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

Abstract : 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.
Type de document :
Communication dans un congrès
IEEE International Workshop on Signal Processing Systems (SiPS), Oct 2018, Cape Town, South Africa
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01837982
Contributeur : Nicolas Gac <>
Soumis le : jeudi 13 septembre 2018 - 15:49:40
Dernière modification le : mardi 18 septembre 2018 - 09:42:38

Fichier

A_Study_on_Convolution_Using_H...
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01837982, version 1

Citation

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), Oct 2018, Cape Town, South Africa. 〈hal-01837982〉

Partager

Métriques

Consultations de la notice

131

Téléchargements de fichiers

38