The Challenge of Multi-Operand Adders in CNNs on FPGAs: How Not to Solve It! - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

The Challenge of Multi-Operand Adders in CNNs on FPGAs: How Not to Solve It!

Kamel Abdelouahab
François Berry

Résumé

Convolutional Neural Networks (CNNs) are computationally intensive algorithms that currently require dedicated hardware to be executed. In the case of FPGA-Based accelerators, we point-out in this work the challenge of Multi-Operand Adders (MOAs) and their high resource utilization in an FPGA implementation of a CNN. To address this challenge, two optimization strategies, that rely on time-multiplexing and approximate computing, are investigated. At first glance, the two strategies looked promising to reduce the footprint of a given architectural mapping, but when synthesized on the device, none of them gave the expected results. Experimental sections analyze the reasons of these unexpected results.
Fichier principal
Vignette du fichier
1807.00217.pdf (681.76 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01902952 , version 1 (18-01-2019)

Identifiants

Citer

Kamel Abdelouahab, Maxime Pelcat, François Berry. The Challenge of Multi-Operand Adders in CNNs on FPGAs: How Not to Solve It!. 18th International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation (SAMOS '18), Jul 2018, Pythagorion, Greece. pp.157-160, ⟨10.1145/3229631.3235024⟩. ⟨hal-01902952⟩
95 Consultations
188 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More