Overview:

• The goal of this summary is to allow users to start using MSI quickly.
• This introduction is non-extensive. Consult the MSI website for details and other uses.
• By the end of this document you should be able to connect to MSI interactively and run GPU-demanding code on it.
• One will need to be connected to the University of Minnesota’s network directly or through a Virtual Private Network (VPN).

Demo

• Steps that the video demonstrates:

edited-msi-demo.mp4

• mnist_classification.py

MSI’s High-performance Computing

“MSI's High Performance Computing (HPC) systems are designed with high speed networks, high performance storage, GPUs, and large amounts of memory in order to support some of the most compute and memory intensive programs developed today.

MSI currently has three main HPC systems: Agate, Mesabi, and Mangi” reference

To get access, first your instructor or a “Principal Investigators (PIs)” has to added you to an MSI user group.

Interacting with MSI's High Performance Computing

Since we would like to develop code on MSI we care about interactive ways of accessing resources. MSI offers multiple ways to connect interactively to its resources.

**Open OnDemand**

Jupyter Notebooks

Interactive srun query

Understanding resources available

As we have said there are multiple clusters and each might have its own pros. To decide which cluster to use, check out the services/clusters/<cluster-name>(example) tabs on the MSI website. A good reference is also the slurm partitions page where you could see all available nodes and their capabilities.