Bind the Gap: Compiling Real Software to Hardware FFT Accelerators

Jackson Woodruff; Jordi Armengol Estapé; Sam Ainsworth; Michael F P O'Boyle

doi:10.1145/3519939.3523439

Bind the Gap: Compiling Real Software to Hardware FFT Accelerators

Jackson Woodruff, Jordi Armengol Estapé, Sam Ainsworth, Michael F P O'Boyle

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

Specialized hardware accelerators continue to be a source of performance improvement. However, such specialization comes at a programming price. The fundamental issue is that of a mismatch between the diversity of user code and the functionality of fixed hardware, limiting its wider uptake. Here we focus on a particular set of accelerators: those for Fast Fourier Transforms. We present FACC (Fourier ACcelerator Compiler), a novel approach to automatically map legacy code to Fourier Transform accelerators. It automatically generates drop-in replacement adapters using InputOutput (IO)-based program synthesis that bridge the gap between user code and accelerators. We apply FACC to unmodified GitHub C programs of varying complexity and compare against two existing approaches. We target FACC to a high-performance library, FFTW, and two hardware accelerators, the NXP PowerQuad and the Analog Devices FFTA, and demonstrate mean speedups of 9x, 17x and 27x respectively.

Original language	English
Title of host publication	Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation
Editors	Ranjit Jhala, Işil Dillig
Publisher	ACM
Pages	687-702
Number of pages	16
ISBN (Print)	978-1-4503-9265-5
DOIs	https://doi.org/10.1145/3519939.3523439
Publication status	Published - 9 Jun 2022
Event	43rd ACM SIGPLAN Conference on Programming Language Design and Implementation - San Diego, United States Duration: 15 Jun 2022 → 17 Jun 2022 Conference number: 43 https://pldi22.sigplan.org/

Conference

Conference	43rd ACM SIGPLAN Conference on Programming Language Design and Implementation
Abbreviated title	PLDI 2022
Country/Territory	United States
City	San Diego
Period	15/06/22 → 17/06/22
Internet address	https://pldi22.sigplan.org/

Keywords / Materials (for Non-textual outputs)

Hardware Accelerators
Program Synthesis
FFT

Access to Document

10.1145/3519939.3523439Licence: All Rights Reserved

Bind the Gap_WOODRUFF_DOA25022022_AFVAccepted author manuscript, 1.15 MBLicence: All Rights Reserved

https://dl.acm.org/doi/10.1145/3519939.3523439Licence: All Rights Reserved

Transmuter: A Reconfigurable Computer
Cole, M. & O'Boyle, M.
Non-EU other
6/08/18 → 20/01/23
Project: Research

Cite this

@inproceedings{ebe3a4c809da408ea1d6434fe5f6a3ac,

title = "Bind the Gap: Compiling Real Software to Hardware FFT Accelerators",

abstract = "Specialized hardware accelerators continue to be a source of performance improvement. However, such specialization comes at a programming price. The fundamental issue is that of a mismatch between the diversity of user code and the functionality of fixed hardware, limiting its wider uptake. Here we focus on a particular set of accelerators: those for Fast Fourier Transforms. We present FACC (Fourier ACcelerator Compiler), a novel approach to automatically map legacy code to Fourier Transform accelerators. It automatically generates drop-in replacement adapters using InputOutput (IO)-based program synthesis that bridge the gap between user code and accelerators. We apply FACC to unmodified GitHub C programs of varying complexity and compare against two existing approaches. We target FACC to a high-performance library, FFTW, and two hardware accelerators, the NXP PowerQuad and the Analog Devices FFTA, and demonstrate mean speedups of 9x, 17x and 27x respectively.",

keywords = "Hardware Accelerators, Program Synthesis, FFT",

author = "Jackson Woodruff and {Armengol Estap{\'e}}, Jordi and Sam Ainsworth and O'Boyle, {Michael F P}",

