Parallel MIMD Architectures Research Articles

Hybrid Regression Testing Based on Path Pruning

Regression testing has been studied by various researchers for developing and testing the quality of software. Regression testing aims at re-execution of evolved software code to ensure that no new errors had been introduced during the process of modification. Since re-execution of all test cases is not feasible, selecting manageable number of test cases to execute modified code with good fault detection rate is a problem. In past few years, various hybrid based regression testing approaches have been proposed and successfully employed for software testing, aiming at reduction in the number of test cases and higher fault detection capabilities. These techniques are based on sequence of selections, prioritizations and minimization of test suite. However, these techniques suffer from major drawbacks like improper consideration of control dependencies, neglection of unaffected fragments of code for testing purpose. Further, these techniques have been employed on hypothetical or simple programs with test suites of smaller size. Present paper proposes hybrid regression testing, a combination of test case selections, test case prioritizations and test suite minimization. The technique works at statement level and is based on finding the paths containing statements that affects or gets affected by the addition/deletion or modification (both control and data dependency) of variables in statements. The modification in the code may cause ripple effect thereby resulting into faulty execution of the code. The hybrid regression testing approach is aimed at detecting such faults with lesser number of test cases. Reduction in number of test cases is possible because of the decreased number of paths to be tested. A web based framework to automate and parallelize this testing technique to maximum extend, making it well suited for globally distributed environments is also proposed in the present paper. Framework when implemented as a tool can handle large pool of test cases and will make use of parallel MIMD architectures like multicore systems. Technique is applied on prototype live system and results are compared with recently proposed hybrid regression testing approach against parameters of interest. Obtained optimized results are indicators of effectiveness of approach in terms of reduction in effort, cost as well as testing time in general and increment delivery time in particular.

Flexible Scheduling and Thread Allocation for Synchronous Parallel Tasks

We describe a task model and dynamic scheduling and resource allocation mechanism for synchronous parallel tasks to be executed on SPMD-programmed synchronous shared-memory MIMD parallel architectures with uniform, unit-time memory access and strict memory consistency, also known in the literature as PRAMs (Parallel Random Access Machines). Our task model provides a two-tier programming model for PRAMs that flexibly combines SPMD and fork-join parallelism within the same application. It offers flexibility by dynamic scheduling and late resource binding while preserving the PRAM execution properties within each task, the only limitation being that the maximum number of threads that can be assigned to a task is limited to what the underlying architecture provides. In particular, our approach opens for automatic performance tuning at run-time by controlling the thread allocation for tasks based on run-time predictions. By a prototype implementation of a synchronous parallel task API in the SPMD-based PRAM language Fork and experimental evaluation with example programs on the SBPRAM simulator, we show that a realization of the task model on a SPMD-programmable PRAM machine is feasible with moderate runtime overhead per task.

A Unified Formulation for Massively Parallel Rigid Multibody Dynamics of O(log2n) Computational Complexity

A novel algorithm for the solution of the inverse dynamics problem is presented and augmented to the solution of the equations of motion (EOM) for rigid multibody chains using explicit constraint components of force. The unified model corresponds to an optimal, strictly parallel, time, space, and processor lower bound solution to the dynamics of accelerated rigid multibodies, i.e., computation time of O(log2n) using O(n) processors for an n body system. Complex topological structures are supported in the form of multiple degree-of-freedom (DOF) joints/hinges, free-floating, hyper-branched, and/or closed-chain systems, with applications ranging from multibody molecular dynamics simulations and computational molecular nanotechnology, to real-time control and simulation of spatial robotic manipulators. In addition to the theoretical significance, the algorithms presented are shown to be very efficient for practical implementation on MIMD parallel architectures for large-scale systems.

