High level synthesis of stereo matching: Productivity, performance, and software constraints

Kyle Rupnow, Yun Liang, Yinan Li, Dongbo Min, Minh N Do, Deming Chen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

FPGAs are an attractive platform for applications with high computation demand and low energy consumption requirements. However, design effort for FPGA implementations remains high often an order of magnitude larger than design effort using high level languages. Instead of this time-consuming process, high level synthesis (HLS) tools generate hardware implementations from high level languages (HLL) such as C/C/SystemC. Such tools reduce design effort: high level descriptions are more compact and less error prone. HLS tools promise hardware development abstracted from software designer knowledge of the implementation platform. In this paper, we examine several implementations of stereo matching, an active area of computer vision research that uses techniques also common for image de-noising, image retrieval, feature matching and face recognition. We present an unbiased evaluation of the suitability of using HLS for typical stereo matching software, usability and productivity of AutoPilot (a state of the art HLS tool), and the performance of designs produced by AutoPilot. Based on our study, we provide guidelines for software design, limitations of mapping general purpose software to hardware using HLS, and future directions for HLS tool development. For the stereo matching algorithms, we demonstrate between 3.5X and 67.9X speedup over software (but less than achievable by manual RTL design) with a five-fold reduction in design effort vs. manual hardware design.

Original languageEnglish (US)
Title of host publication2011 International Conference on Field-Programmable Technology, FPT 2011
DOIs
StatePublished - Dec 1 2011
Event2011 International Conference on Field-Programmable Technology, FPT 2011 - New Delhi, India
Duration: Dec 12 2011Dec 14 2011

Publication series

Name2011 International Conference on Field-Programmable Technology, FPT 2011

Other

Other2011 International Conference on Field-Programmable Technology, FPT 2011
CountryIndia
CityNew Delhi
Period12/12/1112/14/11

Fingerprint

High-level Synthesis
Stereo Matching
Productivity
Software
Autopilot
Hardware
High level languages
Field programmable gate arrays (FPGA)
SystemC
Feature Matching
FPGA Implementation
Hardware Design
Image Denoising
Software Design
Hardware Implementation
Image Retrieval
Image denoising
Matching Algorithm
Face Recognition
Computer Vision

ASJC Scopus subject areas

  • Computational Mathematics

Cite this

Rupnow, K., Liang, Y., Li, Y., Min, D., Do, M. N., & Chen, D. (2011). High level synthesis of stereo matching: Productivity, performance, and software constraints. In 2011 International Conference on Field-Programmable Technology, FPT 2011 [6132716] (2011 International Conference on Field-Programmable Technology, FPT 2011). https://doi.org/10.1109/FPT.2011.6132716

High level synthesis of stereo matching : Productivity, performance, and software constraints. / Rupnow, Kyle; Liang, Yun; Li, Yinan; Min, Dongbo; Do, Minh N; Chen, Deming.

2011 International Conference on Field-Programmable Technology, FPT 2011. 2011. 6132716 (2011 International Conference on Field-Programmable Technology, FPT 2011).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Rupnow, K, Liang, Y, Li, Y, Min, D, Do, MN & Chen, D 2011, High level synthesis of stereo matching: Productivity, performance, and software constraints. in 2011 International Conference on Field-Programmable Technology, FPT 2011., 6132716, 2011 International Conference on Field-Programmable Technology, FPT 2011, 2011 International Conference on Field-Programmable Technology, FPT 2011, New Delhi, India, 12/12/11. https://doi.org/10.1109/FPT.2011.6132716
Rupnow K, Liang Y, Li Y, Min D, Do MN, Chen D. High level synthesis of stereo matching: Productivity, performance, and software constraints. In 2011 International Conference on Field-Programmable Technology, FPT 2011. 2011. 6132716. (2011 International Conference on Field-Programmable Technology, FPT 2011). https://doi.org/10.1109/FPT.2011.6132716
Rupnow, Kyle ; Liang, Yun ; Li, Yinan ; Min, Dongbo ; Do, Minh N ; Chen, Deming. / High level synthesis of stereo matching : Productivity, performance, and software constraints. 2011 International Conference on Field-Programmable Technology, FPT 2011. 2011. (2011 International Conference on Field-Programmable Technology, FPT 2011).
@inproceedings{869dace308764feaaf7ae16cec07370e,
title = "High level synthesis of stereo matching: Productivity, performance, and software constraints",
abstract = "FPGAs are an attractive platform for applications with high computation demand and low energy consumption requirements. However, design effort for FPGA implementations remains high often an order of magnitude larger than design effort using high level languages. Instead of this time-consuming process, high level synthesis (HLS) tools generate hardware implementations from high level languages (HLL) such as C/C/SystemC. Such tools reduce design effort: high level descriptions are more compact and less error prone. HLS tools promise hardware development abstracted from software designer knowledge of the implementation platform. In this paper, we examine several implementations of stereo matching, an active area of computer vision research that uses techniques also common for image de-noising, image retrieval, feature matching and face recognition. We present an unbiased evaluation of the suitability of using HLS for typical stereo matching software, usability and productivity of AutoPilot (a state of the art HLS tool), and the performance of designs produced by AutoPilot. Based on our study, we provide guidelines for software design, limitations of mapping general purpose software to hardware using HLS, and future directions for HLS tool development. For the stereo matching algorithms, we demonstrate between 3.5X and 67.9X speedup over software (but less than achievable by manual RTL design) with a five-fold reduction in design effort vs. manual hardware design.",
author = "Kyle Rupnow and Yun Liang and Yinan Li and Dongbo Min and Do, {Minh N} and Deming Chen",
year = "2011",
month = "12",
day = "1",
doi = "10.1109/FPT.2011.6132716",
language = "English (US)",
isbn = "9781457717406",
series = "2011 International Conference on Field-Programmable Technology, FPT 2011",
booktitle = "2011 International Conference on Field-Programmable Technology, FPT 2011",

