A Study on Convolution Operator Using Half Precision Floating Point Numbers on GPU for Radioastronomy Deconvolution - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

A Study on Convolution Operator Using Half Precision Floating Point Numbers on GPU for Radioastronomy Deconvolution

Mickael Seznec
Nicolas Gac
André Ferrari

Résumé

The use of IEEE 754-2008 half-precision floating-point numbers is an emerging trend in Graphical Processing Units' architecture. Being such a compact way of representing data, its use may speed up programs by reducing the memory bandwidth usage and allowing hardware designers to fit more computing units within the same die space. In this paper, we highlight the acceleration offered by the use of half floating-point numbers over different implementations of the same operation, a 2D convolution. We show that even though it may lead up to a significant speed-up, the degradation brought by this new format is not always negligible. Then, we choose a deconvolution problem inspired by the SKA radio-telescope processing pipeline to show how half floats behave in a more complex application.
Fichier principal
Vignette du fichier
A_Study_on_Convolution_Using_Half_Precision_Floating_Point_Numbers_on_GPU_for_Radio_Astronomy_Deconvolution.pdf (1.26 Mo) Télécharger le fichier
Presentation__A_Study_on_Convolution.pdf (1.52 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-01837982 , version 1 (13-09-2018)

Identifiants

Citer

Mickael Seznec, Nicolas Gac, André Ferrari, François Orieux. A Study on Convolution Operator Using Half Precision Floating Point Numbers on GPU for Radioastronomy Deconvolution. IEEE International Workshop on Signal Processing Systems (SiPS 2018), Oct 2018, Le Cap, South Africa. ⟨10.1109/sips.2018.8598342⟩. ⟨hal-01837982⟩
242 Consultations
734 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More