Applying cellular automata to complex environmental problems: The simulation of the bioremediation of contaminated soils

Cellular automata can be applied to modelling the dynamics of spatially extended physical systems, and represent an alternative to the classical PDE approach. In this paper a macroscopic cellular automata model for simulating the bioremediation of contaminated soils is introduced. The choice of macroscopic automata is motivated by the aim to simulate large-scale systems. It is suggested that in some cases, where the basic laws of continuum mechanics cannot be directly applied without adding phenomenological assumptions, and where the equation system is not amenable to analytical solution, direct discrete modelling may represent a convenient alternative to the use of continuum models, followed by numerical discretization. This hypothesis is empirically tested in the bioremediation case. The model describes the bioremediation of contaminated soils, which relies upon the use of indigeneous microorganisms to degrade the contaminant: bioremediation models pose particular challenges as several physical, chemical and biological phenomena interact in a disordered and partially unknown matrix (the soil). The model is hierarchical, and is composed by a fluid dynamical layer, a solute description layer and a biological layer. The model has been tested in a pilot plant, in the case of contamination by phenol. The values of the phenomenological parameters have been determined by the use of genetic algorithms. The model has proven capable to carefully describe experimental results in a wide range of experimental conditions. It has also been run on a MIMD parallel architecture, achieving a high speed-up. It therefore represents an example of application of cellular automata to a real-world problem which has a very high social and economic importance, and where progresses in modelling may greatly improve the effectiveness of the decontamination interventions.

Understanding parallel supercomputing

From the Publisher: UNDERSTANDING PARALLEL SUPERCOMPUTING is an exhaustive, applications-oriented survey of the world's largest and fastest computers. Beginning with the evolution of parallel supercomputing technology in recent history, author R. Michael Hord goes on to illustrate architectural concepts and implementations at the very center of today's cutting-edge technology. Topics featured include: technology benefits and drawbacks, software tools and programming languages, major programming concepts, sample parallel programs, algorithmic methods, both SIMD and MIMD architectures. This carefully written text will be of interest to engineers, scientists, and program managers involved in geologic exploration, aircraft design, image processing, weather modeling, operations, research, chemical synthesis, and medical applications. It will also be of practical use to computer specialists.

Design and implementation of a parallel cellular language for MIMD architectures

In this paper we present a new language, called ParCeL-1, intended to make easier the implementation of computation-intensive and artificial intelligence applications on highly parallel MIMD architectures. The computational model of ParCeL-1 is object-oriented and synchronous, based on virtual processors called cells that compute and communicate alternately. The current prototype is implemented on several multi-processor systems (Cray T3D, Intel Paragon, Telmat T-Node, workstation networks using PVM, and SGI Power Challenge Array). Several applications written in ParCeL-1 are described, together with their performances obtained on a 256-processor Cray T3D.

Implementation of an environment for Monte Carlo simulation of fully 3-D positron tomography on a high-performance parallel platform

This paper describes the implementation of the Eidolon Monte Carlo program designed to simulate fully three-dimensional (3-D) cylindrical positron tomographs on a MIMD parallel architecture. The original code was written in Objective-C and developed under the NeXTSTEP development environment. Different steps involved in porting the software on a parallel architecture based on PowerPC 604 processors running under AIX 4.1 are presented. Basic aspects and strategies of running Monte Carlo calculations on parallel computers are described. A linear decrease of the computing time was achieved with the number of computing nodes. The improved time performances resulting from parallelisation of the Monte Carlo calculations makes it an attractive tool for modelling photon transport in 3-D positron tomography. The parallelisation paradigm used in this work is independent from the chosen parallel architecture.

A new method of multibeamforming for a swath bathymetry sonar system and its MIMD parallel architecture

In a swath bathymetry sonar system, real-time multibeamforming is required in order to increase the coverage and the efficiency. To determine the direction of seafloor echo accurately, traditional beamforming must work at a high sample rate, the interpolation is used to reduce computing amounts, but when there are tens of receiving channels and beams, the interpolation filter algorithm will rapidly increase computing amounts to result in the real-time multibeamforming, this is very difficult. Combining with a swath bathymetry sonar system, a new method of quadrature beamforming is described in this paper. It combines time domain beamforming with phase shifting beamforming to process the available bandpass signal. To reduce sample rate, a mixing frequency technique is used. A quadrature component is obtained from a single-channel A/D converter utilizing Hilbert transformation and its coefficient is weighted with Hamming window function, therefore, amplitude and phase is highly accurate. Comparing with the traditional beamforming method, the method presented in the paper is characterized by high accuracy, less sample rate, and less computation complexity, easily realized with a high speed DSP device. The parallel pipeline architecture is given based on TMS320C30. Simulating and lake trial results show that the method is right and efficient. The method is universal for active sonar.

