Title
Implications of application usage characteristics for collective communication offload
Abstract
The global, synchronous nature of some collective operations implies that they will become the bottleneck when scaling to hundreds of thousands of nodes. One approach improves collective performance using a programmable network interface to directly implement collectives. While these implementations improve micro-benchmark performance, accelerating applications will require deeper understanding of application behaviour. We describe several characteristics of applications that impact collective communication performance. We analyse network resource usage data to guide the design of collective offload engines and their associated programming interfaces. In particular, we provide an analysis of the potential benefit of non-blocking collective communication operations for MPI.
Year
DOI
Venue
2006
10.1504/IJHPCN.2006.010633
IJHPCN
Keywords
Field
DocType
collective performance,associated programming interface,collective operation,collective offload engine,application usage characteristic,programmable network interface,micro-benchmark performance,application behaviour,analyse network resource usage,impact collective communication performance,collective communication offload,collective communication operation,resource management,mpi,computer network,network interface,indexation
Resource management,Bottleneck,Data transmission,Computer science,Parallel computing,Collective communication,Computer network,Implementation,Usage data,Distributed computing,Network interface
Journal
Volume
Issue
Citations 
4
3/4
18
PageRank 
References 
Authors
0.91
23
4
Name
Order
Citations
PageRank
Ron Brightwell1106094.72
Sue P. Goudy2180.91
Arun Rodrigues3362.80
Keith D. Underwood484777.39