2011
DOI: 10.1007/978-3-642-24449-0_8
|View full text |Cite
|
Sign up to set email alerts
|

Multi-core and Network Aware MPI Topology Functions

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
28
0

Year Published

2013
2013
2017
2017

Publication Types

Select...
4
4

Relationship

0
8

Authors

Journals

citations
Cited by 47 publications
(28 citation statements)
references
References 5 publications
0
28
0
Order By: Relevance
“…These give the computational scientist a portable way to map an MPI program onto a large parallel machine in a way that provides for efficient (low contention) communication. Some of this work was reported on in [22]. Some related work looked at hot spots in two--level direct networks, such as those for the IBM PERCS system, which was originally to be installed at Illinois.…”
Section: Some Of the Most Interesting Results From This Project Addrementioning
confidence: 99%
“…These give the computational scientist a portable way to map an MPI program onto a large parallel machine in a way that provides for efficient (low contention) communication. Some of this work was reported on in [22]. Some related work looked at hot spots in two--level direct networks, such as those for the IBM PERCS system, which was originally to be installed at Illinois.…”
Section: Some Of the Most Interesting Results From This Project Addrementioning
confidence: 99%
“…Other algorithms have been proposed, such as Treematch [35] and Zoltan [36]. We selected Scotch because it has a short execution time (less than 1 ms to map 32 processes) while providing good results [37][38][39]13,40], and is therefore suitable for online mapping. Scotch has a complexity of OðN 3 Þ, where N is the number of processes to be mapped [38].…”
Section: Calculating the Mappingmentioning
confidence: 99%
“…The MPI provides a set of interfaces that are designed to assist the library in constructing a virtual topology from the application's communication pattern and in supporting remapping (reordering) of the processes to the available cores to optimize performance [35]. Communication among the cores is achieved by sending and receiving messages among the corresponding processes.…”
Section: Parallel Mpso Displacement Back-analysis Proceduresmentioning
confidence: 99%