A Study on Convolution using Half-Precision Floating-Point Numbers on GPU for Radio Astronomy Deconvolution

摘要

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.