High-Performance Computing (HPC)

Using supercomputers or clusters of many processors, often with GPUs, to solve problems too large or too slow for a single machine, such as simulation, weather modelling and AI training.

HPC systems use fast interconnects and parallel file systems for tightly coupled jobs. Cloud HPC offers similar hardware on demand, which suits bursty or experimental work, while on-premises systems suit steady heavy use.

MPI passes messages between processes on different nodes, while OpenMP parallelises code across cores on one node. Large applications often combine both with GPU programming.

Users submit jobs to a workload manager such as Slurm, which allocates nodes and queues work fairly across users.

The Dell HPC link now redirects to a community event page, so it was removed. InsideHPC could not be opened in the browser pane but was kept from the earlier version, and the NetApp page blocks scripts but loaded in a browser.