Accelerating High-Order Stencils on GPUs

TL;DR

This paper investigates high-order stencil computations on GPUs, focusing on seismic modeling, and presents optimized CUDA implementations that outperform existing proprietary solutions while maintaining portability.

Contribution

It provides a detailed analysis and handcrafted CUDA implementations for high-order stencils, addressing boundary conditions and performance optimization on GPUs.

Findings

Achieved twice the performance of a proprietary C/OpenACC code.

Demonstrated excellent performance portability across GPU architectures.

Provided insights into memory and data-fetching patterns for high-order stencils.

Abstract

Stencil computations are widely used in HPC applications. Today, many HPC platforms use GPUs as accelerators. As a result, understanding how to perform stencil computations fast on GPUs is important. While implementation strategies for low-order stencils on GPUs have been well-studied in the literature, not all of proposed enhancements work well for high-order stencils, such as those used for seismic modeling. Furthermore, coping with boundary conditions often requires different computational logic, which complicates efficient exploitation of the thread-level parallelism on GPUs. In this paper, we study high-order stencils and their unique characteristics on GPUs. We manually crafted a collection of implementations of a 25-point seismic modeling stencil in CUDA and related boundary conditions. We evaluate their code shapes, memory hierarchy usage, data-fetching patterns, and other performance attributes. We conducted an empirical evaluation of these stencils using several mature and emerging tools and discuss our quantitative findings. Among our implementations, we achieve twice the performance of a proprietary code developed in C and mapped to GPUs using OpenACC. Additionally, several of our implementations have excellent performance portability.