JIT Costing Adaptive Skeletons for Performance Portability

Patrick Maier, John Magnus Morton, Phil Trinder

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The proliferation of widely available, but very different, parallel architectures makes the ability to deliver good parallel performance on a range of architectures, or performance portability, highly desirable. Irregular parallel problems, where the number and size of tasks is unpredictable, are particularly challenging and require dynamic coordination. The paper outlines a novel approach to delivering portable parallel performance for irregular parallel programs. The approach combines JIT compiler technology with dynamic scheduling and dynamic transformation of declarative parallelism. We specify families of algorithmic skeletons plus equations for rewriting skeleton expressions. We present the design of a framework that unfolds skeletons into task graphs, dynamically schedules tasks, and dynamically rewrites skeletons, guided by a lightweight JIT trace-based cost model, to adapt the number and granularity of tasks for the architecture. We outline the system architecture and prototype implementation in Racket/Pycket. As the current prototype does not yet automatically perform dynamic rewriting we present results based on manual offline rewriting, demonstrating that (i) the system scales to hundreds of cores given enough parallelism of suitable granularity, and (ii) the JIT trace cost model predicts granularity accurately enough to guide rewriting towards a good adaptive transformation.
Original languageEnglish
Title of host publicationProceedings of the 5th International Workshop on Functional High-Performance Computing
Place of PublicationNew York, NY, USA
PublisherACM Association for Computing Machinery
Pages23–30
ISBN (Print)9781450344333
DOIs
Publication statusPublished - 8 Sep 2016
Event5th ACM SIGPLAN Workshop on Functional High-Performance Computing - Nara, Japan
Duration: 22 Sep 201622 Sep 2016
https://sites.google.com/site/fhpcworkshops/

Workshop

Workshop5th ACM SIGPLAN Workshop on Functional High-Performance Computing
Abbreviated titleFHPC 2016
CountryJapan
CityNara
Period22/09/1622/09/16
Internet address

Keywords

  • cost model
  • parallelism
  • performance portability

Fingerprint Dive into the research topics of 'JIT Costing Adaptive Skeletons for Performance Portability'. Together they form a unique fingerprint.

Cite this