Three parallel algorithms for classical molecular dynamics are presented. The first assigns each processor a fixed subset of atoms; the second assigns each a fixed subset of inter-atomic forces to compute; the third assigns each a fixed spatial region. The algorithms are suitable for molecular dynamics models which can be difficult to parallelize efficiently—those with short-range forces where the neighbors of each atom change rapidly. They can be implemented on any distributed-memory parallel machine which allows for message-passing of data between independently executing processors. The algorithms are tested on a standard Lennard-Jones benchmark problem for system sizes ranging from 500 to 100,000,000 atoms on several parallel supercomputers--the nCUBE 2, Intel iPSC/860 and Paragon, and Cray T3D. Comparing the results to the fastest reported vectorized Cray Y-MP and C90 algorithm shows that the current generation of parallel machines is competitive with conventional vector supercomputers even for small problems. For large problems, the spatial algorithm achieves parallel efficiencies of 90% and a 1840-node Intel Paragon performs up to 165 faster than a single Cray C9O processor. Trade-offs between the three algorithms and guidelines for adapting them to more complex molecular dynamics simulations are also discussed.

Fast parallel algorithms for short-range molecular dynamics

This study focuses on the routing and WavelengthAssignment (RWA) problem in wavelength-routed optical WDM networks. Most of the attention is devoted to such networks operating under the wavelength-continuity constraint, in which lightpaths are set up for connection requests between node pairs, and a single lightpath must occupy the same wavelength on all of the links that it spans. In setting up a lightpath, a route must be selected and a wavelength must be assigned to the lightpath. If no wavelength is available for this lightpath on the selected route, then the connection request is blocked. We examine the RWA problem and review various routing approaches and wavelengthassignment approaches proposed in the literature. We also briefly consider the characteristics of wavelength-converted networks (which do not have the wavelength-continuity constraint), and we examine the associated research problems and challenges. Finally, we propose a new wavelengthassignment scheme, called Distributed Relative Capacity Loss (DRCL), which works well in distributed-controlled networks, and we demonstrate the performance of DRCL through simulation.

A Review of Routing and Wavelength Assignment Approaches for Wavelength- Routed Optical WDM Networks

Due to its potential to greatly accelerate a wide variety of applications, reconfigurable computing has become a subject of a great deal of research. Its key feature is the ability to perform computations in hardware to increase performance, while retaining much of the flexibility of a software solution. In this survey, we explore the hardware aspects of reconfigurable computing machines, from single chip architectures to multi-chip systems, including internal structures and external coupling. We also focus on the software that targets these machines, such as compilation tools that map high-level algorithms directly to the reconfigurable substrate. Finally, we consider the issues involved in run-time reconfigurable systems, which reuse the configurable hardware during program execution.

/pdf/reconfigurable-computing-a-survey-of-systems-and-software-1bqra3qfdt.pdf

Reconfigurable computing: a survey of systems and software

International Technology Roadmap for Semiconductors 2003の要求清浄度について － シリコンウエハ表面と雰囲気環境に要求される清浄度, 分析方法の現状について －

Several existing volume rendering algorithms operate by factoring the viewing transformation into a 3D shear parallel to the data slices, a projection to form an intermediate but distorted image, and a 2D warp to form an undistorted final image. We extend this class of algorithms in three ways. First, we describe a new object-order rendering algorithm based on the factorization that is significantly faster than published algorithms with minimal loss of image quality. Shear-warp factorizations have the property that rows of voxels in the volume are aligned with rows of pixels in the intermediate image. We use this fact to construct a scanline-based algorithm that traverses the volume and the intermediate image in synchrony, taking advantage of the spatial coherence present in both. We use spatial data structures based on run-length encoding for both the volume and the intermediate image. Our implementation running on an SGI Indigo workstation renders a 2563 voxel medical data set in one second. Our second extension is a shear-warp factorization for perspective viewing transformations, and we show how our rendering algorithm can support this extension. Third, we introduce a data structure for encoding spatial coherence in unclassified volumes (i.e. scalar fields with no precomputed opacity). When combined with our shear-warp rendering algorithm this data structure allows us to classify and render a 2563 voxel volume in three seconds. The method extends to support mixed volumes and geometry and is parallelizable.

