Seminario di analisi numerica
ore
10:00
presso Seminario II
ABSTRACT
The ability of the modern graphics processing units (GPUs) to operate on
large data can be exploited to perform computations in High Performance
Computing (HPC) applications.
Given the relevant number of libraries already available,
little effort is required to parallelize numerical analysis approaches.
We briefly describe NVIDIA GPU architecture, available software and
programming tools. Fast Fourier Transform (FFT) speed-ups will be
presented as an application example.