Unity Services

Unity is an over 20,000 core cluster based on Ubuntu 20.04 LTS and Slurm with a heterogeneous network containing ethernet and IB linked compute nodes. Unity hosts a variety of compute node architectures. The majority of compute nodes are Intel or AMD x86-64 nodes, with a small number of ARM and Power9 nodes to supplement. Additionally, Unity contains ~1500 Nvidia GPUs, including over 100 A100 and V100 GPUs. Visit the node list for a complete description of Unity’s compute nodes.

Unity contains 1.5 PB high performance VAST, for home, work, and scratch directories, and 2 PB storage from the New England Storage Exchange (NESE), a regionally managed Ceph cluster located at the same data center as Unity, the Massachusetts Green High Performance Computing Center (MGHPCC). We have a variety of storage options that suit different performance and size needs.

The Unity team provides access to research computing facilitators to assist Unity users with deploying their workloads onto Unity and advising on workflow optimization, software tuning, and technical issues. In addition, the facilitation team provides HPC workshops and manages an active Slack for the Unity Community. For groups needing mid to long term project support, facilitator time is available on an hourly basis. For help, contact us.

Research groups with compute or storage needs beyond the basic Unity access provided by their institution can contact their institution’s point of contact, our help desk at hpc@umass.edu, or, for institution-level inquiries, Tom Bernardin. Hardware purchases must be approved by the Unity team prior to purchase.

Unity resources and purchases can be written into grants. Contact hpc@umass.edu to discuss your grant requirements and get the Unity NSF boilerplate document.