Enabling communication concurrency through flexible MPI endpoints
- Intel Corporation, Hudson, MA (United States)
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Argonne National Lab. (ANL), Argonne, IL (United States)
- Cisco Systems Inc., San Jose, CA (United States)
- International Business Machines Corporation, Rochester, MN (United States)
MPI defines a one-to-one relationship between MPI processes and ranks. This model captures many use cases effectively; however, it also limits communication concurrency and interoperability between MPI and programming models that utilize threads. Our paper describes the MPI endpoints extension, which relaxes the longstanding one-to-one relationship between MPI processes and ranks. Using endpoints, an MPI implementation can map separate communication contexts to threads, allowing them to drive communication independently. Also, endpoints enable threads to be addressable in MPI operations, enhancing interoperability between MPI and other programming models. Furthermore, these characteristics are illustrated through several examples and an empirical study that contrasts current multithreaded communication performance with the need for high degrees of communication concurrency to achieve peak communication performance.
- Research Organization:
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE National Nuclear Security Administration (NNSA)
- Grant/Contract Number:
- AC04-94AL85000
- OSTI ID:
- 1140752
- Report Number(s):
- SAND2014-0614J; 498469
- Journal Information:
- International Journal of High Performance Computing Applications, Vol. 28, Issue 4; ISSN 1094-3420
- Publisher:
- SAGECopyright Statement
- Country of Publication:
- United States
- Language:
- English
Web of Science
The impact of hybrid-core processors on MPI message rate
|
conference | January 2013 |
Hybrid PGAS runtime support for multicore nodes
|
conference | January 2010 |
Generalized communicators in the message passing interface
|
journal | June 2001 |
Enabling MPI interoperability through flexible communication endpoints
|
conference | January 2013 |
Hybrid parallel programming with MPI and unified parallel C
|
conference | January 2010 |
Enabling Concurrent Multithreaded MPI Communication on Multicore Petascale Systems
|
book | January 2010 |
Ownership passing: efficient distributed memory programming on multi-core systems
|
conference | January 2013 |
Dynamic Communicators in MPI
|
book | January 2009 |
MPI + MPI: a new hybrid approach to parallel programming with MPI plus shared memory
|
journal | May 2013 |
Supporting Hybrid MPI and OpenSHMEM over InfiniBand: Design and Performance Evaluation
|
conference | September 2012 |
Unifying UPC and MPI runtimes: experience with MVAPICH
|
conference | January 2010 |
FG-MPI: Fine-grain MPI for multicore and clusters
|
conference | April 2010 |
PAMI: A Parallel Active Message Interface for the Blue Gene/Q Supercomputer
|
conference | May 2012 |
NUMA-aware shared-memory collective communication for MPI
|
conference | January 2013 |
Multi-threaded UPC runtime with network endpoints: Design alternatives and evaluation on multi-core architectures
|
conference | December 2011 |
Hybrid MPI/OpenMP Parallel Programming on Clusters of Multi-Core SMP Nodes
|
conference | February 2009 |
Development of Mixed Mode MPI / OpenMP Applications
|
journal | January 2001 |
Extending MPI to accelerators
|
conference | January 2011 |
Network Endpoints for Clusters of SMPs
|
conference | October 2012 |
Compact and Efficient Implementation of the MPI Group Operations
|
book | January 2010 |
Evaluating NIC hardware requirements to achieve high message rate PGAS support on multi-core processors
|
conference | January 2007 |
MVAPICH2-GPU: optimized GPU to GPU communication for InfiniBand clusters
|
journal | April 2011 |
Portable, MPI-interoperable coarray fortran
|
conference | January 2014 |
Similar Records
Optimizing point‐to‐point communication between adaptive MPI endpoints in shared memory
Test suite for evaluating performance of multithreaded MPI communication.