What Would You Do with a Million Cores of Compute Capacity?
Clouds and Distributed Computing
TimeTuesday, November 13th2:30pm - 3pm
DescriptionHigh Performance Computing has always been about solving the most complex problems, but for far too long HPC applications and workloads have been constrained by infrastructure capacity to allow for an unfettered imagination of problems could be solved if infrastructure capacity were virtually unlimited. In this session, the speaker will highlight how virtually unlimited capacity and scale, accessed instantly on the cloud, can create a paradigm shift in the way researchers and engineers approach innovation. AWS has helped set up up massive HPC clusters that scale to over a million vCPUs, multiple times. In this session, we will highlight the process of setting up extreme-scale HPC clusters in the cloud and the lessons we learned along the way. We will lay out a broad vision of how the steadily increasing interest in running HPC workloads on the cloud, combined with the advances in AI/ML is a catalyst for sustained innovation.