Expansion of High Performance Compute facility
The HPC team noticed a steady increase in the demand for GPUs in the High Performance Compute facility, which can be used for GPU computing or visualization tasks. To cater to this need, we have recently acquired two additional nodes that both carry two NVIDIA Tesla P100 GPUs. If you are interested in using these nodes, for e.g. visualization tasks, deep learning or image processing, please contact our HPC team, and they will be happy to help.
Fair share policy
With a total compute capacity of 1500 cores and 600TB of HPC storage, the facility is one of the bigger academic compute facilities in the Netherlands. Usage of the HPC is based on a ‘fair share’ basis. Recently, the HPC team has made an effort to quantify what a ‘fair share’ approximately means. Factoring in that users also need non-compute hardware, such as submit and transfer hosts, networking and storage, and some level of support, we have estimated that 1 HPC share would allow a user to use 50.000 CPU hours per year. It is good to know that this is 35% cheaper than Google’s cloud compute service, which does not include support or storage for this price. Moreover, we would like to stress that this compute quota is not strictly enforced, but rather serves as a guideline to ensure fair usage of the HPC facility.
HPC developments, requests and suggestions
In the coming months, the HPC team has tasked itself with extension of our portfolio. Examples include expanding our storage capacity so that we can also offer cheap archive storage to our users. Moreover, we are working towards a cloud analytics platform that provides users with a wide range of tools and easy access to their data in a fair and reproducible fashion. Should you have any ideas, requests or suggestions for new features or future developments, feel free to contact us.
The Utrecht Bioinformatics Center HPC facility provides ample compute resources to accelerate Life Science community in Utrecht. About 35 research groups affiliated with the four UBC partners are currently using the HPC facility. Over the past year we have seen a steady increase in our usage, yet were able to efficiently deal with the peak load usage of the cluster. All users of the compute cluster can inspect the usage of the cluster and investigate their own usage in great detail.
Joep de Ligt, Coordinator UMCU Bioinformatics Expertise Core (UBEC) at UMC Utrecht: technical coordination
Jeroen de Ridder, group leader Center for Molecular Medicine at UMC Utrecht: coordinating community effort