Poster Type: ACM SRC Posters
Author: Maor Mizrahi (Hebrew University of Jerusalem)
Abstract: Domain-decomposed Monte Carlo transport moves particles between MPI ranks as they cross spatial subdomains. Transfers are irregular, bursty, and unpredictable. Existing asynchronous algorithms commonly use matched MPI messages, requiring receiver participation and progress polling.
STORM converts OFI/libfabric one-sided operations into a receiver-passive migration protocol with ordered publication, dynamic buffer growth, and termination accounting. On 2,240–13,440 ranks of Leonardo, STORM retains 88.3% constant global workload scaling efficiency under constant global transport work and more than 97% weak-scaling efficiency, with up to 1.27× speedup over optimized point-to-point MPI. In one three-dimensional Implicit Monte Carlo Hohlraum run, it is 1.41× faster at 4,480 ranks.
Best Poster Finalist (BP): no
Poster: PDF
Poster Summary: PDF