/pdf/fast-volume-rendering-using-a-shear-warp-factorization-of-11o5tav1m5.pdf

Fast volume rendering using a shear-warp factorization of the viewing transformation

Unlike broadcast-and-select networks, wavelength-routing networks offer the advantages of wavelength reuse and scalability and are thus suitable for wide-area networks (WANs) We study the effects of topological connectivity and wavelength conversion in circuit-switched all-optical wavelength-routing networks. A blocking analysis of such networks is given. We first propose an analytical framework for accurate analysis of networks with arbitrary topology. We then introduce a model for networks with a variable number of converters and analyze the effect of wavelength converter density on the blocking probability. This framework is applied to three regular network topologies that have varying levels of connectivity: the ring, the mesh-torus, and the hypercube. The results show that either a relatively small number of converters is sufficient for a certain level of performance or that conversion does not offer a significant advantage. The benefits of conversion are largely dependent on the network load, the number of available wavelengths, and the connectivity of the network. Finally, the tradeoff between physical connectivity, wavelength conversion, and the number of available wavelengths is studied through networks with random topologies.

/pdf/all-optical-networks-with-sparse-wavelength-conversion-15qnov6kce.pdf

All-optical networks with sparse wavelength conversion

Wireless Sensor Networks (WSNs) have become a new information collection and monitoring solution for a variety of applications. Faults occurring to sensor nodes are common due to the sensor device itself and the harsh environment where the sensor nodes are deployed. In order to ensure the network quality of service it is necessary for the WSN to be able to detect the faults and take actions to avoid further degradation of the service. The goal of this paper is to locate the faulty sensors in the wireless sensor networks. We propose and evaluate a localized fault detection algorithm to identify the faulty sensors. The implementation complexity of the algorithm is low and the probability of correct diagnosis is very high even in the existence of large fault sets. Simulation results show the algorithm can clearly identify the faulty sensors with high accuracy.

https://lib.dr.iastate.edu/cgi/viewcontent.cgi?article=1146&context=ece_conf

Distributed fault detection of wireless sensor networks

We present two dynamic routing algorithms based on path and neighborhood link congestion in all-optical networks. In such networks, a connection request encounters higher blocking probability than in circuit-switched networks because of the wavelength-continuity constraint. Much research has focused on the shortest-path routing and alternate shortest-path routing. We consider fixed-paths least-congestion (FPLC) routing in which the shortest path may not be preferred to use. We then extend the algorithm to develop a new routing method: dynamic routing using neighborhood information. It is shown by using both analysis and simulation methods that FPLC routing with the first-fit wavelength-assignment method performs much better than the alternate routing method in mesh-torus networks (regular topology) and in the NSFnet T1 backbone network (irregular topology). Routing using neighborhood information also achieves good performance when compared to alternate shortest-path routing.

Dynamic wavelength routing using congestion and neighborhood information

We consider the problem of establishing dependable connections in WDM networks with dynamic traffic demands. We call a connection with fault-tolerant requirements as a dependable connection (D-connection). We consider the single-link failure model in our study and recommend the use of a proactive approach, wherein a D-connection is identified with the establishment of the primary lightpath and a backup lightpath at the time of honoring the connection request. We develop algorithms to select routes and wavelengths to establish D-connections with improved blocking performance. The algorithms use the backup multiplexing technique to efficiently utilize the wavelength channels. To further improve channel utilization, we propose a new multiplexing technique called primary-backup multiplexing. Here, a connection may not have its backup lightpath readily available throughout its existence. We develop algorithms based on this technique to route D-connections with a specified restoration guarantee. We present an efficient and computationally simple heuristic to estimate the average number of connections per link that do not have backup lightpaths readily available upon a link failure. We conduct extensive simulation experiments on different networks to study the performance of the proposed algorithms.

Efficient algorithms for routing dependable connections in WDM optical networks

Wavelength converters increase the traffic-carrying capacity of circuit-switched optical networks by relaxing the wavelength continuity constraints. We consider the problem of optimally placing a given number of wavelength converters on a path to minimize the call-blocking probability. Using a simple performance model, we first prove that uniform spacing of converters is optimal for the end-to-end performance when link loads are uniform and independent. We then show that significant gains are achievable with optimal placement compared to random placement. For nonuniform link loads, we provide a dynamic programming algorithm for the optimal placement and compare the performance with random and uniform placement. Optimal solutions for bus and ring topologies are also presented. Finally, we discuss the effect of the traffic model on the placement decision.

Arun K. Somani

Papers

All-optical networks with sparse wavelength conversion

Distributed fault detection of wireless sensor networks

Dynamic wavelength routing using congestion and neighborhood information

Efficient algorithms for routing dependable connections in WDM optical networks

On optimal converter placement in wavelength-routed networks