Faculty HPC cluster opens for users
On May 31 the faculty computing cluster Chimera will officially open for access to all members of the faculty. This will be followed on June 3 by a faculty-wide introductory training.
The cluster can be used for all types of HPC applications ranging from single-core to multi-core (parallel) and GPU applications including processing of big data. The cluster currently comprises three virtually connected clusters from Troja (Chimera) and Malá Strana (gpulab and parlab). We are working on connecting more clusters located in Troja and Karlov.
The long-term goal of our HPC community is to build a robust computing platform for our faculty, suitable for both teaching and research, including user support in the form of training and Research Software Engineering for code development. Opening of the faculty cluster is the first step towards this goal.
The introductory training session will be aimed at researchers, including students working on their theses, and lecturers wishing to learn how to utilize the cluster for teaching. We aim to run regular training sessions for new users once every year in the winter semester around November/December.
You can sign up for the introductory training (June 3) here.
The central cluster Chimera is equipped with specialized high speed data storage servers and network infrastructure offering, initially, 40 TB of HOME, 100 TB WORK storage and HDR InfiniBand interconnect on a subset of machines.
Cluster Chimera in Troja has a total of 26 servers but for this year we are planning its significant extension towards a more homogeneous cluster which will improve its readiness for massively parallel applications.
We strive to achieve a maximum possible physical (where possible) and virtual integration of the existing HPC resources of the faculty under Chimera. In the future, we envision their merging and relocation into a new data center on the Troja campus for which we’re seeking funding.
Establishment of the faculty cluster would not have been possible without the support and generous contribution of the faculty (doc. Mirko Rokyta and prof. Ladislav Skrbek), departments and efforts of many of our colleagues, past and present, who have paved the way towards it by decades of HPC and IT developments at our faculty. A special thanks goes to: dr. Jakub Yaghob, Mgr. Pavel Kaňkovský, doc. Ladislav Šubr, prof. Jaroslav Burda, dr. Ludvík Urban, Mgr. Štěpán Balcar, dr. Michal Hájek, dr. Oldřich Ulrych, dr. Ondřej Maršálek and many others for their active involvement, support, useful advice and ideas.
