High Performance Computing Clusters

High Performance Computing Clusters

For computing tasks that require computing power beyond what is currently possible on workstations and servers, you have to move to High Performance Computing (HPC) solutions.

A modern supercomputer is generally built as a HPC cluster. A HPC cluster combines the processing power of many server class computers (compute nodes) to speed up the solution. To benefit from a HPC cluster a program must divide the problem into smaller parts and distributed these to multiple computing nodes.

Moving from a workstation or server to a HPC cluster will always require changes to the workflow and could involve major changes or even complete redesigns of the software stack you use.

If you know that a project is going to need supercomputing you should consider this when choosing the software you use. All supercomputers are Linux based and most of the software for supercomputing is jointly developed by the scientific community as open source. Since development is mostly done on personal computers you can use the same software directly on your laptop from the very start of a project. It will make the transition to a supercomputer much smoother if you have been using the same software right from the start.