Functions in Reconfigurable Hardware for Matrix and Signal Processing Operations in MATLAB ∗

Reconfigurable or adaptive computing is an emerging area of computer science, which promises supercomputing performance from small , relatively low cost systems composed of re-programmable hardware such as field-programmable gate arrays (FPGA). But large scale reconfigurable computing is yet to become a reality, mainly due to the limitations of available reconfigurable hardware and the lack of the tools to aid in the fast development of applications to run on them. The MATCH project at Northwestern University is an attempt to make this application development process easier by providing tools that can quickly map programs written in a high-level language (MATLAB) to code that can be executed simultaneously on a collection of commercialoff-the-shelf (COTS) components that include DSPs and embedded processors, apart from FPGAs. The test bed for such a system has been built and one of its components is the WILDCHILD multi -FPGA system from Annapolis Microsystems Inc. To aid in the development of MATLAB-based applications, a library of matrix and signal processing functions is being developed for reconfigurable hardware that can be used as building blocks for these applications. In this paper we describe four such functions that were developed and their interfaces to the MATLAB compiler which forms the core of our efforts in this project.