Introduction to High Performance Computing¶
Basic High Performance Computing (HPC) System Architecture¶
As you prepare to use TACC systems for this institute, it is important to understand the basic architecture. Think of an HPC resource as a very large and complicated lab instrument. Users need to learn how to:
- Interface with it / push the right buttons (Linux)
- Load samples (data)
- Run experiments (jobs)
- Interpret the results (data analysis / vis)
Login vs. Compute Nodes
As we’ve discussed, an HPC system has login nodes and compute nodes. We cannot run applications on the login nodes because they require too many resources and will interrupt the work of others. Instead, we must submit a job to a queue to run on compute nodes.
Tips for Success¶
Read the documentation.
- Learn node schematics, limitations, file systems, rules
- Learn about the scheduler, queues, policies
- Determine the right resource for the job