Optimizing the hybrid parallelization of BHAC
Autor: | Oliver Porth, Hector Olivares, Salvatore Cielo, Anupam Karmakar, Chun Xia, Luigi Iapichino |
---|---|
Přispěvatelé: | High Energy Astrophys. & Astropart. Phys (API, FNWI) |
Rok vydání: | 2022 |
Předmět: |
Profiling (computer programming)
Scheme (programming language) FOS: Computer and information sciences Computer science Concurrency Astronomy FOS: Physical sciences Astronomy and Astrophysics Parallel computing USable Supercomputer Porting Computer Science Applications Computer Science - Distributed Parallel and Cluster Computing Space and Planetary Science Code (cryptography) x86 Distributed Parallel and Cluster Computing (cs.DC) Astrophysics - Instrumentation and Methods for Astrophysics computer Instrumentation and Methods for Astrophysics (astro-ph.IM) computer.programming_language |
Zdroj: | Astronomy and Computing, 38, 1-9 Astronomy and Computing, 38, pp. 1-9 Astronomy and Computing, 38:100509. Elsevier |
ISSN: | 2213-1337 |
Popis: | We present our experience with the modernization on the GR-MHD code BHAC, aimed at improving its novel hybrid (MPI+OpenMP) parallelization scheme. In doing so, we showcase the use of performance profiling tools usable on x86 (Intel-based) architectures. Our performance characterization and threading analysis provided guidance in improving the concurrency and thus the efficiency of the OpenMP parallel regions. We assess scaling and communication patterns in order to identify and alleviate MPI bottlenecks, with both runtime switches and precise code interventions. The performance of optimized version of BHAC improved by $\sim28\%$, making it viable for scaling on several hundreds of supercomputer nodes. We finally test whether porting such optimizations to different hardware is likewise beneficial on the new architecture by running on ARM A64FX vector nodes. Comment: 10 pages, 9 figures, 1 table; in review |
Databáze: | OpenAIRE |
Externí odkaz: |