A Multicore Architecture for High-Performance Scientific Computing Using FPGAs

A multicomputer architecture is proposed in order to achieve high performance in floating point using FPGA devices. This architecture is used in the system prototype of an instrument that carries out a real-time scientific data analysis aboard the ESA's Solar Or biter space mission. A novel method for parallelizing the scientific algorithm in multicore systems is developed. A software tool that automatizes the architecture design process and configuration is presented.