OpenSHMEM MLIR: A Dialect for Compile-Time Optimization of One-Sided Communications
Communication increasingly limits performance in high-performance computing (HPC), yet mainstream compilers focus on computation because communication intent is lost early in compilation. OpenSHMEM offers a one-sided Partitioned Global Address Space (PGAS) model with symmetric memory and explicit synchronization, but lowering to opaque runtime calls hides these semantics from analysis.
Authors
- B Michałowicz (ORCID: https://orcid.org/0000-0002-7450-5787)
- Wendy Poole (ORCID: https://orcid.org/0009-0007-4984-6209)
- Dhabaleswar K. Panda (ORCID: https://orcid.org/0000-0002-0356-1781)
- Stephen Poole (ORCID: https://orcid.org/0000-0002-4531-7453)
- Michael Beebe (ORCID: https://orcid.org/0009-0004-9151-6607)
- Yong Chen (ORCID: https://orcid.org/0000-0002-9961-9051)
- Y V Ravi Kumar (ORCID: https://orcid.org/0009-0001-2888-3867)
- Andrew McNamara (ORCID: https://orcid.org/0009-0005-6716-491X)
Institutions
- Texas Tech University (US)
- Los Alamos National Laboratory (US)
- New Mexico Consortium (US)
- The Ohio State University (US)
Publication Details
- Published
- 2025-11-07
- DOI
- https://doi.org/10.1145/3731599.3767483
- Citations
- 1
- Primary Topic
- Parallel Computing and Optimization Techniques
- Type
- article
- Field-Weighted Citation Impact
- 2.27
Funders
- National Science Foundation
- Los Alamos National Laboratory