Strelka: The Year in Review

As 2023 comes to a close, it is worth reflecting on ways in which Strelka remains the focus of Swarthmore’s research computing infrastructure. Our flagship high-performance computing cluster has undergone substantial upgrades to its capabilities, and has also been used in a greater diversity of workloads than ever before. Here’s a look at some of the most important achievements from the past year.

As of the end of November, users had logged nearly 3.3 million CPU-hours on Strelka. That’s almost 377 years! Put another way, Strelka runs about a year’s worth of CPU-hours every single day.

The user base has grown as well, now encompassing entire courses, as well as individual students, in addition to the expected researcher and lab groups. Below is an image showing the log of CPU-hours by user over the past year (as a refresher, you can think of the Y-axis as the number of zeros, so 100 hours is 2, 10,000 hours is 4, and 1,000,000 hours is 6). Twenty-seven users logged more than 100 CPU-hours, and ten logged 10,000 or more CPU-hours!

Throughout the year, new hardware was added to increase Strelka’s computational and storage capacities, enabling it to support more concurrent users and jobs, jobs of greater complexity, as well as to store and manage larger datasets. In all, Swarthmore’s infrastructure is among the most robust for an institution of our size – and even eclipses that of several much-larger institutions!

2023 was an important year for Srelka, and we’re just getting started. 2024 promises several exciting changes to our research computing ecosystem, and the high-performance computing cluster will remain the proverbial jewel in the crown. Stay tuned for more information on these projects in the new year!