Heterogeneous systems: an approach using FPGAs as accelerators

Maria Villarroya Gaudo, University of Zaragoza, Spain

Abstract: In this talk I will present the work done at the Computer Architecture Group of the University of Zaragoza (GaZ) in recent years. Our research focus is to increase the performance of the computing systems, while being aware of the energy consumption. During this talk, I will present some results that allow to do more accessible the use of FPGAs as accelerators.

In the talk three different approaches will be faced: first at a low level how an analytical model can help in the execution on FPGA, particularly for memory-bound applications. Next results about standard and portable programming will be shown and finally the execution system and load balancing on a full heterogeneous system composed of a CPU, a GPU, and a FPGA.

The experimental part of this work has been carried out by Mara Angelica Davila-Guzman, former PhD Student of the group, who was supervised by the speaker, Maria Villarroya Gaudo, together with Darío Suarez-Gracia.