FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler

Tan Nguyen; Swathi Gurumani; Kyle Rupnow; Deming Chen

doi:10.1145/2847263.2847344

FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler

Tan Nguyen, Swathi Gurumani, Kyle Rupnow, Deming Chen

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

Abstract

Throughput oriented high level synthesis allows efficient de- sign and optimization using parallel input languages. Par- allel languages offer the benefit of parallelism extraction at multiple levels of granularity, offering effective design space exploration to select efficient single core implementations, and easy scaling of parallelism through multiple core instan- tiations. However, study of high level synthesis for paral- lel languages has concentrated on optimization of core and on-chip communications, while neglecting platform integra- tion, which can have a significant impact on achieved per- formance. In this paper, we create an automated flow to perform efficient platform integration for an existing CUDA- to-RTL throughput oriented HLS, and we open source the FCUDA tool, platform integration, and benchmark appli- cations. We demonstrate platform integration of 16 bench- marks on two Zynq-based systems in bare-metal and OS mode. We study implementation optimization for platform integration, compare to an embedded GPU (Tegra TK1) and verify designs on a Zedboard Zynq 7020 (bare-metal) and Omnitek Zynq 7045 (OS).

Original language	English (US)
Title of host publication	FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays
Publisher	Association for Computing Machinery
Pages	5-14
Number of pages	10
ISBN (Electronic)	9781450338561
DOIs	https://doi.org/10.1145/2847263.2847344
State	Published - Feb 21 2016
Event	2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2016 - Monterey, United States Duration: Feb 21 2016 → Feb 23 2016

Publication series

Name	FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays

Other

Other	2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2016
Country/Territory	United States
City	Monterey
Period	2/21/16 → 2/23/16

ASJC Scopus subject areas

Hardware and Architecture
Electrical and Electronic Engineering

Online availability

10.1145/2847263.2847344

Library availability

Discover UIUC Full Text

Cite this

Nguyen, T., Gurumani, S., Rupnow, K., & Chen, D. (2016). FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler. In FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (pp. 5-14). (FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays). Association for Computing Machinery. https://doi.org/10.1145/2847263.2847344

FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler. / Nguyen, Tan; Gurumani, Swathi; Rupnow, Kyle et al.
FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays. Association for Computing Machinery, 2016. p. 5-14 (FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays).

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

Nguyen, T, Gurumani, S, Rupnow, K & Chen, D 2016, FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler. in FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays. FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, Association for Computing Machinery, pp. 5-14, 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2016, Monterey, United States, 2/21/16. https://doi.org/10.1145/2847263.2847344

Nguyen T, Gurumani S, Rupnow K, Chen D. FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler. In FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays. Association for Computing Machinery. 2016. p. 5-14. (FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays). doi: 10.1145/2847263.2847344

Nguyen, Tan ; Gurumani, Swathi ; Rupnow, Kyle et al. / FCUDA-SoC : Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler. FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays. Association for Computing Machinery, 2016. pp. 5-14 (FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays).

@inproceedings{7623a38d129f40889b5be8fa9f75173b,

title = "FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler",

abstract = "Throughput oriented high level synthesis allows efficient de- sign and optimization using parallel input languages. Par- allel languages offer the benefit of parallelism extraction at multiple levels of granularity, offering effective design space exploration to select efficient single core implementations, and easy scaling of parallelism through multiple core instan- tiations. However, study of high level synthesis for paral- lel languages has concentrated on optimization of core and on-chip communications, while neglecting platform integra- tion, which can have a significant impact on achieved per- formance. In this paper, we create an automated flow to perform efficient platform integration for an existing CUDA- to-RTL throughput oriented HLS, and we open source the FCUDA tool, platform integration, and benchmark appli- cations. We demonstrate platform integration of 16 bench- marks on two Zynq-based systems in bare-metal and OS mode. We study implementation optimization for platform integration, compare to an embedded GPU (Tegra TK1) and verify designs on a Zedboard Zynq 7020 (bare-metal) and Omnitek Zynq 7045 (OS).",

author = "Tan Nguyen and Swathi Gurumani and Kyle Rupnow and Deming Chen",

note = "Publisher Copyright: {\textcopyright} 2016 ACM.; 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2016 ; Conference date: 21-02-2016 Through 23-02-2016",

year = "2016",

month = feb,

day = "21",

doi = "10.1145/2847263.2847344",

language = "English (US)",

series = "FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays",

publisher = "Association for Computing Machinery",

pages = "5--14",

booktitle = "FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays",

address = "United States",

}

TY - GEN

T1 - FCUDA-SoC

T2 - 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays, FPGA 2016

AU - Nguyen, Tan

AU - Gurumani, Swathi

AU - Rupnow, Kyle

AU - Chen, Deming

PY - 2016/2/21

Y1 - 2016/2/21

N2 - Throughput oriented high level synthesis allows efficient de- sign and optimization using parallel input languages. Par- allel languages offer the benefit of parallelism extraction at multiple levels of granularity, offering effective design space exploration to select efficient single core implementations, and easy scaling of parallelism through multiple core instan- tiations. However, study of high level synthesis for paral- lel languages has concentrated on optimization of core and on-chip communications, while neglecting platform integra- tion, which can have a significant impact on achieved per- formance. In this paper, we create an automated flow to perform efficient platform integration for an existing CUDA- to-RTL throughput oriented HLS, and we open source the FCUDA tool, platform integration, and benchmark appli- cations. We demonstrate platform integration of 16 bench- marks on two Zynq-based systems in bare-metal and OS mode. We study implementation optimization for platform integration, compare to an embedded GPU (Tegra TK1) and verify designs on a Zedboard Zynq 7020 (bare-metal) and Omnitek Zynq 7045 (OS).

AB - Throughput oriented high level synthesis allows efficient de- sign and optimization using parallel input languages. Par- allel languages offer the benefit of parallelism extraction at multiple levels of granularity, offering effective design space exploration to select efficient single core implementations, and easy scaling of parallelism through multiple core instan- tiations. However, study of high level synthesis for paral- lel languages has concentrated on optimization of core and on-chip communications, while neglecting platform integra- tion, which can have a significant impact on achieved per- formance. In this paper, we create an automated flow to perform efficient platform integration for an existing CUDA- to-RTL throughput oriented HLS, and we open source the FCUDA tool, platform integration, and benchmark appli- cations. We demonstrate platform integration of 16 bench- marks on two Zynq-based systems in bare-metal and OS mode. We study implementation optimization for platform integration, compare to an embedded GPU (Tegra TK1) and verify designs on a Zedboard Zynq 7020 (bare-metal) and Omnitek Zynq 7045 (OS).

UR - http://www.scopus.com/inward/record.url?scp=84966565331&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84966565331&partnerID=8YFLogxK

U2 - 10.1145/2847263.2847344

DO - 10.1145/2847263.2847344

M3 - Conference contribution

AN - SCOPUS:84966565331

T3 - FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays

SP - 5

EP - 14

BT - FPGA 2016 - Proceedings of the 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays

PB - Association for Computing Machinery

Y2 - 21 February 2016 through 23 February 2016

ER -

FCUDA-SoC: Platform integration for field-programmable SoC with the CUDA-to-FPGA compiler

Abstract

Publication series

Other

ASJC Scopus subject areas

Online availability

Library availability

Related links

Fingerprint

Cite this