Data transfers between processes in an SMP system: performance study and application to MPI (2006)
Cached
Download Links
- [info.mcs.anl.gov]
- [www.mcs.anl.gov]
- [ftp.mcs.anl.gov]
- [www-unix.mcs.anl.gov]
- [www.mcs.anl.gov]
- DBLP
Other Repositories/Bibliography
| Venue: | in: Proceedings of the 2006 International Conference on Parallel Processing (ICPP 2006), IEEE Computer Society |
| Citations: | 28 - 6 self |
BibTeX
@INPROCEEDINGS{Buntinas06datatransfers,
author = {Darius Buntinas and Guillaume Mercier},
title = {Data transfers between processes in an SMP system: performance study and application to MPI},
booktitle = {in: Proceedings of the 2006 International Conference on Parallel Processing (ICPP 2006), IEEE Computer Society},
year = {2006},
pages = {487--496}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract — This paper focuses on the transfer of large data in SMP systems. Achieving good performance for intranode communication is critical for developing an efficient communication system, especially in the context of SMP clusters. We evaluate the performance of five transfer mechanisms: sharedmemory buffers, message queues, the Ptrace system call, kernel module-based copy, and a high-speed network. We evaluate each mechanism based on latency, bandwidth, its impact on application cache usage, and its suitability to support MPI twosided and one-sided messages. I. MOTIVATION AND SCOPE Designing a communication system tailored for a particular architecture requires understanding the achievable performance levels of the underlying hardware and software. Such understanding is key to a more efficient design and
Keyphrases
smp system data transfer performance study good performance efficient communication system motivation scope message queue sharedmemory buffer module-based copy smp cluster communication system transfer mechanism achievable performance level one-sided message efficient design application cache usage large data intranode communication particular architecture high-speed network ptrace system call
