Highly optimized simulations on single- and multi-GPU systems of the 3D Ising spin glass model

@article{Lulli2015HighlyOS,
  title={Highly optimized simulations on single- and multi-GPU systems of the 3D Ising spin glass model},
  author={Matteo Lulli and Massimo Bernaschi and Giorgio Parisi},
  journal={Comput. Phys. Commun.},
  year={2015},
  volume={196},
  pages={290-303}
}
Abstract We present a highly optimized implementation of a Monte Carlo (MC) simulator for the three-dimensional Ising spin-glass model with bimodal disorder, i.e. , the 3D Edwards–Anderson model running on CUDA enabled GPUs. Multi-GPU systems exchange data by means of the Message Passing Interface (MPI). The chosen MC dynamics is the classic Metropolis one, which is purely dissipative, since the aim was the study of the critical off-equilibrium relaxation of the system. We focused on the… Expand
Adaptive multi-GPU Exchange Monte Carlo for the 3D Random Field Ising Model
TLDR
The work proposes an adaptive multi-GPU approach that dynamically builds a proper temperature set free of exchange bottlenecks, based on mid-point insertions at the temperature gaps where the exchange rate is most compromised. Expand
Massively parallel multicanonical simulations
TLDR
This work shows how this approach for the many-thread architecture provided by current generations of graphics processing units (GPUs) can be efficiently employed with of the order of 1 0 4 parallel walkers and beyond, thus constituting a versatile tool for Monte Carlo simulations in the era of massively parallel computing. Expand
GPU accelerated population annealing algorithm
TLDR
This work presents a highly optimized implementation of the population annealing algorithm on GPUs that promises speed-ups of several orders of magnitude as compared to a serial implementation on CPUs. Expand
Monte Carlo methods for massively parallel computers
TLDR
The opportunities and challenges of massively parallel computing for Monte Carlo simulations in statistical physics, with a focus on the simulation of systems exhibiting phase transitions and critical phenomena, are outlined. Expand
Fast, Scalable, and Interactive Software for Landau-de Gennes Numerical Modeling of Nematic Topological Defects
TLDR
An open-source software package that exploits the embarrassingly parallel structure of the lattice discretization of the LdG approach, and makes use of an efficient minimization algorithm, the Fast Inertial Relaxation Engine (FIRE) method, that is well-suited to large-scale parallelization. Expand
Strong ergodicity breaking in aging of mean-field spin glasses
TLDR
It is uncovered that, contrary to common expectation, the off-equilibrium aging dynamics at low temperatures undergoes a strong ergodicity breaking and thus asymptotically remains trapped in a confined region of the configurational space. Expand
Population annealing: Massively parallel simulations in statistical physics
TLDR
An implementation of population annealing on graphics processing units is presented and its behavior for different systems undergoing continuous and first-order phase transitions is discussed. Expand
Bit-vectorized GPU implementation of a stochastic cellular automaton model for surface growth
  • J. Kelling, G. Ódor, S. Gemming
  • Computer Science, Physics
  • 2016 IEEE 20th Jubilee International Conference on Intelligent Engineering Systems (INES)
  • 2016
TLDR
A highly efficient SCA implementation of a surface growth model capable of simulating billions of lattice sites on a single GPU is presented and insight is provided into cases requiring arbitrary random probabilities which are not accessible through bit-vectorization. Expand
Computation of the dynamic critical exponent of the three-dimensional Heisenberg model.
TLDR
The dynamic critical exponent of the three-dimensional Heisenberg model is computed and the integrated autocorrelation time at equilibrium, for lattice sizes L≤64, is obtained and previous estimates of the static critical exponents, η and ν, in the out-of-equilibrium regime are checked. Expand
Benchmarking multi‐GPU applications on modern multi‐GPU integrated systems
TLDR
The results of a set of experiments aimed at assessing the performance of some of these hardware/software platforms using a particularly challenging application as a benchmark are presented. Expand
...
1
2
...

References

SHOWING 1-10 OF 41 REFERENCES
Parallel tempering simulation of the three-dimensional Edwards-Anderson model with compact asynchronous multispin coding on GPU
TLDR
This paper presents optimization and tuning approaches for the CUDA implementation of the spin glass simulation on GPUs, and presents a binary data format, Compact Asynchronous Multispin Coding (CAMSC), which provides an additional 28:4% speedup compared with the traditionally used Asynchronous multispin coding (AMSC). Expand
Multi-GPU accelerated multi-spin Monte Carlo simulations of the 2D Ising model
TLDR
This work extends the implementation of the checkerboard algorithm for the two-dimensional Ising model and is able to accelerate simulations on a single GPU by factors up to 35 compared to an optimized single Central Processor Unit (CPU) core implementation which employs multi-spin coding. Expand
Optimized GPU simulation of continuous-spin glass models
TLDR
A highly optimized code for simulating the Edwards-Anderson Heisenberg model on graphics processing units (GPUs) 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. Expand
GPU accelerated Monte Carlo simulation of the 2D and 3D Ising model
TLDR
The critical temperature of the 2D and 3D Ising model is calculated using finite size scaling techniques and an implementation of the checkerboard algorithm on a GPU is able to generate results up to 35 times faster than on a current CPU core. Expand
Reconfigurable computing for Monte Carlo simulations: Results and prospects of the Janus project
TLDR
Janus, a massively parallel FPGA-based computer optimized for the simulation of spin glasses, theoretical models for the behavior of glassy materials and an assessment of the potential of possible future versions of the Janus architecture, based on state-of-the-art technology are described. Expand
q-state Potts model metastability study using optimized GPU-based Monte Carlo algorithms
TLDR
A GPU-based parallel code to perform Monte Carlo simulations of the two-dimensional q -state Potts model is implemented, based on a checkerboard update scheme, to provide a positive numerical evidence about the existence of metastability on very large systems based on Binderʼs criterion. Expand
Multi-GPU codes for spin systems simulations
TLDR
The results show that a suitable combination of techniques allows to hide almost completely the communication overhead by using the CPU as a communication co-processor of the GPU. Expand
Benchmarking GPU and CPU codes for Heisenberg spin glass over-relaxation
TLDR
A set of possible implementations for Graphics Processing Units of the Over-relaxation technique applied to the 3D Heisenberg spin glass model show that a carefully tuned code can achieve more than 100 GFlops/s of sustained performance and update a single spin in about 0.6 nanoseconds. Expand
Critical behavior of three-dimensional Ising spin glass models
We perform high-statistics Monte Carlo simulations of three-dimensional Ising spin-glass models on cubic lattices of size L: the +- J (Edwards-Anderson) Ising model for two values of the disorderExpand
Performance potential for simulating spin models on GPU
  • M. Weigel
  • Computer Science, Physics
  • J. Comput. Phys.
  • 2012
TLDR
A number of examples, ranging from Metropolis simulations of ferromagnetic Ising models, over continuous Heisenberg and disordered spin-glass systems to parallel-tempering simulations are discussed, and significant speed-ups by factors of up to 1000 compared to serial CPU code as well as previous GPU implementations are observed. Expand
...
1
2
3
4
5
...