Courtesy of the schedule for the SC22 conference, we now have our first glimpse at the finalists for this year’s coveted Gordon Bell Prize. The Gordon Bell Prize, of course, comes with an award of $10,000 courtesy of HPC luminary Gordon Bell. It is awarded annually to an outstanding achievement in high-performance computing, with ACM emphasizing “innovation in applying high-performance computing to applications in science, engineering, and large-scale data analytics.”
The finalist projects for the 2022 ACM Gordon Bell Prize are listed below.
2.5 million-atom ab initio electronic-structure simulation of complex metallic heterostructures with DGDFT
Listed author: Qingcai Jiang (University of Science and Technology of China)
One of two finalists confirmed to have used an exascale supercomputer, this team leveraged Sunway’s OceanLight system to enable a “massively parallel implementation of discontinuous Galerkin … density functional theory,” using that implementation to investigate the quantum electronic structures of complex metals to an ultra-precise scale of hundreds of nanometers. The simulations included 2.5 million atoms, used 28.1 million cores of OceanLight and achieved a parallel efficiency of 72 percent on the system.
The second of the two confirmed exascale finalists, this project used the HPE-built Frontier system to develop a new method for data mining enormous swaths of biomedical literature using graph analytics. “In this context,” the abstract reads, “we present COAST (Exascale Communication Optimized All Pairs Shortest Path), a new high-performance algorithm and implementation of the Floyd-Warshall algorithm for the world’s fastest supercomputer, Frontier.”
Extreme-scale earthquake simulation with uncertainty quantification
Listed author: Tsuyoshi Ichimura (University of Tokyo)
This team developed a stochastic finite element method with “ultra-large degrees of freedom” that is “designed to attain high performance on a variety of CPU/GPU-based supercomputers,” resulting in a 197× speedup for the tool when run on the full Fugaku system compared to state-of-the-art tools running on the full Summit system. “This method, which has shown its effectiveness via solving huge (32-trillion degrees-of-freedom) practical problems, is expected to be a breakthrough in [earthquake] damage mitigation,” the description reads, “and is expected to facilitate the scientific understanding of earthquake phenomena and have a ripple effect on other fields that similarly require [uncertainty quantification].”
Extreme-scale many-against-many protein similarity search
Listed author: Oguz Selvitopi (Lawrence Berkeley National Laboratory)
Supercomputer: unknown Summit
Per the paper’s abstract (shared on Twitter), the project used “over 20,000 GPUs” on the Summit supercomputer to perform a massive protein similarity search on “one of the largest publicly available datasets with 405 million proteins” — all in less than 3.5 hours. “Due to the need to construct and maintain a data structure holding indices to all other sequences, this application has a huge memory footprint that makes it hard to scale the problem sizes,” the description reads. “We overcome this memory limitation by innovative matrix-based blocking techniques, without introducing additional load imbalance.”
Pushing the frontier in laser-based electron accelerators design with groundbreaking mesh-refined particle-in-cell simulations on pre-exascale supercomputers
Listed author: Jean-Luc Vay (Lawrence Berkeley National Laboratory)
Supercomputers: Fugaku, Perlmutter, Summit
This project leveraged a trio of major supercomputers — Fugaku, Perlmutter and Summit, totaling more than 661 Linpack petaflops — to develop and run their first-of-its-kind mesh-refined, massively parallel particle-in-cell code for kinetic plasma simulations. “The … code enabled 3D simulations of laser-matter interactions on Fugaku and Summit, which have so far been out of the reach of standard codes,” the description reads. “These simulations helped remove a major limitation of compact laser-based electron accelerators, which are promising candidates for next generation high-energy physics experiments and ultra-high dose rate FLASH radiotherapy.”
Reshaping geostatistical modeling and prediction for extreme-scale environmental applications
Listed author: David Keyes (King Abdullah University of Science and Technology)
“Geostatistics augments first-principles modeling approaches for the prediction of environmental phenomena given the availability of measurements at a large number of locations,” this project’s description explains; however, traditional approaches “grow cubically in complexity,” limiting their applicability to modern datasets. Using Fugaku, the researchers developed an “adaptive approach” that scales across systems and achieves up to a 12× performance speedup against those traditional implementations.
In the coming weeks and months, we’re sure to learn more about the six finalist projects — as well as the first information about the finalists for the Gordon Bell Special Prize for High Performance Computing-Based COVID-19 Research, which will be awarded for the third time at SC22. All Gordon Bell finalist projects will be presented in advance of the award ceremony. To learn more about the projects and the SC22 schedule, click here.