Jargon

HPC systems often use jargon to describe their systems. Here are some common terms used:

● Allocation - Tthe amount of computing resources formally granted to a researcher or research project for use on a shared supercomputing system.

● Cluster - A group of computers connected together so they function as one powerful system.

● Node - A single computer inside a cluster. Large HPC systems contain hundreds or thousands of nodes.

● Compute Node - A node dedicated to running calculations or research jobs.

● Login Node - A front-end machine where users prepare jobs, edit files, and submit work (not meant for heavy computation).

● Core - An individual processing unit inside a CPU capable of executing tasks independently.

● GPU (Graphics Processing Unit) - A specialized processor that performs many calculations simultaneously, commonly used for AI, simulations, and accelerated science.

● Parallel Computing - Running many parts of a problem simultaneously across multiple processors.

● MPI (Message Passing Interface) - A programming method allowing many nodes to communicate during large distributed computations.

● Job - A computational task submitted to the cluster.

● Job Scheduler - Software that decides when and where jobs run on the system (e.g., Slurm).

● Queue (Partition) - A waiting line for jobs grouped by resource type or priority.

● Batch Job - A job submitted to run automatically without user interaction.

● Interactive Job - A temporary session allowing direct interaction with compute resources.