“XILINX Open Hardware 2017” – PhD category Winner

At NECSTLab, we are happy for the new achievement: on 31 August, at the XILINX Open Hardware 2017, the project “N-BObbiDY boo: Magic Acceleration of N-Body Simulation” by Emanuele Del Sozzo, Marco Rabozzi, Marco Nanni, won the PhD category.

An abstract from this project: N-Body simulation is the process that describes the evolution of a system of forces that is composed of N particles. It is a highly important process in physics as it is as general that it can describe the behavior of multiple particles in fields that range from the astrophysics to molecular dynamics. The main problem with this algorithm is that its most accurate approach, called the All-Pairs method, is particularly compute intensive and pure software implementation on CPUs are inefficient both from a performance and power consumption point of view. An implementation on a hardware accelerator, such as an FPGA, would benefits in both these terms, exploiting a parallel execution at a relative low power profile. Moreover, it would also benefit faster methods with lower computational com- plexity, since many of them exploit different methodologies based on the All-Pairs approach to approximate the calculation of forces.

This work proposes a highly scalable, power efficient and high performance hardware architecture for the N-Body All-Pairs simulation problem. Our final im- plementation can scale up to systems with an arbitrary number of bodies thanks to a tiling approach that allows performance in the order of 4400 Mpairs/s, deliv- ering a speedup of 24X (5.7X) with respect to a single-threaded (multi-threaded) O3-optimized CPU implementation, as well as the best performance per watt ratio with respect to CPUs and GPUs implementations in the state of the art.