Novel algorithms for massively parallel, long-term, simulation of molecular dynamics systems

In this paper, a novel algorithm for solution of the constrained equations of motion with application to simulation of the molecular dynamics systems is presented. The algorithm enables the solution of equations of motion with an internal coordinates model wherein the high-frequency oscillations are frozen by explicit inclusion of hard constraints in the system as well as by clustering of atoms and, thus, allowing a much larger time step in the integration. For a molecular system with N clusters, the algorithm achieves the optimal sequential complexity of O( N). However, the main advantage of this new algorithm is its efficiency for massively parallel computation. In fact, this is the first known algorithm that achieves a both time- and processor-optimal parallel solution for the constrained equations of motion, i.e. an optimal computation time of O(log N) by using an optimal number of O( N) processors. In addition to its theoretical significance, this algorithm is also very efficient for practical implementation on the coarse grain MIMD parallel architectures owing to its highly decoupled computational structure.

EVPI-Based Importance Sampling Solution Procedures for Multistage Stochastic Linear Programmes on Parallel MIMD Architectures

Multistage stochastic linear programming has many practical applications for problems whose current decisions have to be made under future uncertainty. There are a variety of methods for solving the deterministic equivalent forms of these dynamic problems, including the simplex and interior point methods and nested Benders decomposition which decomposes the original problem into a set of smaller linear programming problems and has recently been shown to be superior to the alternatives for large problems. The Benders subproblems can be visualised as being attached to the nodes of a tree which is formed from the realisations of the random data vectors determining the uncertainty in the problem. Parallel versions of the nested Benders algorithm involve two obvious techniques for parallelising the associated tree structure for multiprocessors or multicomputers subtree parallelisation or a nodal parallelisation both of which utilise a farming approach. The nodal parallelisation technique is presented in this paper, as it balances load more efficiently than its alternative. Differing structures of the test problems cause differing levels of speedup on a variety of multicomputing platforms: problems with few variables and constraints per node do not gain from this parallelisation. Stage aggregation has been successfully employed for such problems to improve their parallel solution efficiency by increasing the size of the nodes and therefore the time spent calculating relative to the time spent communicating between processors. A parallel version of an importance sampling solution algorithm based on local EVPI information has been developed for extremely large multistage stochastic linear programmes which either have too many data paths to solve directly or a continuous distribution of possible realisations. It utilises the parallel nested Benders algorithm and a parallel version of an algorithm designed to calculate the local EVPI values for the nodes of the tree and achieves near linear speedup.

A domain decomposition implementation of the SIMPLE method with PVM

A parallel implementation of a finite volume method for the solution of the Navier-Stokes equations on a distributed computing environment through Parallel Virtual Machine (PVM) is reported. The numerical method is implicit and is based on the SIMPLE algorithm in which the system of equations is discretised using a hybrid scheme. An Alternative Direction Implicit (ADI) scheme, and the Thomas tri-diagonal solver are used to solve the algebraic equations. The parallelization of the program is implemented by a domain decomposition strategy on MIMD parallel architectures using PVM platform. The program was tested for laminar flow in a cavity. The parallelisation strategy and performance are discussed. It is concluded that the efficiency is strongly dependent on the grid size, block numbers and the number of processors. Different strategies to improve the computational efficiency are proposed.

ProperTEST: a portable parallel test generator for sequential circuits

Parallel algorithms developed for CAD problems today suffer from two important drawbacks. First, they are machine specific, and tend to perform poorly on architectures other than the one for they were designed. Second, the quality of results degrades significantly during parallel execution. In this paper, we address these two problems for an important CAD application: test generation for sequential circuits, We have developed a new parallel test generator, ProperTEST, that is portable across a range of MIMD parallel architectures. This work is part of the ProperCAD project which aims to develop CAD algorithms that run unchanged on shared and nonshared memory machines. We present performance data for ProperTEST on ISCAS 89 sequential circuits on a Sequent Symmetry, an Intel i860 hypercube, an NCUBE/2 hypercube, a network of Sun workstations, and an Encore Multimax. Parallel processing can also be used to improve on the fault coverage possible on one processor in a given amount of time. This was not possible in earlier approaches due to search anomalies. Using ProperTEST, we provide results on ISCAS 89 benchmark programs demonstrating the improvements in fault coverage as the number of processors is increased.

