Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

University of Guelph

$38,000.00

Enhancing the FPGA CAD Flow via Advanced Deep Learning Probes and Framework

EF 2024-25

Résumé
Field Programmable Gate Arrays (FPGAs) are integrated circuits which can be reconfigured to implement new designs. The Computer-Aided Design (CAD) flow for FPGAs is a several stage process which configures the FPGA into the desired circuit. FPGAs were invented and introduced almost 40 years ago. FPGAs today are enablers for many critical applications including communication systems, cryptography, deep learning, autonomous driving, computer vision and many more crucial technologies. Using FPGAs to accelerate complex application delivers enormous performance per Watt compared to Graphical Processing Units (GPUs) and Central Processing Units (CPUs). However, many challenges face designers today when using FPGAs for such complex applications. Placement and routing are important and challenging steps in the FPGA CAD process. In recent years designers have demanded increased performance as FPGA circuits have grown in size and complexity with logic blocks numbering in the millions. Existing FPGA CAD tools struggle to support these designs, with long compile times and many costly iterations required to reach design closure. While the literature has produced many different placement flows, there is no single flow that exhibits superior performance on all possible circuits. Due to variations in circuit features, as well as the different optimization strategies employed by different placers, flows that perform well on some circuits may perform poorly on others. Also, FPGA CAD algorithm parameters controlling place and route efforts can significantly affect timing results of the circuits. A correct selection of these parameters is challenging. Accordingly, there is a growing need to reduce FPGA compilation times, and further improve the quality of the solutions being produced with the help of machine learning. We anticipate that this research proposal will yield novel ML/Algorithmic FPGA CAD tools that would enable designs that are too large or complex to undertake otherwise, shorter design time and improve product quality and produce near optimal solutions. The proposed research is multi-faceted and capable of concurrent efforts. The novelty and expected significance of the proposed framework can be summarized as following: The proposed machine learning framework for congestion management and algorithm selection and parameter tuning will significantly improve the quality of solutions produced and at the same time reduce the CPU time thus enhancing the compile time of reconfigurable systems. The overall significance of this work will be to provide Canadian industry with efficient, smart and scalable FPGA placement and routing tools that can produce high-quality near optimal solutions, while avoiding excessively long compile times. Finally, the proposed machine learning tools developed can be easily ported and used in more complex ASIC design platforms.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Ontario, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2023-03367
Name nom
Areibi, Shawki
Department département
School of Engineering
Organization id
24
Institution établissement
University of Guelph
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
38000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1510
Committee name en
Electrical and Computer Engineering
Committee name fr
Génie électrique et informatique
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2500
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Electrical and electronic engineering
Research subject fr
Génie électrique et électronique
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Enhancing the FPGA CAD Flow via Advanced Deep Learning Probes and Framework
Keywords
vlsi design; field programmable gate arrays (fpgas); electronic design automation; optimization (exact mathematical/meta heuristics); machine learning; deep learning; fpga placement and routing; high performance computing; algorithm design and analysis; meta heuristics
Application summary
Field Programmable Gate Arrays (FPGAs) are integrated circuits which can be reconfigured to implement new designs. The Computer-Aided Design (CAD) flow for FPGAs is a several stage process which configures the FPGA into the desired circuit. FPGAs were invented and introduced almost 40 years ago. FPGAs today are enablers for many critical applications including communication systems, cryptography, deep learning, autonomous driving, computer vision and many more crucial technologies. Using FPGAs to accelerate complex application delivers enormous performance per Watt compared to Graphical Processing Units (GPUs) and Central Processing Units (CPUs). However, many challenges face designers today when using FPGAs for such complex applications. Placement and routing are important and challenging steps in the FPGA CAD process. In recent years designers have demanded increased performance as FPGA circuits have grown in size and complexity with logic blocks numbering in the millions. Existing FPGA CAD tools struggle to support these designs, with long compile times and many costly iterations required to reach design closure. While the literature has produced many different placement flows, there is no single flow that exhibits superior performance on all possible circuits. Due to variations in circuit features, as well as the different optimization strategies employed by different placers, flows that perform well on some circuits may perform poorly on others. Also, FPGA CAD algorithm parameters controlling place and route efforts can significantly affect timing results of the circuits. A correct selection of these parameters is challenging. Accordingly, there is a growing need to reduce FPGA compilation times, and further improve the quality of the solutions being produced with the help of machine learning. We anticipate that this research proposal will yield novel ML/Algorithmic FPGA CAD tools that would enable designs that are too large or complex to undertake otherwise, shorter design time and improve product quality and produce near optimal solutions. The proposed research is multi-faceted and capable of concurrent efforts. The novelty and expected significance of the proposed framework can be summarized as following: The proposed machine learning framework for congestion management and algorithm selection and parameter tuning will significantly improve the quality of solutions produced and at the same time reduce the CPU time thus enhancing the compile time of reconfigurable systems. The overall significance of this work will be to provide Canadian industry with efficient, smart and scalable FPGA placement and routing tools that can produce high-quality near optimal solutions, while avoiding excessively long compile times. Finally, the proposed machine learning tools developed can be easily ported and used in more complex ASIC design platforms.