Third Annual Peregrine User Group Meeting

The 3rd Annual Peregrine User Group Meeting took place on January 22nd 2020. Continuing a 3 year old tradition, we had an interesting program, which started with Henk-Jan Zilverberg, team-lead HPC, talking about the highlights of the previous year regarding Peregrine and HPC. We have extended Peregrine with about 40 GPU nodes and a “vulture” partition which makes use of idle compute power at the University, and puts it to work for researchers from within Peregrine. We also have a new OpenStack cluster, Merlin, where researchers can request their own virtual machines, customize and use them as needed.

The meeting continued with three presentations by Peregrine users, giving a brief introduction to their research and how Peregrine helps.

Presentations

Jules Tinang (FEB) studies Consumption-based Capital Asset Pricing Models, and he uses Peregrine to build confidence intervals for the bootstrap method. This involves setting up hundreds of thousands of samples and aggregating statistics from these, a task that can make good use of, and indeed needs, the resources that Peregrine provides.

If you tell Ranko Gacesa that his work is s#!t, he won’t mind too much, since he does study the human gut microbiota, but only theoretically. Identifying all the bacterial species that call us home is a herculean task, but of paramount importance, given the huge influence that the gut microbiota has on human health and well-being. This is what Ranko does, by piecing together RNA sequencing data from large samples, and he uses Peregrine to get the job done. Not everything always goes smoothly, and the HPC team took notes on the problems Ranko has faced.

From the gut to the stars, Gijs Verdoes-Klein gave a fascinating overview of Euclid, a world-wide project spearheaded by ESA to search for 25% of the Universe that consists of Dark Matter. Upon its launch in 2022, the Euclid satellite will send down to Earth a large volume of data that needs to be processed, stored and analyzed. A part of this job will be carried out using Peregrine, and work has been underway for quite some time in building and testing data processing pipelines. Our colleague Bob Dröge works part-time on this project, and he’s promised to let us know first when Dark Matter is finally detected.

A look to the future

We closed the formal part of the meeting with a look to the future of Peregrine and of HPC in Groningen. Fokke Dijkstra from the HPC team is leading the project for the next iteration of our computer cluster (name pending), and throughout the summer, he together with CIT trainee Jacob Ziemke has met with several user groups to discuss their views on what that future should look like. He gave an overview of those discussions, as well as the challenges ahead, and the solutions that might be employed to meet those challenges. In the rapidly changing environment of HPC, we need to constantly be on the lookout for interesting innovation opportunities.

The meeting was followed by a borrel with interesting conversations over a couple of drinks and hapjes. We also introduced our miniature portable cluster, Bobolink, consisting of 8 Raspberry Pi 4s connected together. As with any small cute thing, it was very popular!

We’d also like to give a shout-out to our colleagues Egon Rijpkema, Harm Vos and Ger Strikwerda, who organized the second Groningen Tech eXchange meeting later that evening, and which was as well a success!

See you all next year (or later this year?)