HyperFPGA: An Experimental Testbed for Heterogeneous Supercomputing

Abstract

The HyperFPGA is a scalable SoC-FPGA cluster aimed at exploring new architectures for improved performance and energy efficiency in high-performance computing.Recently, it has become evident that the scaling trend of von Neumann-based supercomputers is unsustainable in terms of energy and performance, stressing a change in the computing paradigm.By exploiting the flexibility, reconfigurability, and programmability offered by the HyperFPGA infrastructure, which combines FPGAs, CPUs, and high-speed general-purpose connectors, it is possible to experiment with novel computing paradigms. The Linux OS and custom drivers, along with a Message Passing Interface (MPI), offer a programmable framework for firmware and task deployment.In addition to describing the design and implementation of the HyperFPGA, we report results obtained by testing its scalability with the N-Queens problem, which is a classic benchmark for evaluating the performance of parallel computing systems.Overall, the testing of the HyperFPGA using the N-Queens problem highlights the platform's ability to handle computationally intensive tasks and demonstrates its suitability for its use in supercomputing.