Scalable Asynchronous Domain Decomposition Solvers
Parallel implementations of linear iterative solvers generally alternate between phases of data exchange and phases of local computation. Increasingly large problem sizes and more heterogeneous compute architectures make load balancing and the design of low latency network interconnects that are able to satisfy the communication requirements of linear solvers very challenging tasks. In particular, global communication patterns such as inner products become increasingly limiting at scale. We explore the use of asynchronous communication based on one-sided Message Passing Interface primitives in the context of domain decomposition solvers. In particular, a scalable asynchronous two-level Schwarz method is presented. Here, we discuss practical issues encountered in the development of a scalable solver and show experimental results obtained on a state-of-the-art supercomputer system that illustrate the benefits of asynchronous solvers in load balanced as well as load imbalanced scenarios. Using the novel method, we can observe speedups of up to four times over its classical synchronous equivalent.
- Research Organization:
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States); Temple Univ., Philadelphia, PA (United States); Georgia Institute of Technology, Atlanta, GA (United States)
- Sponsoring Organization:
- USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
- Grant/Contract Number:
- AC04-94AL85000; SC0016578; NA0003525; SC0016564; AC02-05CH11231
- OSTI ID:
- 1769906
- Alternate ID(s):
- OSTI ID: 1832745
- Report Number(s):
- SAND-2020-8220J; 689827
- Journal Information:
- SIAM Journal on Scientific Computing, Vol. 42, Issue 6; ISSN 1064-8275
- Publisher:
- SIAMCopyright Statement
- Country of Publication:
- United States
- Language:
- English
Similar Records
Evaluating asynchronous Schwarz solvers on GPUs
An Asynchronous Task-based Fan-Both Sparse Cholesky Solver