apply what they learn in the course to independent projects and to research for master’s theses and PhD dissertations.
3. Increased rate of producing research results and artifacts.
A number of UTEP researchers in different engineering and science departments currently use laptop or desktop computers, their own servers, or the current UTEP HPC cluster to run simulation and machine learning applications. Their programs can take hours or days to run and produce results using these resources. Because of the often steep learning curve of adopting new HPC and AI hardware and software technologies, or lack of access to these technologies, they may continue to use
outdated hardware and software in the interest of producing results in a timely manner. This project aims to both provide easy access to local state-of-the-art hardware and software as well as provide documentation, tutorials, and user support that will help flatten the learning curve for adoption.
