Coulomb and Landau Gauge Fixing in GPUs using CUDA and MILC

TL;DR

This paper introduces GPU-based implementations of gauge fixing algorithms for SU(N) gauge theories, demonstrating improved performance over CPU methods through CUDA and multi-GPU MPI techniques.

Contribution

It presents novel GPU and multi-GPU implementations of gauge fixing algorithms for SU(N) theories, optimized with CUDA and MPI.

Findings

GPU implementation outperforms CPU MILC code

Multi-GPU approach scales efficiently with MPI

Significant speedup in gauge fixing tasks

Abstract

In this work, we present the GPU implementation of the overrelaxation and steepest descent method with Fourier acceleration methods for Laudau and Coulomb gauge fixing using CUDA for SU(N) with N>2. A multi-GPU implementation of the overrelaxation method is also presented using MPI and CUDA. The GPU performance was measured on BlueWaters and compared against the gauge fixing of the CPU MILC code.