Supercomputing and Computation

SHARE

Towards Parallel Discrete Event Simulation (PDES) at Exascale


Problem Statement:

  • PDES applications form a unique class of supercomputing applications, requiring customized runtime analysis, and core engine software.  Few PDES engines and applications exist in the world that have been shown to scale to peta-scale and beyond (105-106 processor cores).

Technical Approach:

  • Our scalable algorithmic framework for combined optimistic/conservative execution enables sophisticated data structures and algorithms needed for scalable PDES application execution.  Tested on up to 216,000 cores, our unique capabilities in buffering techniques, memory management, flow control, global virtual time algorithms, etc., provide demonstrated scaling and efficiency.

Benefit:

  • Applications with tight runtime needs and large scenario configurations previously unimaginable are made possible.  Multiple proof-of-concept applications to capitalize on the new capabilities stretch modelers’ and users’ solution space.  PDES application areas such as epidemiology, cyber security, logistics, and social behavior modeling are all positively impacted.
ASK ORNL

We're always happy to get feedback from our users. Please use the Comments form to send us your comments, questions, and observations.