High performance computing


Learning to use the High Performance Computing Cluster

Gepubliceerd op
26 juni 2017

The HPC is designed as a highly available and versatile platform. Whether your job needs lots of memory, raw computing power or enormous storage demands, the HPC can handle your requests. Below you will find more on the HPC, the use of this platform, the courses and the experiences of two participants of the basic course.

The High Performance Computing (HPC) cluster is a supercomputer on which large calculation tasks (“jobs”) can be performed, e.g. for bioinformatics and model creation. Through the joint use of the HPC by groups within (and even outside of) WUR, knowledge, methods, and software can be shared more easily. The overall management of this will be provided by Facilities and Services - IT. This will enable researchers to give more focus to their research by not also having to manage a small cluster within their own group. In order to make it easier for groups to connect, it will be possible this year for groups to take out a “subscription”. This allows the HPC to be used for a fixed price.

High Performance Computing Cluster basic course on 7 June

The High Performance Computing Cluster (HPC) basic course on 7 June was attended by 40 researchers interested in using or already using the HPC. After an introduction to the HPC, the researchers received tips & tricks on starting and running computing jobs to speed up working with the HPC. The course was organised by FB-IT with contributions from WUR researchers who were experienced in using the HPC. Read the experiences of two participants of this basic course.

Esther Ellen is a researcher in Animal Breeding and Genetics and wants to begin using the HPC to find areas of DNA that can explain genetic differences:
“I currently use Linux for my analyses, but that takes a long time. My large and complex datasets demand a lot of memory. The HPC works faster and you can run more jobs in tandem, which makes an enormous difference in calculation time. The course was incredibly useful for me, because I received tools with which I could get started. You often do not take the time necessary to delve deeply into such a new method on your own. This course means that you can block off one afternoon in your calendar to actually work with it. The organisers provided a number of assignments and examples, most of which were quite doable. There was also plenty of help available if you got stuck. I was familiar with the terms being used, but I had never worked with the HPC before. For people who already work with the HPC, the course for advanced learners in the autumn is probably more suitable.”

Manon de Visser has worked with the HPC cluster at Animal Breeding & Genetics Group for nearly nine months on her MSc thesis project. The beginning of the course was already familiar to her:
“I had already ‘submitted jobs’ before, but that is definitely crucial if you want to work with the HPC, so it is an important component of the course. For me, it was particularly useful to learn how exactly you execute multiple jobs in tandem, regardless of whether they interact with each other or not. Without a course like this, you would have to figure it all out for yourself and that takes time. If, in that process, if you use the HPC incorrectly, you can also disrupt other users. So, that makes an HPC course very important. The examples and the results of the analyses give you an idea of what you are doing and what you can use the output for. In this way, you can really understand the tasks (“jobs”) and see their use. I am not currently using the HPC, but hopefully, I can apply the knowledge in practice again in the future.”