note = "Funding Information: Acknowledgements The material is partially based on research by Michael O{\textquoteright}Boyle which was sponsored by Defense Advanced Research Projects Agency (DARPA) under agreement number FA8650-18-2-7864. The views and conclusions contained herein are those of the authors and do not represent the official policies or endorsements, either expressed or implied, of DARPA or the U.S. Government. Publisher Copyright: {\textcopyright} 2022 ACM.; 43rd ACM SIGPLAN Conference on Programming Language Design and Implementation, PLDI 2022 ; Conference date: 15-06-2022 Through 17-06-2022",

year = "2022",

month = jun,

day = "9",

doi = "10.1145/3519939.3523439",

language = "English",

isbn = " 978-1-4503-9265-5",

pages = "687--702",

editor = "Ranjit Jhala and I{\c s}il Dillig",

booktitle = "Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation",

publisher = "ACM",

url = "https://pldi22.sigplan.org/",

}

Woodruff, J, Armengol Estapé, J, Ainsworth, S & O'Boyle, MFP 2022, Bind the Gap: Compiling Real Software to Hardware FFT Accelerators. in R Jhala & I Dillig (eds), Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation. ACM, pp. 687-702, 43rd ACM SIGPLAN Conference on Programming Language Design and Implementation, San Diego, California, United States, 15/06/22. https://doi.org/10.1145/3519939.3523439

Bind the Gap: Compiling Real Software to Hardware FFT Accelerators. / Woodruff, Jackson; Armengol Estapé, Jordi; Ainsworth, Sam et al.
Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation. ed. / Ranjit Jhala; Işil Dillig. ACM, 2022. p. 687-702.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Bind the Gap: Compiling Real Software to Hardware FFT Accelerators

AU - Woodruff, Jackson

AU - Armengol Estapé, Jordi

AU - Ainsworth, Sam

AU - O'Boyle, Michael F P

N1 - Conference code: 43

PY - 2022/6/9

Y1 - 2022/6/9

N2 - Specialized hardware accelerators continue to be a source of performance improvement. However, such specialization comes at a programming price. The fundamental issue is that of a mismatch between the diversity of user code and the functionality of fixed hardware, limiting its wider uptake. Here we focus on a particular set of accelerators: those for Fast Fourier Transforms. We present FACC (Fourier ACcelerator Compiler), a novel approach to automatically map legacy code to Fourier Transform accelerators. It automatically generates drop-in replacement adapters using InputOutput (IO)-based program synthesis that bridge the gap between user code and accelerators. We apply FACC to unmodified GitHub C programs of varying complexity and compare against two existing approaches. We target FACC to a high-performance library, FFTW, and two hardware accelerators, the NXP PowerQuad and the Analog Devices FFTA, and demonstrate mean speedups of 9x, 17x and 27x respectively.

AB - Specialized hardware accelerators continue to be a source of performance improvement. However, such specialization comes at a programming price. The fundamental issue is that of a mismatch between the diversity of user code and the functionality of fixed hardware, limiting its wider uptake. Here we focus on a particular set of accelerators: those for Fast Fourier Transforms. We present FACC (Fourier ACcelerator Compiler), a novel approach to automatically map legacy code to Fourier Transform accelerators. It automatically generates drop-in replacement adapters using InputOutput (IO)-based program synthesis that bridge the gap between user code and accelerators. We apply FACC to unmodified GitHub C programs of varying complexity and compare against two existing approaches. We target FACC to a high-performance library, FFTW, and two hardware accelerators, the NXP PowerQuad and the Analog Devices FFTA, and demonstrate mean speedups of 9x, 17x and 27x respectively.

KW - Hardware Accelerators

KW - Program Synthesis

KW - FFT

U2 - 10.1145/3519939.3523439

DO - 10.1145/3519939.3523439

M3 - Conference contribution

SN - 978-1-4503-9265-5

SP - 687

EP - 702

BT - Proceedings of the 43rd ACM SIGPLAN International Conference on Programming Language Design and Implementation

A2 - Jhala, Ranjit

A2 - Dillig, Işil

PB - ACM

T2 - 43rd ACM SIGPLAN Conference on Programming Language Design and Implementation

Y2 - 15 June 2022 through 17 June 2022

ER -

Bind the Gap: Compiling Real Software to Hardware FFT Accelerators

Abstract / Description of output

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Fingerprint

Projects

Transmuter: A Reconfigurable Computer

Cite this