Porting the PLASMA Numerical Library to the OpenMP Standard

Asim YarKhan; Jakub Kurzak; Piotr Luszczek; Jack Dongarra

doi:10.1007/s10766-016-0441-6

Porting the PLASMA Numerical Library to the OpenMP Standard

Asim YarKhan, Jakub Kurzak, Piotr Luszczek, Jack Dongarra

Research output: Contribution to journal › Article › peer-review

28 Scopus citations

Abstract

PLASMA is a numerical library intended as a successor to LAPACK for solving problems in dense linear algebra on multicore processors. PLASMA relies on the QUARK scheduler for efficient multithreading of algorithms expressed in a serial fashion. QUARK is a superscalar scheduler and implements automatic parallelization by tracking data dependencies and resolving data hazards at runtime. Recently, this type of scheduling has been incorporated in the OpenMP standard, which allows to transition PLASMA from the proprietary solution offered by QUARK to the standard solution offered by OpenMP. This article studies the feasibility of such transition.

Original language	English
Pages (from-to)	612-633
Number of pages	22
Journal	International Journal of Parallel Programming
Volume	45
Issue number	3
DOIs	https://doi.org/10.1007/s10766-016-0441-6
State	Published - Jun 1 2017
Externally published	Yes

Funding

This work has been supported in part by the National Science Foundation Grants Numbers: 1339822 and 1527706.

Funders	Funder number
National Science Foundation	1339822, 1527706
National Science Foundation

Keywords

Linear algebra
Multicore processors
Multithreading
Numerical libraries
Parallel computing
Programming models
Runtime systems
Task scheduling

Access to Document

10.1007/s10766-016-0441-6

Cite this

@article{958889e576a74af5931dcdc7529253dd,

title = "Porting the PLASMA Numerical Library to the OpenMP Standard",

abstract = "PLASMA is a numerical library intended as a successor to LAPACK for solving problems in dense linear algebra on multicore processors. PLASMA relies on the QUARK scheduler for efficient multithreading of algorithms expressed in a serial fashion. QUARK is a superscalar scheduler and implements automatic parallelization by tracking data dependencies and resolving data hazards at runtime. Recently, this type of scheduling has been incorporated in the OpenMP standard, which allows to transition PLASMA from the proprietary solution offered by QUARK to the standard solution offered by OpenMP. This article studies the feasibility of such transition.",

keywords = "Linear algebra, Multicore processors, Multithreading, Numerical libraries, Parallel computing, Programming models, Runtime systems, Task scheduling",

author = "Asim YarKhan and Jakub Kurzak and Piotr Luszczek and Jack Dongarra",

note = "Publisher Copyright: {\textcopyright} 2016, Springer Science+Business Media New York.",

year = "2017",

month = jun,

day = "1",

doi = "10.1007/s10766-016-0441-6",

language = "English",

volume = "45",

pages = "612--633",

journal = "International Journal of Parallel Programming",

issn = "0885-7458",

publisher = "Springer",

number = "3",

}

TY - JOUR

T1 - Porting the PLASMA Numerical Library to the OpenMP Standard

AU - YarKhan, Asim

AU - Kurzak, Jakub

AU - Luszczek, Piotr

AU - Dongarra, Jack

PY - 2017/6/1

Y1 - 2017/6/1

N2 - PLASMA is a numerical library intended as a successor to LAPACK for solving problems in dense linear algebra on multicore processors. PLASMA relies on the QUARK scheduler for efficient multithreading of algorithms expressed in a serial fashion. QUARK is a superscalar scheduler and implements automatic parallelization by tracking data dependencies and resolving data hazards at runtime. Recently, this type of scheduling has been incorporated in the OpenMP standard, which allows to transition PLASMA from the proprietary solution offered by QUARK to the standard solution offered by OpenMP. This article studies the feasibility of such transition.

AB - PLASMA is a numerical library intended as a successor to LAPACK for solving problems in dense linear algebra on multicore processors. PLASMA relies on the QUARK scheduler for efficient multithreading of algorithms expressed in a serial fashion. QUARK is a superscalar scheduler and implements automatic parallelization by tracking data dependencies and resolving data hazards at runtime. Recently, this type of scheduling has been incorporated in the OpenMP standard, which allows to transition PLASMA from the proprietary solution offered by QUARK to the standard solution offered by OpenMP. This article studies the feasibility of such transition.

KW - Linear algebra

KW - Multicore processors

KW - Multithreading

KW - Numerical libraries

KW - Parallel computing

KW - Programming models

KW - Runtime systems

KW - Task scheduling

UR - http://www.scopus.com/inward/record.url?scp=84974853449&partnerID=8YFLogxK

U2 - 10.1007/s10766-016-0441-6

DO - 10.1007/s10766-016-0441-6

M3 - Article

AN - SCOPUS:84974853449

SN - 0885-7458

VL - 45

SP - 612

EP - 633

JO - International Journal of Parallel Programming

JF - International Journal of Parallel Programming

IS - 3

ER -

Porting the PLASMA Numerical Library to the OpenMP Standard

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this