QCD simulations with staggered fermions on GPUs

TL;DR

This paper presents a GPU-based implementation of the RHMC algorithm for lattice QCD with staggered fermions, achieving efficient full trajectory simulations and exploring multiGPU and OpenCL extensions.

Contribution

The authors developed a GPU-centric implementation of the RHMC algorithm for lattice QCD, performing entire molecular dynamics trajectories on GPUs, which is a novel approach.

Findings

Achieved significant performance improvements on GPUs.

Explored multiGPU and OpenCL extensions for scalability.

Provided preliminary results and future development directions.

Abstract

We report on our implementation of the RHMC algorithm for the simulation of lattice QCD with two staggered flavors on Graphics Processing Units, using the NVIDIA CUDA programming language. The main feature of our code is that the GPU is not used just as an accelerator, but instead the whole Molecular Dynamics trajectory is performed on it. After pointing out the main bottlenecks and how to circumvent them, we discuss the obtained performances. We present some preliminary results regarding OpenCL and multiGPU extensions of our code and discuss future perspectives.