JANUS: an FPGA-based System for High Performance Scientific Computing

Abstract

This paper describes JANUS, a modular massively parallel and reconfigurable FPGA-based computing system. Each JANUS module has a computational core and a host. The computational core is a 4x4 array of FPGA-based processing elements with nearest-neighbor data links. Processors are also directly connected to an I/O node attached to the JANUS host, a conventional PC. JANUS is tailored for, but not limited to, the requirements of a class of hard scientific applications characterized by regular code structure, unconventional data manipulation instructions and not too large data-base size. We discuss the architecture of this configurable machine, and focus on its use on Monte Carlo simulations of statistical mechanics. On this class of application JANUS achieves impressive performances: in some cases one JANUS processing element outperfoms high-end PCs by a factor ~ 1000. We also discuss the role of JANUS on other classes of scientific applications.

11 pages, 6 figures. Improved version, largely rewritten, submitted to Computing in Science & Engineering

References in corpus (3)

• Nonequilibrium spin glass dynamics from picoseconds to 0.1 seconds
• Simulating spin systems on IANUS, an FPGA-based computer
• QCD on the Cell Broadband Engine

Cited by in corpus (45)

• Critical parameters of the three-dimensional Ising spin glass
• An in-depth view of the microscopic dynamics of Ising spin glasses at fixed temperature
• Nature of the spin-glass phase at experimental length scales
• Weighted p-bits for FPGA implementation of probabilistic circuits
• Performance potential for simulating spin models on GPU
• Simulating spin models on GPU
• Unraveling Quantum Annealers using Classical Hardness
• GPU accelerated population annealing algorithm
• Efficient Quantum Walk Circuits for Metropolis-Hastings Algorithm
• The three dimensional Ising spin glass in an external magnetic field: the role of the silent majority
• Aging rate of spin glasses from simulations matches experiments
• Temperature chaos in 3D Ising Spin Glasses is driven by rare events
• Dynamical transition in the D = 3 Edwards-Anderson spin glass in an external magnetic field
• A statics-dynamics equivalence through the fluctuation-dissipation ratio provides a window into the spin-glass phase from nonequilibrium measurements
• Practical engineering of hard spin-glass instances
• Dynamic Variational Study of Chaos: Spin Glasses in Three Dimensions
• FPGA Hardware Acceleration of Monte Carlo Simulations for the Ising Model
• Connected component identification and cluster update on GPU
• Reconfigurable computing for Monte Carlo simulations: results and prospects of the Janus project
• A review of High Performance Computing foundations for scientists
• Tethered Monte Carlo: computing the effective potential without critical slowing down
• GPU accelerated Monte Carlo simulations of lattice spin models
• Optimized GPU simulation of continuous-spin glass models
• Ageing at the Spin-Glass/Ferromagnet Transition: Monte Carlo Simulation using GPUs
• The Spin Glass Phase in the Four-State, Three-Dimensional Potts Model
• The cumulative overlap distribution function in realistic spin glasses
• Estimating the Density of States of Frustrated Spin Systems
• An experiment-oriented analysis of 2D spin-glass dynamics: a twelve time-decades scaling study
• Tethered Monte Carlo: Managing rugged free-energy landscapes with a Helmholtz-potential formalism
• Critical Behavior of Three-Dimensional Disordered Potts Models with Many States
• Percolation of Fortuin-Kasteleyn clusters for the random-bond Ising model
• Accurate Sampling with Noisy Forces from Approximate Computing
• Spin-glass dynamics: experiment, theory and simulation
• Explicit generation of the branching tree of states in spin glasses
• On the superposition principle and non-linear response in spin glasses
• Replica symmetry breaking for Ulam's problem
• General Oscillator-Based Ising Machine Models with Phase-Amplitude Dynamics and Polynomial Interactions
• SWAP algorithm for lattice spin models
• Solving Spin Glasses with Optimized Trees of Clustered Spins
• Massively parallel simulations for disordered systems
• Logically Synthesized, Hardware-Accelerated, Restricted Boltzmann Machines for Combinatorial Optimization and Integer Factorization
• Computing Canonical Averages with Quantum and Classical Optimizers: Thermodynamic Reweighting for QUBO Models of Physical Systems
• Noise-injected analog Ising machines enable ultrafast statistical sampling and machine learning
• Numerical study of the overlap Lee-Yang singularities in the three-dimensional Edwards-Anderson model
• Jump Events in a 3D Edwards-Anderson Spin Glass