Purdue University

Purdue University’s newest community cluster Gautschi has debuted at number 157 on the Top500 list of the world’s most powerful supercomputers, which was announced this week at SC24, the International Conference for High-Performance Computing, Networking, Storage and Analysis, taking place in Atlanta.

Gautschi also ranks number 43 on the Green500 list of the most energy-efficient supercomputers in the world.

“Thanks to Purdue’s continued investment in high performance computing, Gautschi will provide the resources needed to recruit faculty as part of Purdue Computes, and to enable new discoveries in artificial intelligence. We look forward to supporting our faculty and students in applying this system to their research and learning”, says Preston Smith, Executive Director of the Rosen Center for Advanced Computing in Purdue IT, and Director of Computing Infrastructure for the Institute for Physical AI.

The Gautschi cluster features cutting-edge technology for traditional modeling and simulation applications, while also offering NVIDIA's top-tier solutions for AI applications. Gautschi, named after Walter Gautschi, professor emeritus of computer science and mathematics, is Purdue’s most powerful cluster built so far, with a peak performance of 10.86 petaFLOPs.

The powerful Gautschi system consists of 338 Dell compute nodes, each outfitted with two 96-core AMD Epyc “Genoa” processors, along with 20 nodes equipped with NVIDIA’s cutting-edge H100 GPUs, which are known for their transformative potential in AI research. This combination of hardware enables Purdue researchers to tackle increasingly data-intensive and computationally demanding projects.

Gautschi enables the transformative nature of AI in research by providing state-of-the-art GPUs to faculty. The 160, 8-way connected H100 GPUs allow adoption of intensive AI workloads such as major Large Language Models that have become key to research in several disciplines. The H100 utilizes NVIDIA Hopper architecture and a Transformer Engine in order to provide training and speeds that are four times faster than previous generation models.

“These devices supercharge Purdue's investigation into AI,” says Alex Younts, principal research engineer for RCAC.

“Whether it's customizing the next big LLM for language processing, training bots to drive around, or evaluating the health of fields of crops, Gautschi has all the resources faculty need.”

Gautschi is now part of Purdue’s community cluster program, which is celebrating its 20 year anniversary and serves more than 261 active partners from 66 departments across all three Purdue campuses. Last year, 73% of Purdue’s grant awards were given to faculty using high-performance computing.

Gautschi’s capabilities have been eagerly anticipated by Purdue researchers, particularly those in AI and machine learning fields. The processing power of its H100 GPUs is critical for training and deploying AI models that demand high precision and scalability, benefiting disciplines that range from climate science to bioinformatics.

To learn more about the Gautschi cluster, please contact rcac-help@purdue.edu or visit the Gautschi home page.