How HPC and Hyak can pay off in multiple ways for researchers

Last updated: August 11, 2022

Photo of David BakerIn a move that underscores the cost effectiveness and efficiency of Hyak, the UW’s shared high-performance computing cluster for large-scale compute and data analysis, UW Professor David Baker is transitioning nearly all of his lab’s computing to the UW-IT-managed resource. Baker directs the UW’s Institute for Protein Design (IPD), a world leader in its field.

Baker is decommissioning around 280 independently operated compute “nodes”—individual machines in a computing cluster—in favor of Hyak, confirming its growing value to researchers. The 280 compute nodes occupy eight racks and consume 80KW of power, while Hyak needs only 28 nodes in just half of one rack to deliver the same performance—consuming 10 times less power.

For researchers like Baker who already have systems, UW-IT makes it easy to migrate to Hyak. Others are finding that they don’t need to deploy and maintain separate systems in the first place.”—Chance Reschke, UW-IT Director of Research Computing