The Impact of New and Emerging Architectures on Computational Physics
ORAL
Abstract
We present experiences with Ookami, a testbed machine providing open access to the A64FX processor developed by Fujitsu for the Japanese path to exascale computing. We provide an overview of the project with details of the user base and the diverse applications of users. We also present a case study of experience with running the community simulation code FLASH, developed at the University of Chicago for multi-scale multi-physics applications. We present results with different compilers and other software, our initial scaling results, and attempts to utilize the A64FX's SVE instructions and NUMA architecture. We present results from several applications including scaling studies, compiler comparisons, and tuning for SVE and communication patterns. We also share best practices for getting the most out of this new architecture.
*Ookami is supported by the US NSF under grant OAC 1927880. FLASH was developed at the Flash Center for Computational Science at the University of Chicago. Additional support came from the US DOE under grant DE-FG02-87ER40317.
–
Presenters
-
Eva Siegmann
- Stony Brook University