}

TY - GEN

T1 - High level synthesis of stereo matching

T2 - Productivity, performance, and software constraints

AU - Rupnow, Kyle

AU - Liang, Yun

AU - Li, Yinan

AU - Min, Dongbo

AU - Do, Minh N

AU - Chen, Deming

PY - 2011/12/1

Y1 - 2011/12/1

N2 - FPGAs are an attractive platform for applications with high computation demand and low energy consumption requirements. However, design effort for FPGA implementations remains high often an order of magnitude larger than design effort using high level languages. Instead of this time-consuming process, high level synthesis (HLS) tools generate hardware implementations from high level languages (HLL) such as C/C/SystemC. Such tools reduce design effort: high level descriptions are more compact and less error prone. HLS tools promise hardware development abstracted from software designer knowledge of the implementation platform. In this paper, we examine several implementations of stereo matching, an active area of computer vision research that uses techniques also common for image de-noising, image retrieval, feature matching and face recognition. We present an unbiased evaluation of the suitability of using HLS for typical stereo matching software, usability and productivity of AutoPilot (a state of the art HLS tool), and the performance of designs produced by AutoPilot. Based on our study, we provide guidelines for software design, limitations of mapping general purpose software to hardware using HLS, and future directions for HLS tool development. For the stereo matching algorithms, we demonstrate between 3.5X and 67.9X speedup over software (but less than achievable by manual RTL design) with a five-fold reduction in design effort vs. manual hardware design.

AB - FPGAs are an attractive platform for applications with high computation demand and low energy consumption requirements. However, design effort for FPGA implementations remains high often an order of magnitude larger than design effort using high level languages. Instead of this time-consuming process, high level synthesis (HLS) tools generate hardware implementations from high level languages (HLL) such as C/C/SystemC. Such tools reduce design effort: high level descriptions are more compact and less error prone. HLS tools promise hardware development abstracted from software designer knowledge of the implementation platform. In this paper, we examine several implementations of stereo matching, an active area of computer vision research that uses techniques also common for image de-noising, image retrieval, feature matching and face recognition. We present an unbiased evaluation of the suitability of using HLS for typical stereo matching software, usability and productivity of AutoPilot (a state of the art HLS tool), and the performance of designs produced by AutoPilot. Based on our study, we provide guidelines for software design, limitations of mapping general purpose software to hardware using HLS, and future directions for HLS tool development. For the stereo matching algorithms, we demonstrate between 3.5X and 67.9X speedup over software (but less than achievable by manual RTL design) with a five-fold reduction in design effort vs. manual hardware design.

UR - http://www.scopus.com/inward/record.url?scp=84863182412&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863182412&partnerID=8YFLogxK

U2 - 10.1109/FPT.2011.6132716

DO - 10.1109/FPT.2011.6132716

M3 - Conference contribution

AN - SCOPUS:84863182412

SN - 9781457717406

T3 - 2011 International Conference on Field-Programmable Technology, FPT 2011

BT - 2011 International Conference on Field-Programmable Technology, FPT 2011

ER -