Deep and Shallow convections in Atmosphere Models on Intel Xeon Phi Coprocessor Systems

TL;DR

This paper presents methods to accelerate atmospheric convection calculations on Intel Xeon Phi coprocessor systems, achieving significant load balancing and performance improvements through dynamic scheduling, data management, and vectorization.

Contribution

The work introduces a comprehensive approach combining dynamic scheduling, data categorization, false sharing elimination, and proportional partitioning to optimize convection calculations on heterogeneous CPU-coprocessor systems.

Findings

10% increase in overall speedup

30% improvement in convection calculation performance

Effective load balancing with dynamic scheduling

Abstract

Deep and shallow convection calculations occupy significant times in atmosphere models. These calculations also present significant load imbalances due to varying cloud covers over different regions of the grid. In this work, we accelerate these calculations on Intel{\textregistered} Xeon Phi{\texttrademark} Coprocessor Systems. By employing dynamic scheduling in OpenMP, we demonstrate large reductions in load imbalance and about 10% increase in speedups. By careful categorization of data as private, firstprivate and shared, we minimize data copying overheads for the coprocessors. We identify regions of false sharing among threads and eliminate them by loop rearrangements. We also employ proportional partitioning of independent column computations across both the CPU and coprocessor cores based on the performance ratio of the computations on the heterogeneous resources. These techniques along with various vectorization strategies resulted in about 30% improvement in convection calculations.