Seismic wave propagation simulations on low-power and performance-centric manycores

Abstract

The large processing requirements of seismic wave propagation simulations make High Performance Computing (HPC) architectures a natural choice for their execution. However, to keep both the current pace of performance improvements and the power consumption under a strict power budget, HPC systems must be more energy e cient than ever. As a response to this need, energye cient and low-power processors began to make their way into the market. In this paper we employ a novel low-power processor, the MPPA-256 manycore, to perform seismic wave propagation simulations. It has 256 cores connected by a NoC, no cache-coherence and only a limited amount of on-chip memory. We describe how its particular architectural characteristics influenced our solution for an energy-e cient implementation. As a counterpoint to the low-power MPPA-256 architecture, we employ Xeon Phi, a performance-centric manycore. Although both processors share some architectural similarities, the challenges to implement an e cient seismic wave propagation kernel on these platforms are very di↵erent. In this work we compare the performance and energy e ciency of our implementations for these processors to proven and optimized solutions for other hardware platforms such as general-purpose processors and a GPU. Our experimental results show that MPPA-256 has the best energy e ciency, consuming at least 77 % less energy than the other evaluated platforms, whereas the performance of our solution for the Xeon Phi is on par with a state-of-the-art solution for GPUs.

Otimização de Aplicações Paralelas em Aceleradores Vetoriais NEC SX-Aurora

“…Técnicas de alinhamento de memória e cache blocking são apresentadas por Castro et al, aprimorando o desempenho de uma aplicação de propagação de ondas acústicas [Castro et al 2016]. Semelhante a Castro et al, com o intuito de alcançar melhor desempenho, neste trabalho será implementado duas técnicas, loop unrolling e inlining.…”

Optimization strategies for geophysics models on manycore systems

“…Research efforts such as the presented in Castro et al (2016) improved and evaluated the performance of the acoustic wave propagation equation on Intel Xeon Phi and compared it with MPPA-256, general-purpose processors, and a GPU. The optimizations include cache blocking, memory alignment with pointer shifting, and thread affinity.…”

Portabilidade e Eficiência do Método Fletcher de Aplicações Sísmicas em Arquiteturas Multicore e GPU

“…Esforços de pesquisa como o apresentado em [Castro et al 2016] melhoraram e avaliaram o desempenho da equação de propagação de onda acústica no coprocessador Intel Xeon Phi e compararam com o processador manycore MPPA-256, com processadores de uso geral e com uma GPU. As otimizações incluem cache blocking, alinhamento de memória com deslocamento de ponteiro e afinidade de thread.…”