Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor

Alastair Donaldson; Paul Keir; Anton Lokhmotov

doi:10.1007/978-3-642-00955-6_21

Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor

Alastair Donaldson, Paul Keir, Anton Lokhmotov

Research output: Chapter in Book/Report/Conference proceeding › Chapter › peer-review

5 Citations (Scopus)

Abstract

We describe compiler and run-time optimisations for effective auto-parallelisation of C++ programs on the Cell BE architecture. Auto-parallelisation is made easier by annotating sieve scopes, which abstract the "read in, compute in parallel, write out" processing paradigm. We show that the semantics of sieve scopes enables data movement optimisations, such as re-organising global memory reads to minimise DMA transfers and streaming reads from uniformly accessed arrays. We also describe run-time optimisations for committing side-effects to main memory. We provide experimental results showing the benefits of our optimisations, and compare the Sieve-Cell system with IBM’s OpenMP implementation for Cell.

Original language	English
Title of host publication	Euro-Par 2008
Subtitle of host publication	Euro-Par 2008 Workshops - Parallel Processing
Editors	Eduardo César, Michael Alexander, Achim Streit, Jesper Larsson Träff, Christophe Cérin, Andreas Knüpfer, Dieter Kranzlmüller, Shantenu Jha
Publisher	Springer-Verlag
Pages	163-173
Number of pages	11
Volume	5415
ISBN (Print)	978-3-642-00955-6
DOIs	https://doi.org/10.1007/978-3-642-00955-6_21
Publication status	Published - 2009
Externally published	Yes

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
ISSN (Print)	0302-9743

Access to Document

10.1007/978-3-642-00955-6_21

Cite this

Donaldson, A., Keir, P., & Lokhmotov, A. (2009). Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor. In E. César, M. Alexander, A. Streit, J. L. Träff, C. Cérin, A. Knüpfer, D. Kranzlmüller, & S. Jha (Eds.), Euro-Par 2008: Euro-Par 2008 Workshops - Parallel Processing (Vol. 5415, pp. 163-173). (Lecture Notes in Computer Science). Springer-Verlag. https://doi.org/10.1007/978-3-642-00955-6_21

Donaldson, Alastair ; Keir, Paul ; Lokhmotov, Anton. / Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor. Euro-Par 2008: Euro-Par 2008 Workshops - Parallel Processing. editor / Eduardo César ; Michael Alexander ; Achim Streit ; Jesper Larsson Träff ; Christophe Cérin ; Andreas Knüpfer ; Dieter Kranzlmüller ; Shantenu Jha. Vol. 5415 Springer-Verlag, 2009. pp. 163-173 (Lecture Notes in Computer Science).

@inbook{ecc5d8eb12574aada758797df942e938,

title = "Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor",

abstract = "We describe compiler and run-time optimisations for effective auto-parallelisation of C++ programs on the Cell BE architecture. Auto-parallelisation is made easier by annotating sieve scopes, which abstract the {"}read in, compute in parallel, write out{"} processing paradigm. We show that the semantics of sieve scopes enables data movement optimisations, such as re-organising global memory reads to minimise DMA transfers and streaming reads from uniformly accessed arrays. We also describe run-time optimisations for committing side-effects to main memory. We provide experimental results showing the benefits of our optimisations, and compare the Sieve-Cell system with IBM{\textquoteright}s OpenMP implementation for Cell.",

author = "Alastair Donaldson and Paul Keir and Anton Lokhmotov",

year = "2009",

doi = "10.1007/978-3-642-00955-6_21",

language = "English",

isbn = "978-3-642-00955-6",

volume = "5415",

series = "Lecture Notes in Computer Science",

publisher = "Springer-Verlag",

pages = "163--173",

editor = "Eduardo C{\'e}sar and Michael Alexander and Achim Streit and Tr{\"a}ff, {Jesper Larsson } and C{\'e}rin, {Christophe } and Kn{\"u}pfer, {Andreas } and Kranzlm{\"u}ller, {Dieter } and Jha, {Shantenu }",

booktitle = "Euro-Par 2008",

address = "Germany",

}

Donaldson, A, Keir, P & Lokhmotov, A 2009, Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor. in E César, M Alexander, A Streit, JL Träff, C Cérin, A Knüpfer, D Kranzlmüller & S Jha (eds), Euro-Par 2008: Euro-Par 2008 Workshops - Parallel Processing. vol. 5415, Lecture Notes in Computer Science, Springer-Verlag, pp. 163-173. https://doi.org/10.1007/978-3-642-00955-6_21

Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor. / Donaldson, Alastair; Keir, Paul; Lokhmotov, Anton.

Euro-Par 2008: Euro-Par 2008 Workshops - Parallel Processing. ed. / Eduardo César; Michael Alexander; Achim Streit; Jesper Larsson Träff; Christophe Cérin; Andreas Knüpfer; Dieter Kranzlmüller; Shantenu Jha. Vol. 5415 Springer-Verlag, 2009. p. 163-173 (Lecture Notes in Computer Science).

Research output: Chapter in Book/Report/Conference proceeding › Chapter › peer-review

TY - CHAP

T1 - Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor

AU - Donaldson, Alastair

AU - Keir, Paul

AU - Lokhmotov, Anton

PY - 2009

Y1 - 2009

N2 - We describe compiler and run-time optimisations for effective auto-parallelisation of C++ programs on the Cell BE architecture. Auto-parallelisation is made easier by annotating sieve scopes, which abstract the "read in, compute in parallel, write out" processing paradigm. We show that the semantics of sieve scopes enables data movement optimisations, such as re-organising global memory reads to minimise DMA transfers and streaming reads from uniformly accessed arrays. We also describe run-time optimisations for committing side-effects to main memory. We provide experimental results showing the benefits of our optimisations, and compare the Sieve-Cell system with IBM’s OpenMP implementation for Cell.

AB - We describe compiler and run-time optimisations for effective auto-parallelisation of C++ programs on the Cell BE architecture. Auto-parallelisation is made easier by annotating sieve scopes, which abstract the "read in, compute in parallel, write out" processing paradigm. We show that the semantics of sieve scopes enables data movement optimisations, such as re-organising global memory reads to minimise DMA transfers and streaming reads from uniformly accessed arrays. We also describe run-time optimisations for committing side-effects to main memory. We provide experimental results showing the benefits of our optimisations, and compare the Sieve-Cell system with IBM’s OpenMP implementation for Cell.

U2 - 10.1007/978-3-642-00955-6_21

DO - 10.1007/978-3-642-00955-6_21

M3 - Chapter

SN - 978-3-642-00955-6

VL - 5415

T3 - Lecture Notes in Computer Science

SP - 163

EP - 173

BT - Euro-Par 2008

A2 - César, Eduardo

A2 - Alexander, Michael

A2 - Streit, Achim

A2 - Träff, Jesper Larsson

A2 - Cérin, Christophe

A2 - Knüpfer, Andreas

A2 - Kranzlmüller, Dieter

A2 - Jha, Shantenu

PB - Springer-Verlag

ER -

Donaldson A, Keir P, Lokhmotov A. Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor. In César E, Alexander M, Streit A, Träff JL, Cérin C, Knüpfer A, Kranzlmüller D, Jha S, editors, Euro-Par 2008: Euro-Par 2008 Workshops - Parallel Processing. Vol. 5415. Springer-Verlag. 2009. p. 163-173. (Lecture Notes in Computer Science). doi: 10.1007/978-3-642-00955-6_21

Compile-time and run-time issues in an auto-parallelisation system for the cell BE processor

Abstract

Publication series

Access to Document

Fingerprint

Cite this