National HPC Hackathon

Each team will have its own AWS clusters with Intel processor-based instances (c5n) for comparing results. Each team can have a total of 10,000 core hours during the hack phase to build and run code successfully. However, what the team does to utilize the hardware to extract performance will make the most difference for the team. Instances will also have 100 Gbit/s networking with the Elastic Fabric Adapter (EFA) to support efficient MPI communications. Teams can also leverage very high throughput Lustre file system. 

Teams will be provided support to set up AWS clusters with development tools and optimized libraries from Intel and others for porting, running, and benchmarking the selected code(s). Teams will need to provide realistic inputs and data for test cases, and measure how a code performs multiple-node runs. Points will be awarded for every successful code-build and test case, and the number of points earned will increase for performance improvements. Spack recipe improvements will go upstream, and test-cases and workloads will go to a public GitHub repo where all the artifacts will get a DOI so you can reference them in a paper.

Teams will also be assigned a mentor, and have guidance from a community of technology specialists from within Intel and AWS to enable them to get the most out of this experience.