GPU-Accelerated Simulations of Isolated Black Holes

TL;DR

This paper details the development of a GPU-accelerated version of the SpEC numerical relativity code for simulating isolated black holes, using automated porting strategies to minimize code changes and achieve efficient GPU performance.

Contribution

The authors introduce a novel approach combining automated code generation and matrix caching to port complex tensor computations to GPUs with minimal modifications.

Findings

Achieved significant speedups in black hole simulations on NVIDIA GPUs.

Demonstrated the effectiveness of automated porting strategies for complex scientific codes.

Provided benchmarks showing performance improvements across GPU generations.

Abstract

We present a port of the numerical relativity code SpEC which is capable of running on NVIDIA GPUs. Since this code must be maintained in parallel with SpEC itself, a primary design consideration is to perform as few explicit code changes as possible. We therefore rely on a hierarchy of automated porting strategies. At the highest level we use TLoops, a C++ library of our design, to automatically emit CUDA code equivalent to tensorial expressions written into C++ source using a syntax similar to analytic calculation. Next, we trace out and cache explicit matrix representations of the numerous linear transformations in the SpEC code, which allows these to be performed on the GPU using pre-existing matrix-multiplication libraries. We port the few remaining important modules by hand. In this paper we detail the specifics of our port, and present benchmarks of it simulating isolated black hole spacetimes on several generations of NVIDIA GPU.