Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures

Siegfried Benkner; Sabri Pllana; Jesper Larsson Träff; Philippas Tsigas; Andrew Richards; Raymond Namyst; H. Cornelius; Christoph Kessler; David Moloney; Peter Sanders

doi:10.1002/9781119332015.ch12

Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures

Siegfried Benkner
, Sabri Pllana
, Jesper Larsson Träff
, Philippas Tsigas
, Andrew Richards
, Raymond Namyst
, H. Cornelius
, Christoph Kessler
, David Moloney
, Peter Sanders

Research Group Scientific Computing

Publications: Contribution to book › Chapter › Peer Reviewed

Abstract

PEPPHER takes a pluralistic and parallelization agnostic approach to programmability and performance portability for heterogeneous many-core architectures. The PEPPHER framework is in principle language independent but focuses on supporting C++ code with PEPPHER-specific annotations as pragmas or external annotations. The framework is open and extensible; the PEPPHER methodology details how new architectures are incorporated. The PEPPHER methodology consists of rules for how to extend the framework for new architectures. This mainly concerns adaptivity and autotuning for algorithm libraries, the necessary hooks and extensions for the run-time system and any supporting algorithms and data structures that this relies on. Offloading is a specific technique for programming heterogeneous platforms that can sometimes be applied with high efficiency. Offload as developed by the PEPPHER partner Codeplay is a particular, nonintrusive C++ extension allowing portable C++ code to support diverse heterogeneous multicore architectures in a single code base.

Original language	English
Title of host publication	Programming Multi-Core and Many-Core Computing Systems
Editors	Sabri Pllana, Fatos Xhafa
Place of Publication	New York, NY
Publisher	Wiley
Pages	241-260
Number of pages	20
Volume	1
Edition	1
ISBN (Electronic)	9781119332015
ISBN (Print)	9780470936900
DOIs	https://doi.org/10.1002/9781119332015.ch12
Publication status	Published - Jan 2017

Publication series

Series	Wiley Series on Parallel and Distributed Computing

Austrian Fields of Science 2012

102023 Supercomputing
102029 Practical computer science

Keywords

External annotations
Heterogeneous many-core architectures
Multicore programmability
Offloading
Parallel language extensions
PEPPHER framework
Performance portability

Access to Document

10.1002/9781119332015.ch12

Cite this

Benkner, S., Pllana, S., Träff, J. L., Tsigas, P., Richards, A., Namyst, R., Cornelius, H., Kessler, C., Moloney, D., & Sanders, P. (2017). Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures. In S. Pllana, & F. Xhafa (Eds.), Programming Multi-Core and Many-Core Computing Systems (1 ed., Vol. 1, pp. 241-260). Wiley. https://doi.org/10.1002/9781119332015.ch12

@inbook{aa5a60f61b3b473585ea24279dc1251b,

title = "Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures",

abstract = "PEPPHER takes a pluralistic and parallelization agnostic approach to programmability and performance portability for heterogeneous many-core architectures. The PEPPHER framework is in principle language independent but focuses on supporting C++ code with PEPPHER-specific annotations as pragmas or external annotations. The framework is open and extensible; the PEPPHER methodology details how new architectures are incorporated. The PEPPHER methodology consists of rules for how to extend the framework for new architectures. This mainly concerns adaptivity and autotuning for algorithm libraries, the necessary hooks and extensions for the run-time system and any supporting algorithms and data structures that this relies on. Offloading is a specific technique for programming heterogeneous platforms that can sometimes be applied with high efficiency. Offload as developed by the PEPPHER partner Codeplay is a particular, nonintrusive C++ extension allowing portable C++ code to support diverse heterogeneous multicore architectures in a single code base.",

keywords = "External annotations, Heterogeneous many-core architectures, Multicore programmability, Offloading, Parallel language extensions, PEPPHER framework, Performance portability",

author = "Siegfried Benkner and Sabri Pllana and Tr{\"a}ff, \{Jesper Larsson\} and Philippas Tsigas and Andrew Richards and Raymond Namyst and H. Cornelius and Christoph Kessler and David Moloney and Peter Sanders",

year = "2017",

month = jan,

doi = "10.1002/9781119332015.ch12",

language = "English",

isbn = "9780470936900",

volume = "1",

series = "Wiley Series on Parallel and Distributed Computing",

publisher = "Wiley",

pages = "241--260",

editor = "Sabri Pllana and Fatos Xhafa",

booktitle = "Programming Multi-Core and Many-Core Computing Systems",

edition = "1",

}

Benkner, S, Pllana, S, Träff, JL, Tsigas, P, Richards, A, Namyst, R, Cornelius, H, Kessler, C, Moloney, D & Sanders, P 2017, Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures. in S Pllana & F Xhafa (eds), Programming Multi-Core and Many-Core Computing Systems. 1 edn, vol. 1, Wiley, New York, NY, Wiley Series on Parallel and Distributed Computing, pp. 241-260. https://doi.org/10.1002/9781119332015.ch12

Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures. / Benkner, Siegfried; Pllana, Sabri; Träff, Jesper Larsson et al.
Programming Multi-Core and Many-Core Computing Systems. ed. / Sabri Pllana; Fatos Xhafa. Vol. 1 1. ed. New York, NY: Wiley, 2017. p. 241-260 (Wiley Series on Parallel and Distributed Computing).

Publications: Contribution to book › Chapter › Peer Reviewed

TY - CHAP

T1 - Peppher

T2 - Performance Portability and Programmability for Heterogeneous Many-Core Architectures

AU - Benkner, Siegfried

AU - Pllana, Sabri

AU - Träff, Jesper Larsson

AU - Tsigas, Philippas

AU - Richards, Andrew

AU - Namyst, Raymond

AU - Cornelius, H.

AU - Kessler, Christoph

AU - Moloney, David

AU - Sanders, Peter

PY - 2017/1

Y1 - 2017/1

N2 - PEPPHER takes a pluralistic and parallelization agnostic approach to programmability and performance portability for heterogeneous many-core architectures. The PEPPHER framework is in principle language independent but focuses on supporting C++ code with PEPPHER-specific annotations as pragmas or external annotations. The framework is open and extensible; the PEPPHER methodology details how new architectures are incorporated. The PEPPHER methodology consists of rules for how to extend the framework for new architectures. This mainly concerns adaptivity and autotuning for algorithm libraries, the necessary hooks and extensions for the run-time system and any supporting algorithms and data structures that this relies on. Offloading is a specific technique for programming heterogeneous platforms that can sometimes be applied with high efficiency. Offload as developed by the PEPPHER partner Codeplay is a particular, nonintrusive C++ extension allowing portable C++ code to support diverse heterogeneous multicore architectures in a single code base.

AB - PEPPHER takes a pluralistic and parallelization agnostic approach to programmability and performance portability for heterogeneous many-core architectures. The PEPPHER framework is in principle language independent but focuses on supporting C++ code with PEPPHER-specific annotations as pragmas or external annotations. The framework is open and extensible; the PEPPHER methodology details how new architectures are incorporated. The PEPPHER methodology consists of rules for how to extend the framework for new architectures. This mainly concerns adaptivity and autotuning for algorithm libraries, the necessary hooks and extensions for the run-time system and any supporting algorithms and data structures that this relies on. Offloading is a specific technique for programming heterogeneous platforms that can sometimes be applied with high efficiency. Offload as developed by the PEPPHER partner Codeplay is a particular, nonintrusive C++ extension allowing portable C++ code to support diverse heterogeneous multicore architectures in a single code base.

KW - External annotations

KW - Heterogeneous many-core architectures

KW - Multicore programmability

KW - Offloading

KW - Parallel language extensions

KW - PEPPHER framework

KW - Performance portability

UR - https://www.scopus.com/pages/publications/85019645755

U2 - 10.1002/9781119332015.ch12

DO - 10.1002/9781119332015.ch12

M3 - Chapter

SN - 9780470936900

VL - 1

T3 - Wiley Series on Parallel and Distributed Computing

SP - 241

EP - 260

BT - Programming Multi-Core and Many-Core Computing Systems

A2 - Pllana, Sabri

A2 - Xhafa, Fatos

PB - Wiley

CY - New York, NY

ER -

Benkner S, Pllana S, Träff JL, Tsigas P, Richards A, Namyst R et al. Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures. In Pllana S, Xhafa F, editors, Programming Multi-Core and Many-Core Computing Systems. 1 ed. Vol. 1. New York, NY: Wiley. 2017. p. 241-260. (Wiley Series on Parallel and Distributed Computing). doi: 10.1002/9781119332015.ch12

Peppher: Performance Portability and Programmability for Heterogeneous Many-Core Architectures

Abstract

Publication series

Austrian Fields of Science 2012

Keywords

Access to Document

Other files and links

Fingerprint

Cite this