IndySCC

The IndySCC is a virtual companion competition where teams use cloud resources to compete and shares many of the same goals as the SCC. Unlike the SCC, teams are not required to partner with a vendor, build their own cluster, and bring that hardware to the SC conference. If selected to participate in the IndySCC,  Teams will have the opportunity to attend SC in person and participate in the broader conference. Conference attendance is not required to participate in the IndySCC.

Teams applying to the SCC may indicate they would like to be considered for IndySCC if they are not selected to the SCC. Teams who do not indicate interest in the IndySCC will not be considered if not selected for the SCC. Indicating you would like to be considered for the IndySCC is not a guarantee to be selected for the competition.

Teams may also apply directly to the IndySCC, without being considered for the SCC. This serves as a lower bar for entry for teams that may not have existing strong vendor relationships or sufficient funding to travel to the conference or who are looking to gain a footing in the cluster competition world before applying to the SCC. The goal for teams participating in the IndySCC is they will be able to travel to the conference and compete in the SCC in a later year.

The IndySCC is intended for less experienced teams, and final selections will be made considering the strength of the application, motivations as they relate to the goals of IndySCC, and the team’s level of experience with prior cluster competitions.

Selected teams are invited to participate in a remote competition from late summer to November:

• Phase 1 (August/September): Teams develop familiarity with the provided resources. They learn how to run real-world applications, the science behind them, and how to optimize them for performance.
• Phase 2 (October/November): Teams learn how to run and optimize industry-standard benchmarks on the provided resources.
• Phase 3 (November 17–19): Teams compete in a final contest similar to and held concurrently with the SCC.

Students, with the guidance of their advisor, will craft a proposal that describes their team, their approach to the competition, and what they hope to get out of the experience. This proposal is submitted as a team application for review by the SCC committee.

Further guidance will be provided at a later date.

All teams will be provided remote or cloud-based HPC resources and guidance on accessing the resources. Continuous support is provided by HPC experts in the months leading up to the competition. Webinars are recorded for the virtual platform, the benchmarks, and the two applications. At the end of each phase, teams will turn in a report and receive feedback from the experts.

Participants are encouraged to source travel support from their local institution and supporters to attend the conference in person. Reduced rate in-person and Digital Experience registration is available for IndySCC student participants.

Benchmarks

Teams will run an industry-standard benchmark on their provided resources. After an initial practice period, each team will complete a 24-hour “hero run”, running the benchmark application across a larger set of resources and competing for the best score.

Applications

Exascale Climate Emulator

Conventional climate models solve PDEs representing the interactions in Earth’s systems, a task whose computational and storage requirements increase dramatically with increasing resolution. Climate emulators alleviate those requirements to predict the results of a more-complex conventional climate model. 

One such emulator, trained on 318 billion hourly and 31 billion daily data points of surface temperature, generates hourly climate emulations that are statistically consistent with simulations, and was awarded the 2024 Gordon Bell Prize for Climate Modelling. The emulator utilizes the spherical harmonic transform to statistically model spatio-temporal variations in climate data and the PaRSEC runtime system to support efficient parallel matrix operations on exascale systems. SCC teams will demonstrate some aspects of the climate emulator on their own clusters.

The Gordon Bell paper is at https://dl.acm.org/doi/pdf/10.1109/SC41406.2024.00008. This application is supported by Qinglei Cao, Sameh Abdulah, and Zubair Khalid.

Structural Simulation Toolkit (SST)

The Structural Simulation Toolkit (SST) is a parallel discrete-event simulator used to model large-scale and high-performance computing platforms. It was developed to explore innovations in highly concurrent systems where the ISA, microarchitecture, and memory system interact with the programming model and communications system. SST enables this through a modular design, allowing for interoperability between various simulation components. Scalable simulation of large systems is supported through both MPI and multithreading. 

While many high-profile applications are suited to GPU acceleration, many important applications — including SST — are not, and SCC teams should take this into consideration while designing their clusters.

The SST web page is at https://sst-simulator.org/. SST and SCC25 are supported by Patrick Lavin, Shannon Kinkead, Clay Hughes, Gwen Voskuilen, and Scott Hemmert.