Accelerating Weather Prediction using Near-Memory Reconfigurable Fabric

TL;DR

This paper presents NERO, an FPGA+HBM-based near-memory accelerator that significantly improves performance and energy efficiency for weather prediction simulations by overcoming traditional memory bottlenecks.

Contribution

The paper introduces NERO, a novel near-memory reconfigurable fabric accelerator for weather prediction, demonstrating substantial performance and energy efficiency gains over CPU systems.

Findings

NERO outperforms POWER9 by 5.3x and 12.7x on two kernels.

NERO reduces energy consumption by 12x and 35x.

Achieves high energy efficiency of up to 21.01 GFLOPS/Watt.

Abstract

Ongoing climate change calls for fast and accurate weather and climate modeling. However, when solving large-scale weather prediction simulations, state-of-the-art CPU and GPU implementations suffer from limited performance and high energy consumption. These implementations are dominated by complex irregular memory access patterns and low arithmetic intensity that pose fundamental challenges to acceleration. To overcome these challenges, we propose and evaluate the use of near-memory acceleration using a reconfigurable fabric with high-bandwidth memory (HBM). We focus on compound stencils that are fundamental kernels in weather prediction models. By using high-level synthesis techniques, we develop NERO, an FPGA+HBM-based accelerator connected through OCAPI (Open Coherent Accelerator Processor Interface) to an IBM POWER9 host system. Our experimental results show that NERO outperforms a 16-core POWER9 system by 5.3x and 12.7x when running two different compound stencil kernels. NERO reduces the energy consumption by 12x and 35x for the same two kernels over the POWER9 system with an energy efficiency of 1.61 GFLOPS/Watt and 21.01 GFLOPS/Watt. We conclude that employing near-memory acceleration solutions for weather prediction modeling is promising as a means to achieve both high performance and high energy efficiency.