Saved in:
Bibliographic Details
Main Authors: Magalhães, Philippe, Fresse, Virginie, Suffran, Benoît, Alata, Olivier
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.15930
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909854088560640
author Magalhães, Philippe
Fresse, Virginie
Suffran, Benoît
Alata, Olivier
author_facet Magalhães, Philippe
Fresse, Virginie
Suffran, Benoît
Alata, Olivier
contents Implementing convolutional neural networks (CNNs) on field-programmable gate arrays (FPGAs) has emerged as a promising alternative to GPUs, offering lower latency, greater power efficiency and greater flexibility. However, this development remains complex due to the hardware knowledge required and the long synthesis, placement and routing stages, which slow down design cycles and prevent rapid exploration of network configurations, making resource optimisation under severe constraints particularly challenging. This paper proposes a library of configurable convolution Blocks designed to optimize FPGA implementation and adapt to available resources. It also presents a methodological framework for developing mathematical models that predict FPGA resources utilization. The approach is validated by analyzing the correlation between the parameters, followed by error metrics. The results show that the designed blocks enable adaptation of convolution layers to hardware constraints, and that the models accurately predict resource consumption, providing a useful tool for FPGA selection and optimized CNN deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2510_15930
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Implémentation Efficiente de Fonctions de Convolution sur FPGA à l'Aide de Blocs Paramétrables et d'Approximations Polynomiales
Magalhães, Philippe
Fresse, Virginie
Suffran, Benoît
Alata, Olivier
Hardware Architecture
Artificial Intelligence
Neural and Evolutionary Computing
Implementing convolutional neural networks (CNNs) on field-programmable gate arrays (FPGAs) has emerged as a promising alternative to GPUs, offering lower latency, greater power efficiency and greater flexibility. However, this development remains complex due to the hardware knowledge required and the long synthesis, placement and routing stages, which slow down design cycles and prevent rapid exploration of network configurations, making resource optimisation under severe constraints particularly challenging. This paper proposes a library of configurable convolution Blocks designed to optimize FPGA implementation and adapt to available resources. It also presents a methodological framework for developing mathematical models that predict FPGA resources utilization. The approach is validated by analyzing the correlation between the parameters, followed by error metrics. The results show that the designed blocks enable adaptation of convolution layers to hardware constraints, and that the models accurately predict resource consumption, providing a useful tool for FPGA selection and optimized CNN deployment.
title Implémentation Efficiente de Fonctions de Convolution sur FPGA à l'Aide de Blocs Paramétrables et d'Approximations Polynomiales
topic Hardware Architecture
Artificial Intelligence
Neural and Evolutionary Computing
url https://arxiv.org/abs/2510.15930