A High-Performance Implementation of Atomistic Spin Dynamics Simulations on x86 CPUs
arXiv:2304.10966 · doi:10.1016/j.cpc.2023.108851
Abstract
Atomistic spin dynamics simulations provide valuable information about the energy spectrum of magnetic materials in different phases, allowing one to identify instabilities and the nature of their excitations. However, the time cost of evaluating the dynamical correlation function increases quadratically as the number of spins , leading to significant computational effort, making the simulation of large spin systems very challenging. In this work, we propose to use a highly optimized general matrix multiply (GEMM) subroutine to calculate the dynamical spin-spin correlation function that can achieve near-optimal hardware utilization. Furthermore, we fuse the element-wise operations in the calculation of into the in-house GEMM kernel, which results in further performance improvements of 44\% - 71\% on several relatively large lattice sizes when compared to the implementation that uses the GEMM subroutine in OpenBLAS, which is the state-of-the-art open source library for Basic Linear Algebra Subroutine (BLAS).
18 (short) pages, 6 figures
References in corpus (22)
- Antiferromagnetic spintronics
- Resonant Inelastic X-ray Scattering Studies of Elementary Excitations
- Atomistic spin model simulations of magnetic nanomaterials
- Colloquium: Spintronics in graphene and other two-dimensional materials
- A method for atomistic spin dynamics simulations: implementation and examples
- Spirit: Multifunctional Framework for Atomistic Spin Simulations
- Massively parallel symplectic algorithm for coupled magnetic spin dynamics and molecular dynamics
- The Mpemba effect in spin glasses is a persistent memory effect
- General method for atomistic spin-lattice dynamics with first principles accuracy
- Dynamical structure factor of the three-dimensional quantum spin liquid candidate NaCaNiF
- The Landau-Lifshitz equation in atomistic models
- Comprehensive study of the dynamics of a classical Kitaev spin liquid
- Simulating spin models on GPU
- Quantum-accurate magneto-elastic predictions with classical spin-lattice dynamics
- Classical and quantum spin dynamics of the honeycomb model
- GPU accelerated Monte Carlo simulations of lattice spin models
- Signatures of a liquid-crystal transition in spin-wave excitations of skyrmions
- Symplectic integrators for classical spin systems
- A GPU-based Large-scale Monte Carlo Simulation Method for Systems with Long-range Interactions
- Spatial correlation functions in 3-d Ising spin glasses
- Spin dynamics of the antiferromagnetic Heisenberg model on a kagome bilayer
- Parallelization and implementation of multi-spin Monte Carlo simulation of 2D square Ising model using MPI and C++