Paper Info
Title
Dynamic Task and Data Placement over NUMA Architectures: An OpenMP Runtime Perspective
Abstract
Exploiting the full computational power of current hierarchical multiprocessor machines requires a very careful distribution of threads and data among the underlying non-uniform architecture so as to avoid memory access penalties. Directive-based programming languages such as OpenMPprovide programmers with an easy way to structure the parallelism of their application and to transmit this information to the runtime system. Our runtime, which is based on a multi-level thread scheduler combined with a NUMA-aware memory manager, converts this information into "scheduling hints" to solve thread/memory affinity issues. It enables dynamic load distribution guided by application structure and hardware topology, thus helping to achieve performance portability. First experiments show that mixed solutions (migrating threads and data) outperform next-touch -based data distribution policies and open possibilities for new optimizations.
Year
DOI
Venue
2009
10.1007/978-3-642-02303-3_7
IWOMP
Keywords
Field
DocType
dynamic load distribution,migrating thread,dynamic task,careful distribution,data distribution policy,memory access penalty,multi-level thread scheduler,numa-aware memory manager,openmp runtime perspective,application structure,numa architectures,runtime system,data placement,memory affinity issue,multi core,memory,numa,memory management,programming language
Interleaved memory,Computer architecture,Scheduling (computing),Computer science,Parallel computing,Cache-only memory architecture,Thread (computing),Memory model,Memory management,Software portability,Runtime system
Conference
Volume
ISSN
Citations 
5568
0302-9743
33
PageRank 
References 
Authors
1.50
12
5
Name
Order
Citations
PageRank
François Broquedis115711.99
Nathalie Furmento230027.19
Brice Goglin322621.78
Raymond Namyst4140583.04
Pierre-andré Wacrenier576636.69