Assessing Performance and Porting Strategies for Gravitational $N$-Body Simulations on the RISC-V-Based Tenstorrent Wormhole\textsuperscript{\texttrademark}

TL;DR

This paper evaluates three strategies for scaling gravitational N-body simulations on RISC-V-based Tenstorrent Wormhole accelerators, analyzing performance and energy efficiency to identify optimal configurations.

Contribution

It introduces and assesses three novel strategies for scaling N-body simulations specifically on RISC-V-based accelerators, focusing on performance and energy efficiency.

Findings

Identified the most energy-efficient scaling strategy.

Measured execution time improvements across strategies.

Provided insights into RISC-V accelerator performance for scientific computing.

Abstract

While RISC-V-based accelerators were initially designed with artificial intelligence applications in mind, they are increasingly being recognized as promising platforms for high performance scientific computing. In this work, we present three strategies for scaling an $N$-body code across multiple Tenstorrent Wormhole accelerators based on the RISC-V architecture. We assess the performance of these approaches by measuring both the execution time and the energy consumption required to complete a representative simulation, ultimately identifying the configuration that offers the most favorable balance between efficiency and performance.