UMBC HPCF

Overview

As of April 2025, the UMBC HPCF is composed of a single cluster, named chip.

Anyone may [request an account] on the cluster to access cluster resources.

First-time users should familiarize themselves with the [default user environment].

All of the machines (or computers) composing the cluster are called “cluster nodes”. The “compute nodes” are those intended for computationally intense tasks. In contrast, the “login nodes” (a.k.a., edge nodes or user nodes) are not meant for intense processing and are only intended for file editing, navigating the [file system and storage], and kicking-off slurm jobs (see below). Attempting to run load-generating tasks on the login node can cause a noticeable slow down for other users. These tasks will be killed by system administrators. Repeated offenses may result in a [lock of your cluster account]. Instead, computationally intense tasks should be run via [slurm, the cluster workload manager].

Many scientific software packages are pre-compiled and available on the cluster via the [Lmod module system].

Users may also opt to build their own packages on the cluster with tools like:

• GNU Compiler Suite

• Intel Compiler Toolchain

• Python Virtual Environments

A variety of [sample slurm scripts] and workflows exist on the cluster to help new users get started.