Purdue University

Purdue’s Anvil supercomputer has debuted at number 143 on the Top500 list of the world’s most powerful supercomputers, announced this week at the ISC22 supercomputing conference.

With a peak processing speed of 5.1 petaFLOPs, Anvil will deliver over 1 billion core hours to the National Science Foundation’s Extreme Science and Engineering Discovery Environment (XSEDE), in which Purdue is a partner.

Anvil is the tenth Purdue supercomputer to make the Top 500 list, and the seventh highest ranked academic system in the United States. Only 14 institutions in the U.S. currently have a supercomputer on the Top 500 list.

“Anvil not only provides users with the performance they need, as evidenced by its place among the world’s most powerful supercomputers, it was also built with a focus on usability and easy access for those in traditionally under-served domains,” says Carol Song, senior research scientist for the Rosen Center for Advanced Computing and project director for Anvil. “It’s the greatest feeling when users tell us Anvil exceeded their expectations, for example, by reducing months of machine learning model training time to hours.”

After being in early user testing earlier this year, Anvil is now widely available to researchers in the United States. Researchers may request access to Anvil via the XSEDE allocation process.

Early users of Anvil like Yinglong Miao of the University of Kansas, used Anvil's power to do in weeks what would take months, if not years, on a personal computer. Anvil has already been used by researchers for projects including modeling the virus that causes Covid-19, drug discovery, and astrophysical modeling.

To date, more than 137 million CPU core hours and 310,000 GPU core hours on Anvil have been allocated to research projects. Research projects around the US have used approximately 40 million CPU core hours on Anvil for research projects and education.

Anvil consists of 1,000 nodes with two 64-core third-generation AMD Epyc processors each. Anvil's nodes are interconnected with 100 Gbps Mellanox HDR InfiniBand. The supercomputer ecosystem also includes 32 large memory nodes, each with 1 TB of RAM, and 16 nodes each with four NVIDIA A100 Tensor Core GPUs providing 1.5 PF of single-precision performance to support machine learning and artificial intelligence applications.

Research on Anvil will be able to leverage a diverse set of storage technologies, anchored by a 10-plus PB parallel filesystem, boosted with over 3 PB of flash disk. Novel workflows will benefit from block and object storage systems also supported by Anvil.

Anvil makes all of these powerful capabilities accessible through familiar interactive access interfaces, such as Open OnDemand, Jupyter Notebooks, R Studio and remote desktops via Thinlinc. Anvil also makes it easy for researchers to access tools such as the NVIDIA NGC containers for machine learning workflows, and the biocontainers deployed by the Anvil team that provide more than 400 bioinformatics tools.

Anvil’s powerful hardware and versatile software ecosystem is complemented by a team of experts in high-performance computing and domain sciences ready to help researchers.

More information about Anvil is available on Purdue’s Anvil website. Anyone with questions should contact anvil@purdue.edu. Anvil is funded under NSF award No. 2005632.

Writer: Adrienne Miller, science and technology writer, Rosen Center for Advanced Computing, mill2027@purdue.edu.