Skip to topic | Skip to bottom


Home

Main
Main.DenisDeLeeuwr1.1 - 08 Oct 2007 - 12:01 - GerdGross

Start of topic | Skip to actions

-- GerdGross - 08 Oct 2007

Denis De Leeuw

Mapping Algorithms to Silicon

Abstract

FPGA-based Custom Computing Machines are application-specific computing devices that can outperform conventional software implementations of algorithms by several orders of magnitude. Developing these CCMs is a time consuming creative process that requires expert knowledge and experience. This thesis describes a systematic approach to the synthesis of FPGA-based CCM designs from conventional algorithm descriptions, based on common design patterns found in efficient hand-crafted CCM designs. It presents the imperative programming language ‘Julia’ that is based on this design method. The compilation process of Julia is described, and static analysis techniques are presented to predict the runtime and circuit growth of Julia programs. Performance experiments are presented for a number of sequential and parallel algorithms, which show that Julia is capable of delivering the expected speedups in practice.


You are here: Main > MastersProjects > PastAndCurrentMScProjects > DenisDeLeeuw

to top

Copyright © 2003-2017, Software Engineering Research Group, Delft University of Technology, The Netherlands