Glossary¶

Common terms in UCT eResearch¶

This glossary defines key terms used across research computing, data, and software workflows at UCT.

HPC and compute¶

High Performance Computing (HPC)¶

A shared computing environment designed for running large-scale or resource-intensive analyses across multiple compute nodes.

Cluster¶

A group of interconnected computers (nodes) that work together to perform computational tasks.

The entry point to the HPC system where users connect, prepare jobs, and manage files. Not intended for heavy computation.

Compute node (worker node)¶

A machine within the cluster where jobs are executed.

Scheduler¶

A system that manages job queues and allocates compute resources. At UCT, this is based on SLURM.

Job¶

A unit of work submitted to the scheduler for execution on the cluster.

Job script¶

A file that defines how a job should run, including required resources and commands.

Partition¶

A subset of compute resources within the cluster, often grouped by capability (e.g. CPU vs GPU).

GPU (Graphics Processing Unit)¶

A specialised processor used for parallel workloads such as machine learning and simulation.

Storage and data¶

/home¶

A user’s personal storage space on HPC for configuration files, code, and small datasets.

/scratch¶

High-performance temporary storage used during active computations. Not intended for long-term storage.

Quota¶

A limit on the amount of storage a user or project can use.

Research Data Storage (RDS)¶

A managed storage service for longer-term research data that needs to be retained and organised.

Access and transfer¶

SSH (Secure Shell)¶

A protocol used to securely connect to remote systems such as HPC.

SCP (Secure Copy)¶

A method for copying files between local and remote systems over SSH.

Rsync¶

A tool for efficiently transferring and synchronising files between systems.

Globus¶

A platform for reliable, high-performance data transfer between storage systems.

VPN (Virtual Private Network)¶

A secure connection required for accessing certain UCT services from off-campus.

Software and environments¶

Module¶

A system used on HPC to load and manage software environments.

Module system¶

A tool that allows users to dynamically load different software versions (e.g. using module avail and module load).

Environment¶

A configured set of software and dependencies used to run a workflow.

General research computing¶

Workflow¶

A sequence of steps combining code, data, and compute resources to produce research outputs.

Reproducibility¶

The ability to run the same workflow again and obtain consistent results.

Intermediate data¶

Temporary data generated during processing that may not need to be retained long-term.

Notes¶

Terms may appear in multiple sections of the documentation.
Where needed, pages link back to more detailed explanations in Services or How-to guides.