Peter Altevogt

Parallelization of the two-dimensional Ising Model on a cluster of IBM RISC System/6000 workstations

Abstract Using the PVM programming environment for parallel applications, we have parallelized a simulation of the two-dimensional Ising Model on a cluster of IBM RISC System/6000 workstations connected by a Token Ring (16Mb/sec) and by Serial Optical Channels (220 Mb/sec) via a NSC DX Router. The parallelization is done by dividing the lattice into sublattices, each sublattice being associated with one workstation. On each sublattice, a Metropolis algorithm using Multispin Coding techniques is used to generate new configurations. We provide numerical results concerning the number of spin updates per second, speedups, and efficiencies for various numbers of processors and lattice sizes.

An algorithm for dynamic load balancing of synchronous Monte Carlo simulations on multiprocessor systems

Abstract We describe an algorithm for dynamic load balancing of geometrically parallelized synchronous Monte Carlo simulations of physical models. This algorithm is designed for a (heterogeneous) multiprocessor system of the MIMD type with distributed memory. The algorithm is based on a dynamic partitioning of the domain of the algorithm, taking into account the actual processor resources of the various processors of the multiprocessor system.

Mesoscale performance simulation of multicore processor systems

Modern microprocessor design relies heavily on detailed full-chip performance simulations to evaluate complex trade-offs. Typically, different design alternatives are tried out for a specific sub-system or component, while keeping the rest of the system unchanged. We observe that full-chip simulations for such studies is overkill. This paper introduces mesoscale simulation, which employs high-level modeling for the unchanged parts of a design and uses detailed cycle-accurate simulations for the components being modified. This combination of high-level and low-level modeling enables accuracy on par with detailed full-chip modeling while achieving much higher simulation speeds than detailed full-chip simulations. Consequently, mesoscale models can be used to quickly explore vast areas of the design space with high fidelity. We describe a proof-of-concept mesoscale implementation of the memory subsystem of the Cell/B.E. processor and discuss results from running various workloads.

Simulating very large Ising systems for short timescales

Abstract We present a new approach for investigating very large lattices of the Ising type with local update algorithms for short time scales. Instead of the conventional storing of configurations and iterating through time, we calculate spins for all time steps simultaneously and propagate through the lattice. We apply the algorithm to the two-dimensional Ising model with Metropolis dynamics and simulate systems of up to (4 × 10 5 ) 2 lattice sites for 10 MC steps.

Modular performance simulations of clouds

Performance and scalability are essential non-functional features of contemporary cloud solutions. Performance modeling and simulation techniques provide the tools required for cloud capacity planning and design. In this publication we describe a modular approach to simulate the hardware and software components of clouds. This approach supports the rapid construction of new cloud models by combining available simple or compound simulation modules, adding new cloud modules when required and adapting the implementation of existing ones if necessary. Key design points are a careful separation between hardware (infrastructure) modules and modules representing software workflows as well as the introduction of a system of a hierarchical request execution phases separating the simulation of high-level cloud workflows from the simulation of workflows at hardware component level.

Monte Carlo Simulations of Lattice Gauge Theories on Multiprocessor Systems

A large scale numerical simulation in the field of lattice gauge theory has been performed, relevant for elementary particle physics. Lattices of size 64×243 have been studied on a cluster of IBM RISC System/6000-workstations, on an IBM 9076 SP1 parallel computer with 8 nodes and on other parallel computers. A sustained performance of 30 MFLOPS/node has been reached (without special tuning steps), and a speedup of 7.1 has been found for 8 nodes on the IBM 9076 SP1. Details on the computational aspects are given. We investigate the static quark-antiquark potential up to the distance of 8 lattice spacings for pure SU(2) lattice gauge theory. Numerical simulations are performed in a large range of bare coupling constants. The action is the Wilson action with an asymmetric coupling for timelike plaquettes. The potential is obtained by fitting ’cooled’ Wilson loops with up to 3 exponential terms. An interpolation of the potentials by a sum of a perturbative and a linear term shows only approximate scaling in comparison with the symmetric