SIMD architecture on FPGA for scientific computing aboard a space instrument

In this paper we propose a SIMD multiprocessor architecture to reach high performance in floating point operations by using FPGA devices. This architecture is used in an instrument that carries out the scientific analysis aboard the ESA's Solar Orbiter mission. We present a programming language and a compiler able to automatize the SIMD configuration process by using an initial sequential code. The proposed architecture squeezes the FPGA resources in order to reach the time constraints. The achieved FPGA system improves the ground-based system performance based on commercial CPUs regarding time and power consumption. © 2015 Elsevier B.V.