Reconfigurable Computing
Overview
This set of research projects concentrates on Reconfigurable Computing, or the use of FPGAs, in high performance computing. Here FPGAs are used as computing accelerators in combination with a general purpose processor. Special focus is on high-level design and synthesis techniques to achieve this.
SKA -- Square Kilometer Array
This project focuses on FPGA-based high-performance computing and high-level design for the world's largest radio astronomy project--Square Kilometre Array (SKA).
The PARC lab is officially involved in the CSP (Central Signal Processor) package of the SKA. Inside this package we work on the correlator design and pulsar search, in collaboration with the University of Manchester. Based on high-level design and synthesis techniques, involving the use of OpenCL, we investigate high-performance and low-power signal processing solutions on FPGAs for the immense computing problems.
Our involvement is supported by MBIE funding for the NZ SKA Alliance.
cellml-opt - Domain-specific optimisation of CellML models
cellml-opt is a source-to-source compiler. It applies a combination of standard LLVM passes and domain-specific optimisations to the equation systems found in CellML models. Input and ouput are C files as generated by the CellML framework's C Code Generation Service.
This project is done in collaboration with the Technical University of Darmstadt, with Julian Oppermann and Andreas Koch.