SU (2) lattice gauge theory simulations on Fermi GPUs

Abstract

In this work we explore the performance of CUDA in quenched lattice SU(2) simulations. CUDA, NVIDIA Compute Unified Device Architecture, is a hardware and software architecture developed by NVIDIA for computing on the GPU. We present an analysis and performance comparison between the GPU and CPU in single and double precision. Analyses with multiple GPUs and two different architectures (G200 and Fermi architectures) are also presented. In order to obtain a high performance, the code must be optimized for the GPU architecture, i.e., an implementation that exploits the memory hierarchy of the CUDA programming model.We produce codes for the Monte Carlo generation of SU(2) lattice gauge configurations, for the mean plaquette, for the Polyakov Loop at finite T and for the Wilson loop. We also present results for the potential using many configurations (50 000) without smearing and almost 2 000 configurations with APE smearing. With two Fermi GPUs we have achieved an excellent performance of 200× the speed over one CPU, in single precision, around 110 Gflops/s. We also find that, using the Fermi architecture, double precision computations for the static quark-antiquark potential are not much slower (less than 2× slower) than single precision computations.

Color fields computed in SU(3) lattice QCD for the static tetraquark system

“…We present our results in lattice spacing units of a, with a = 0.07261(85) fm or a −1 = 2718 ± 32 MeV. We generate our configurations in NVIDIA GPUs of the FERMI series (480, 580 and Tesla 2070) with a SU(3) CUDA code upgraded from our SU(2) combination of Cabibbo-Marinari pseudoheatbath and over-relaxation algorithm [28,29]. Our SU(3) updates involve three SU(2) subgroups, we work with 9 complex numbers, and we reunitarize the matrix.…”

Landau gauge fixing on GPUs

“…To store the lattice array in global memory, we use a SOA type array as described in [18]. The main reason to do this is due to the FFT implementation algorithm, Algo.…”

Portable LQCD Monte Carlo code using OpenACC

“…Lattice QCD simulations is a typical and well known HPC grand challenge, where physics results are strongly limited by available computational resources [3,4]; over the years, several generations of parallel machines, optimized for LQCD, have been developed [5,6], while the development of LQCD codes running on many core architectures, in particular GPUs, has seen large efforts in the last decade [7][8][9]. Our target is to have a single code able to run on several processors without any major code change while looking for an acceptable trade-off between portability and efficiency [10].…”