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 (10)

• Feedback-optimized parallel tempering Monte Carlo
• Study of the de Almeida-Thouless line using power-law diluted one-dimensional Ising spin glasses
• Make life simple: unleash the full power of the parallel tempering algorithm
• The spin glass transition of the three dimensional Heisenberg spin glass
• Finding Low-Temperature States with Parallel Tempering, Simulated Annealing and Simple Monte Carlo
• Large-scale Monte Carlo simulations of the isotropic three-dimensional Heisenberg spin glass
• Simulating spin models on GPU
• Random number generators for massively parallel simulations on GPU
• Large-scale Monte Carlo simulations of the three-dimensional XY spin glass
• Speeding up parallel tempering simulations

Cited by in corpus (1)

• Random number generators for massively parallel simulations on GPU