Optimized GPU simulation of continuous-spin glass models

Abstract

We develop a highly optimized code for simulating the Edwards-Anderson Heisenberg model on graphics processing units (GPUs). Using a number of computational tricks such as tiling, data compression and appropriate memory layouts, the simulation code combining over-relaxation, heat bath and parallel tempering moves achieves a peak performance of 0.29 ns per spin update on realistic system sizes, corresponding to a more than 150 fold speed-up over a serial CPU reference implementation. The optimized implementation is used to study the spin-glass transition in a random external magnetic field to probe the existence of a de Almeida-Thouless line in the model, for which we give benchmark results.

to appear in Eur. Phys. J. Spec. Top. issue "Computer simulations on GPU"

References in corpus (28)

• Exchange Monte Carlo Method and Application to Spin Glass Simulations
• Feedback-optimized parallel tempering Monte Carlo
• Critical Behavior of the Three-Dimensional Ising Spin Glass
• Critical exponents of the three dimensional diluted Ising model
• Replica Symmetry Breaking in Short-Range Spin Glasses: Theoretical Foundations and Numerical Evidences
• Phase Transition in the Three-Dimensional $\pm J$ Ising Spin Glass
• Spin and link overlaps in 3-dimensional spin glasses
• Single spin- and chiral-glass transition in vector spin glasses in three-dimensions
• Study of the de Almeida-Thouless line using power-law diluted one-dimensional Ising spin glasses
• JANUS: an FPGA-based System for High Performance Scientific Computing
• Make life simple: unleash the full power of the parallel tempering algorithm
• Ising spin glass transition in magnetic field out of mean-field
• The spin glass transition of the three dimensional Heisenberg spin glass
• Performance potential for simulating spin models on GPU
• Large-scale Monte Carlo simulations of the isotropic three-dimensional Heisenberg spin glass
• Finding Low-Temperature States with Parallel Tempering, Simulated Annealing and Simple Monte Carlo
• Random number generators for massively parallel simulations on GPU
• Simulating spin models on GPU
• Numerical evidence of the spin-chirality decoupling in the three-dimensional Heisenberg spin glass
• Canonical local algorithms for spin systems: Heat Bath and Hasting's methods
• Large-scale Monte Carlo simulations of the three-dimensional XY spin glass
• The de Almeida-Thouless line in vector spin glasses
• A geometrical picture for finite dimensional spin glasses
• Connected component identification and cluster update on GPU
• GPU accelerated Monte Carlo simulations of lattice spin models
• Ground States and Defect Energies of the Two-dimensional XY Spin Glass from a Quasi-Exact Algorithm
• Speeding up parallel tempering simulations
• Genetic embedded matching approach to ground states in continuous-spin systems

Cited by in corpus (11)

• Random number generators for massively parallel simulations on GPU
• GPU accelerated population annealing algorithm
• Phase Transition in 3d Heisenberg Spin Glasses with Strong Random Anisotropies, through a Multi-GPU Parallelization
• Highly optimized simulations on single- and multi-GPU systems of 3D Ising spin glass
• Adaptive Multi-GPU Exchange Monte Carlo for the 3D Random Field Ising Model
• Directly observing replica symmetry breaking in a vector quantum-optical spin glass
• Dynamic scaling in the 2D Ising spin glass with Gaussian couplings
• Superdiffusion-like behavior in zero-temperature coarsening of the $d=3$ Ising model
• Dynamical heterogeneities as fingerprints of a backbone structure in Potts models
• Massively parallel simulations for disordered systems
• Massive-Scale Simulations of 2D Ising and Blume-Capel Models on Rack-Scale Multi-GPU Systems