Efficient Dynamic Pinning of Parallelized Applications by Distributed Reinforcement Learning

This paper introduces a resource allocation framework specifically tailored for addressing the problem of dynamic placement (or pinning) of parallelized applications to processing units. Under the proposed setup each thread of the parallelized application constitutes an independent decision maker (o...

Full description

Saved in:
Bibliographic Details
Published in:International journal of parallel programming Vol. 47; no. 1; pp. 24 - 38
Main Authors: Chasparis, Georgios C., Rossbory, Michael
Format: Journal Article
Language:English
Published: New York Springer US 01.02.2019
Springer Nature B.V
Subjects:
Affinity
Central processing units
Computer Science
CPUs
Decision making
Machine learning
Parallel processing
Placement
Processor Architectures
Resource allocation
Resource scheduling
Software Engineering/Programming and Operating Systems
Special Issue on High-Level Programming for Heterogeneous Parallel Systems
Theory of Computation
Dynamic pinning
Reinforcement learning
Parallel applications
ISSN:0885-7458, 1573-7640
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper introduces a resource allocation framework specifically tailored for addressing the problem of dynamic placement (or pinning) of parallelized applications to processing units. Under the proposed setup each thread of the parallelized application constitutes an independent decision maker (or agent), which (based on its own prior performance measurements and its own prior CPU-affinities) decides on which processing unit to run next. Decisions are updated recursively for each thread by a resource manager/scheduler which runs in parallel to the application’s threads and periodically records their performances and assigns to them new CPU affinities. For updating the CPU-affinities, the scheduler uses a distributed reinforcement-learning algorithm, each branch of which is responsible for assigning a new placement strategy to each thread. The proposed framework is flexible enough to address alternative optimization criteria, such as maximum average processing speed and minimum speed variance among threads. We demonstrate analytically that convergence to locally-optimal placements is achieved asymptotically. Finally, we validate these results through experiments in Linux platforms.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0885-7458
1573-7640
DOI:10.1007/s10766-017-0541-y