Exploiting Atomic Operations for Barrier on Cray XE/XK Systems
Conference
·
· EuroMPI 2012: Recent Advances in the Message Passing Interface
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
- Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Barrier is a collective operation used by many scientific applications and parallel libraries for synchronization. Typically, a Barrier operation is implemented by exchanging a short data message that requires demultiplexing, thereby adding undesired latency to the operation. In this work, we reduce the latency of Barrier operations for Cray XE/XK systems by leveraging the atomic operations provided by the Gemini interconnect, tailoring algorithms to utilize these capabilities, and utilizing a hierarchical design to arrive at an efficient implementation. Our micro-benchmark evaluation shows that for a 4,096 process Barrier operation, the atomic-operations-based Barrier outperforms the data exchange Barrier by 52% and the native Barrier by 111%.
- Research Organization:
- Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States). Oak Ridge Leadership Computing Facility (OLCF); Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- USDOE Office of Science; USDOE
- OSTI ID:
- 1567525
- Conference Information:
- Journal Name: EuroMPI 2012: Recent Advances in the Message Passing Interface
- Country of Publication:
- United States
- Language:
- English
Two algorithms for barrier synchronization
|
journal | February 1988 |
Open MPI: Goals, Concept, and Design of a Next Generation MPI Implementation
|
book | January 2004 |
The Gemini System Interconnect
|
conference | August 2010 |
Cheetah: A Framework for Scalable Hierarchical Collective Operations
|
conference | May 2011 |
Similar Records
Open MPI for Cray XE/XK Systems
A uGNI-based Asynchronous Message-driven Runtime System for Cray Supercomputers with Gemini Interconnect
A uGNI-Based Asynchronous Message-driven Runtime System for Cray Supercomputers with Gemini Interconnect
Conference
·
Sat Dec 31 23:00:00 EST 2011
·
OSTI ID:1049087
A uGNI-based Asynchronous Message-driven Runtime System for Cray Supercomputers with Gemini Interconnect
Conference
·
Thu Aug 16 00:00:00 EDT 2012
· 2012 IEEE 26TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS)
·
OSTI ID:1567313
A uGNI-Based Asynchronous Message-driven Runtime System for Cray Supercomputers with Gemini Interconnect
Conference
·
Sat Dec 31 23:00:00 EST 2011
·
OSTI ID:1056922