Compiling APL to accelerate through a typed array intermediate language

Michael Budde; Martin Dybdal; Martin Elsman

doi:10.1145/2774959.2774966

Compiling APL to accelerate through a typed array intermediate language

Michael Budde, Martin Dybdal, Martin Elsman

Datalogisk Institut

1 Citationer (Scopus)

Abstract

We present an approach for compiling a rich subset of APL into dataparallel programs that can be executed on GPUs. The compiler is based on the AplTail compiler, which compiles APL programs into a typed array intermediate language, called TAIL [11]. We translate TAIL programs into Haskell source code, employing Accelerate [6], a Haskell-library for general purpose GPU-programming. We demonstrate the feasibility of the approach by presenting some encouraging results for a number of smaller benchmarks. We also outline some problems that we need to overcome in order for the approach to result in competitive code for larger benchmarks.

Originalsprog	Engelsk
Titel	Proceedings of the 2nd ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming
Antal sider	6
Forlag	Association for Computing Machinery
Publikationsdato	13 jun. 2015
Sider	13-18
ISBN (Trykt)	978-1-4503-3584-3
DOI	https://doi.org/10.1145/2774959.2774966
Status	Udgivet - 13 jun. 2015
Begivenhed	ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming 2015 - Portland, USA Varighed: 13 jun. 2015 → 17 jun. 2015 Konferencens nummer: 2

Konference

Konference	ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming 2015
Nummer	2
Land/Område	USA
By	Portland
Periode	13/06/2015 → 17/06/2015

Emneord

APL compilation, Data parallelism, GPGPU

Adgang til dokumentet

10.1145/2774959.2774966

Citationsformater

Compiling APL to accelerate through a typed array intermediate language. / Budde, Michael; Dybdal, Martin; Elsman, Martin.
Proceedings of the 2nd ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming. Association for Computing Machinery, 2015. s. 13-18.

Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › peer review

Budde, M, Dybdal, M & Elsman, M 2015, Compiling APL to accelerate through a typed array intermediate language. i Proceedings of the 2nd ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming. Association for Computing Machinery, s. 13-18, ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming 2015, Portland, USA, 13/06/2015. https://doi.org/10.1145/2774959.2774966

@inproceedings{7d540667d65d47edaa84615a261a0f1a,

title = "Compiling APL to accelerate through a typed array intermediate language",

abstract = "We present an approach for compiling a rich subset of APL into dataparallel programs that can be executed on GPUs. The compiler is based on the AplTail compiler, which compiles APL programs into a typed array intermediate language, called TAIL [11]. We translate TAIL programs into Haskell source code, employing Accelerate [6], a Haskell-library for general purpose GPU-programming. We demonstrate the feasibility of the approach by presenting some encouraging results for a number of smaller benchmarks. We also outline some problems that we need to overcome in order for the approach to result in competitive code for larger benchmarks.",

keywords = "APL compilation, Data parallelism, GPGPU",

author = "Michael Budde and Martin Dybdal and Martin Elsman",

year = "2015",

month = jun,

day = "13",

doi = "10.1145/2774959.2774966",

language = "English",

isbn = "978-1-4503-3584-3",

pages = "13--18",

booktitle = "Proceedings of the 2nd ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming",

publisher = "Association for Computing Machinery",

note = "ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming 2015 ; Conference date: 13-06-2015 Through 17-06-2015",

}

TY - GEN

T1 - Compiling APL to accelerate through a typed array intermediate language

AU - Budde, Michael

AU - Dybdal, Martin

AU - Elsman, Martin

N1 - Conference code: 2

PY - 2015/6/13

Y1 - 2015/6/13

N2 - We present an approach for compiling a rich subset of APL into dataparallel programs that can be executed on GPUs. The compiler is based on the AplTail compiler, which compiles APL programs into a typed array intermediate language, called TAIL [11]. We translate TAIL programs into Haskell source code, employing Accelerate [6], a Haskell-library for general purpose GPU-programming. We demonstrate the feasibility of the approach by presenting some encouraging results for a number of smaller benchmarks. We also outline some problems that we need to overcome in order for the approach to result in competitive code for larger benchmarks.

AB - We present an approach for compiling a rich subset of APL into dataparallel programs that can be executed on GPUs. The compiler is based on the AplTail compiler, which compiles APL programs into a typed array intermediate language, called TAIL [11]. We translate TAIL programs into Haskell source code, employing Accelerate [6], a Haskell-library for general purpose GPU-programming. We demonstrate the feasibility of the approach by presenting some encouraging results for a number of smaller benchmarks. We also outline some problems that we need to overcome in order for the approach to result in competitive code for larger benchmarks.

KW - APL compilation, Data parallelism, GPGPU

U2 - 10.1145/2774959.2774966

DO - 10.1145/2774959.2774966

M3 - Article in proceedings

SN - 978-1-4503-3584-3

SP - 13

EP - 18

BT - Proceedings of the 2nd ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming

PB - Association for Computing Machinery

T2 - ACM SIGPLAN International Workshop on Libraries, Languages, and Compilers for Array Programming 2015

Y2 - 13 June 2015 through 17 June 2015

ER -

Compiling APL to accelerate through a typed array intermediate language

Abstract

Konference

Emneord

Adgang til dokumentet

Fingeraftryk

Citationsformater