Skip to main content

High-Performance Computing (HPC)

server portsIn service of The Rockefeller University’s goal of “science for the benefit of humanity”, the High Performance Computing Resource Center (HPCRC) (RRID:SCR_025889) provides our scientists with access to infrastructure capable of running computationally demanding workflows. These workflows would be prohibitively slow, or even impossible, to execute on an individual investigator’s personal computer or laboratory workstations. The parallelization possible with an enterprise-level compute cluster allows our faculty, students, and research staff to rapidly scale their scientific work to match current high-throughput methods. Big data necessitates big storage, and the HPCRC supports its constituency with over 10 petabytes of storage that are networked to thousands of CPU processors and over 100 GPUs using a high-speed, low latency InfiniBand fabric to optimize analyses.

The HPC cluster catalyzes researchers’ simulation and analysis work by providing a broad constellation of pre-installed open-source and commercial software tools spanning a variety of scientific domains. This is accomplished using conda environments, containers, and a module system available to all users, which makes it easy for users to load in the pre-compiled binaries as well as be aware of the versions in their software stack, facilitating reproducibility and unburdening scientists from the details of managing many software and pipeline installations with complex dependencies.

The HPCRC also provides support to university researchers via a variety of training opportunities. Users can use our documentation to guide their usage of the cluster in a way that aligns with best practices. We regularly maintain and update this series of guides to reflect feedback from our user community. We also offer yearly instructor-led courses for more structured learning, as well as one-on-one consulting to implement and optimize specific workloads. We are regularly acknowledged by our constituency for these collaborative software and computer engineering efforts.

For further information, including how to contact us, please visit our website, which is only accessible from campus internet or VPN access.