Semi-automatic tool to ease the creation and optimization of GPU programs

Jacob Jepsen

doi:10.1109/ICPPW.2014.36

Semi-automatic tool to ease the creation and optimization of GPU programs

Jacob Jepsen^*

^*Corresponding author af dette arbejde

Datalogisk Institut

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

Abstract

We present a tool that reduces the development time of GPU-executable code. We implement a catalogue of common optimizations specific to the GPU architecture. Through the tool, the programmer can semi-automatically transform a computationally-intensive code section into GPU-executable form and apply optimizations thereto. Based on experiments, the code generated by the tool can be 3-256X faster than code generated by an OpenACC compiler, 4-37X faster than optimized CPU code, and attain up to 25% of peak performance of the GPU. We found that by using pattern-matching rules, many of the transformations can be performed automatically, which makes the tool usable for both novices and experts in GPU programming.

Originalsprog	Engelsk
Titel	Proceedings of the 43rd International Conference on Parallel Processing Workshops : ICPPW 2014
Antal sider	10
Forlag	IEEE
Publikationsdato	2014
Sider	196-205
ISBN (Trykt)	978-1-4799-5615-9
DOI	https://doi.org/10.1109/ICPPW.2014.36
Status	Udgivet - 2014
Begivenhed	43rd International Conference on Parallel Processing Workshops, ICPPW 2014 - Minneapolis, USA Varighed: 9 sep. 2014 → 12 sep. 2014

Konference

Konference	43rd International Conference on Parallel Processing Workshops, ICPPW 2014
Land/Område	USA
By	Minneapolis
Periode	09/09/2014 → 12/09/2014

Adgang til dokumentet

10.1109/ICPPW.2014.36

Biblioteksadgang?

Tjek tilgængelighed hos det Kgl. Bibliotek

Citationsformater

@inproceedings{fb142d0db43f4ffab201955a33a54902,

title = "Semi-automatic tool to ease the creation and optimization of GPU programs",

abstract = "We present a tool that reduces the development time of GPU-executable code. We implement a catalogue of common optimizations specific to the GPU architecture. Through the tool, the programmer can semi-automatically transform a computationally-intensive code section into GPU-executable form and apply optimizations thereto. Based on experiments, the code generated by the tool can be 3-256X faster than code generated by an OpenACC compiler, 4-37X faster than optimized CPU code, and attain up to 25\% of peak performance of the GPU. We found that by using pattern-matching rules, many of the transformations can be performed automatically, which makes the tool usable for both novices and experts in GPU programming.",

author = "Jacob Jepsen",

year = "2014",

doi = "10.1109/ICPPW.2014.36",

language = "English",

isbn = "978-1-4799-5615-9",

pages = "196--205",

booktitle = "Proceedings of the 43rd International Conference on Parallel Processing Workshops",

publisher = "IEEE",

note = "43rd International Conference on Parallel Processing Workshops, ICPPW 2014 ; Conference date: 09-09-2014 Through 12-09-2014",

}

TY - GEN

T1 - Semi-automatic tool to ease the creation and optimization of GPU programs

AU - Jepsen, Jacob

PY - 2014

Y1 - 2014

N2 - We present a tool that reduces the development time of GPU-executable code. We implement a catalogue of common optimizations specific to the GPU architecture. Through the tool, the programmer can semi-automatically transform a computationally-intensive code section into GPU-executable form and apply optimizations thereto. Based on experiments, the code generated by the tool can be 3-256X faster than code generated by an OpenACC compiler, 4-37X faster than optimized CPU code, and attain up to 25% of peak performance of the GPU. We found that by using pattern-matching rules, many of the transformations can be performed automatically, which makes the tool usable for both novices and experts in GPU programming.

AB - We present a tool that reduces the development time of GPU-executable code. We implement a catalogue of common optimizations specific to the GPU architecture. Through the tool, the programmer can semi-automatically transform a computationally-intensive code section into GPU-executable form and apply optimizations thereto. Based on experiments, the code generated by the tool can be 3-256X faster than code generated by an OpenACC compiler, 4-37X faster than optimized CPU code, and attain up to 25% of peak performance of the GPU. We found that by using pattern-matching rules, many of the transformations can be performed automatically, which makes the tool usable for both novices and experts in GPU programming.

U2 - 10.1109/ICPPW.2014.36

DO - 10.1109/ICPPW.2014.36

M3 - Article in proceedings

SN - 978-1-4799-5615-9

SP - 196

EP - 205

BT - Proceedings of the 43rd International Conference on Parallel Processing Workshops

PB - IEEE

T2 - 43rd International Conference on Parallel Processing Workshops, ICPPW 2014

Y2 - 9 September 2014 through 12 September 2014

ER -