Skip to main content

What is Discovery?

Overview​

Discovery cluster is a high-performance computing (HPC) resource for Northeastern University research community, which is physically located in Massachusetts Green High Performance Computing Center (MGHPCC) in Holyoke, MA. MGHPCC is a joint venture of multiple universities around Boston—including Boston University, Harvard, MIT, Northeastern, the UMass system, and Yale—and provides an intercollegiate high-performance computing for these universities1.

For those who aren't familiar with the HPC cluster, a cluster is a collection of many separate servers (computers) interconnected with each other through a network, which provides capability for conducting intensive and/or parallel computing. This makes researchers possible to run analysis that can't be run locally on personal devices.

Here's the official overview description of the cluster, provided by the university's Research Computing office2.

The Discovery cluster provides access to over 50,000 CPU cores and over 525 GPUs to all Northeastern faculty and students free of charge. Hardware currently available for research consists of a combination of Intel Xeon (Cascadelake, Skylake, Broadwell, Haswell, Sandybridge, and Ivybridge) and AMD (Zen, Zen2) CPU microarchitectures. Additionally, a selection of NVIDIA Pascal (P100), Volta (V100), Turing (T4), Ampere (A100), and Hopper (H100) GPUs. Discovery is connected to the university network over 10 Gbps Ethernet (GbE) for high-speed data transfer, and Discovery provides 6 PB of available storage on a high-performance file system. Compute nodes are connected with either 10 GbE or high data rate InfiniBand (200 Gbps or 100 Gbps), supporting all types and scales of computational workloads.

Official documentation​

The official documentation for Discovery is available here. https://rc-docs.northeastern.edu/en/latest/index.html

Environment​

The cluster is a Linux (RHEL 6) based system, and is consisted

  • Nodes:

Footnotes​

  1. MGHPCC. "About The MGHPCC". https://www.mghpcc.org/about-the-mghpcc/ ↩

  2. Northeastern University Research Computing. "Research Computing at Northeastern University". https://rc.northeastern.edu/ ↩