A Multi-FPGA High Performance Computing System for 3D FFT-based Numerical Simulations

TL;DR

This paper presents a specialized multi-FPGA system with a custom network interface and parallel architecture to accelerate 3D FFT computations, addressing communication bottlenecks in large-scale scientific simulations.

Contribution

It introduces a novel multi-FPGA high-performance computing system with a custom network interface and a parallel-pipelined complex arithmetic unit for efficient 3D FFT-based simulations.

Findings

Reduced communication latency in large-scale simulations

Enhanced scalability with increased computing elements

Achieved high throughput in 3D FFT computations

Abstract

In the field of High Performance Computing, communications among processes represent a typical bottleneck for massively parallel scientific applications. Object of this research is the development of a network interface card with specific offloading capabilities that could help large scale simulations in terms of communication latency and scalability with the number of computing elements. In particular this work deals with the development of a double precision floating point complex arithmetic unit with a parallel-pipelined architecture, in order to implement a massively parallel computing system tailored for three dimensional Fast Fourier Transform.