Over the past decade, SMU has pushed forward in high-performance computing (HPC), first with ManeFrame, then ManeFrame II, the NVIDIA DGX SuperPOD Advantage, and now M3. With M3 coming online this month, with its state-of-the-art CPUs, networking technologies, high memory capacity per node, and advanced interactive experiences, it has doubled the number of CPU Cores in the Data Center available for optimized high-level programming environments such as MATLAB, Python, and R. Another improvement is the unified Work storage system which allows researchers to more seamlessly transfer data between the SuperPod and M3!
M3 Configuration
| Resource | Standard-Memory | High-Memory |
|---|---|---|
| Nodes | 170 | 8 |
| Processors | AMD EPYC 7763 | AMD EPYC 7763 |
| Frequency | 2.45 GHz | 2.45 GHz |
| CPUs/Node | 2 | 2 |
| Cores/Node | 128 | 128 |
| Memory/Node | 512 GB | 2 TB |
| Local Scratch/Node | None | 4.3 TB |
| Interconnect | 200 Gb/s | 200 Gb/s |
Over the coming months, the Office of Information Technology’s Research and Data Science Services team will be providing direct and dedicated support to researchers wishing to engage with M3. The first step will be the migration of accounts, access, and directories from M2 to M3.
Currently, our Infrastructure team is hard at work on installing and making software available on M3. There are a couple of changes from the previous environment. For users using Python and R, it is recommended they set up their own environments with any specific packages needed. Also, users will need to recompile performance-critical applications, as M3 uses AMD CPUs, while M2 uses Intel. Of course, if you run into issues, as always, please feel free to ask for assistance.
Just like M2, M3 is available to faculty, staff, and students of SMU. If you haven’t used a system like M3 before, new training will be offered soon to learn how to take advantage of M3’s high-performance computing in your studies or academics.
This is an exciting time for campus. With new solutions like M3 and the SuperPOD, data sciences, digital humanities, and many more areas of study are going to have easy access to R1-level high-performance computing. Just another way we are furthering SMU’s mission of supporting world-changing research and innovation for Dallas, North Texas, and the World.
