Optimization for multi-thread data-flow software

Authors:
Helmut Hlavacs;Michael Nussbaumer
Affiliations:
Research Group Entertainment Computing, University of Vienna, Vienna, Austria;Research Group Entertainment Computing, University of Vienna, Vienna, Austria
Venue:
EPEW'11 Proceedings of the 8th European conference on Computer Performance Engineering
Year:
2011

Citing 6
Cited 0

Automatically tuned linear algebra software

SC '98 Proceedings of the 1998 ACM/IEEE conference on Supercomputing
A review on queueing network models with finite capacity queues for software architectures performance prediction

Performance Evaluation
Queueing Networks and Markov Chains

Queueing Networks and Markov Chains
Optimizing system configurations quickly by guessing at the performance

Proceedings of the 2007 ACM SIGMETRICS international conference on Measurement and modeling of computer systems
System Modeling and Analysis: Foundations of System Performance Evaluation

System Modeling and Analysis: Foundations of System Performance Evaluation
Optimization towards Consolidation or Throughput for Multi-thread Software

PAAP '10 Proceedings of the 2010 3rd International Symposium on Parallel Architectures, Algorithms and Programming

Quantified Score

Hi-index	0.00

Visualization

Abstract

This work presents an optimization tool that finds the optimal number of threads for multi-thread data-flow software. Threads are assumed to encapsulate parallel executable key functionalities, are connected through finite capacity queues, and require certain hardware resources. We show how a combination of measurement and calculation, based on queueing theory, leads to an algorithm that recursively determines the best combination of threads, i.e. the best configuration of the multi-thread data-flow software on a given host. The algorithm proceeds on the directed graph of a queueing network that models this software. Experiments on different machines verify our optimization approach.