Finite temperature lattice QCD with GPUs

TL;DR

This paper demonstrates the use of GPUs for lattice QCD simulations, showing performance improvements and presenting results for SU(2) observables across different temperatures.

Contribution

It provides a performance comparison between GPU and CPU implementations and presents finite temperature SU(2) results using GPU acceleration.

Findings

GPU implementations outperform CPU in generating lattice configurations.

Single and multiple GPU analyses show scalable performance.

Finite temperature SU(2) results match theoretical expectations.

Abstract

Graphics Processing Units (GPUs) are being used in many areas of physics, since the performance versus cost is very attractive. The GPUs can be addressed by CUDA which is a NVIDIA's parallel computing architecture. It enables dramatic increases in computing performance by harnessing the power of the GPU. We present a performance comparison between the GPU and CPU with single precision and double precision in generating lattice SU(2) configurations. Analyses with single and multiple GPUs, using CUDA and OPENMP, are also presented. We also present SU(2) results for the renormalized Polyakov loop, colour averaged free energy and the string tension as a function of the temperature.