NERO: Accelerating Weather Prediction using Near-Memory Reconfigurable Fabric

Ongoing climate change calls for fast and accurate weather and climate modeling. However, when solving large-scale weatherprediction 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 fundamentalchallenges to acceleration. To overcome these challenges, we propose and evaluate the use of near-memory acceleration using areconfigurable fabric with high-bandwidth memory (HBM). We focus on compound stencils that are fundamental kernels in weatherprediction models. By using high-level synthesis techniques, we develop NERO, an FPGA+HBM-based accelerator connected throughIBM OCAPI (Open Coherent Accelerator Processor Interface) to an IBM POWER9 host system. Our experimental results show thatNERO outperforms a 16-core POWER9 system by5.3×and12.7×when running two different compound stencil kernels. NEROreduces the energy consumption by12×and35×for the same two kernels over the POWER9 system with an energy efficiency of 1.61GFLOPS/Watt and 21.01 GFLOPS/Watt . We conclude that employing near-memory acceleration solutions for weather predictionmodeling is promising as a means to achieve both high performance and high energy efficiency.