Robust Algorithms for Solving Stochastic Partial Differential Equations

A robust semi-implicit central partial difference algorithm for the numerical solution of coupled stochastic parabolic partial differential equations (PDEs) is described. This can be used for calculating correlation functions of systems of interacting stochastic fields. Such field equations can arise in the description of Hamiltonian and open systems in the physics of nonlinear processes, and may include multiplicative noise sources. The algorithm can be used for studying the properties of nonlinear quantum or classical field theories. The general approach is outlined and applied to a specific example, namely the quantum statistical fluctuations of ultra-short optical pulses in χ(2)parametric waveguides. This example uses a non-diagonal coherent state representation, and correctly predicts the sub-shot noise level spectral fluctuations observed in homodyne detection measurements. It is expected that the methods used will be applicable for higher-order correlation functions and other physical problems as well. A stochastic differencing technique for reducing sampling errors is also introduced. This involves solving nonlinear stochastic parabolic PDEs in combination with a reference process, which uses the Wigner representation in the example presented here. A computer implementation on MIMD parallel architectures is discussed.

Two-dimensional block partitionings for the parallel sparse Cholesky factorization

This paper presents a discussion on 2D block mappings for the sparse Cholesky factorization on parallel MIMD architectures with distributed memory. It introduces the fan-in algorithm in a general manner and proposes several mapping strategies. The grid mapping with row balancing, inspired by Rothberg's work (1994), is proved to be more robust than the original fan-out algorithm. Even more efficient is the proportional mapping, as shown by the experiments on a 32 processor IBM SP1 and on a Cray T3D. Subforest-to-subcube mappings are also considered and give good results on the T3D.

ParCeL-1

In this paper we present a new language, called ParCeL-1, intended to make easier the implementation of computation-intensive applications on highly parallel MIMD architectures. The computational model of ParCeL-1 is object-oriented and synchronous. The current prototype is implemented on several multi-processor systems (Cray T3D, Intel Paragon, Telmat T-Node and workstation networks using PVM). Benchmarks are presented for several parallel programs.

Wavelet packet best basis selection on moderate parallel MIMD architectures

Wavelet packet best basis selection has become a popular method in image compression. This paper introduces a subband based parallelization which overcomes most of the difficulties of a straightforward parallel version of the sequential algorithm. Beside the higher efficiency the algorithm is easier to implement than its classical version.

Comparison of MIMD and SIMD strategies for Monte Carlo modelling of kinetically coupled reactions

The atomic-level description of a complex heavy hydrocarbon feedstock and its complete temporal product behavior is a challenging and computationally intensive problem which requires robust reaction simulations to accurately account for kinetic interactions and efficient utilization of available computer hardware for solution. Herein we discuss the parallel implementation of different Monte Carlo reaction algorithms for two prototype kinetic problems and examine their efficiency on MIMD and SIMD parallel architectures. The concepts developed are subsequently applied to a complex resid pyrolysis model. Both fixed- and variable-time step Monte Carlo approaches to the stochastic simulation algorithm were investigated. The prototype kinetic problems consisted of A i⇌B i with Ldngmuir-Hinshelwood-Hougen-Watson kinetics for i = 1–3 and a hypothetical 1–5 ring aro matic hydrogenation simulation. The two parallel architectures utilized were BBN's TC2000 (MIMD) and Thinking Machines CM-2 (SIMD). A new modelling approach that stochastically samples the reaction environment to provide an estimate of species' interactions is introduced. This “partial system” approach exhibited excellent parallel efficiency ( > 90%) on the TC2000. The simple prototype A i⇌B i kinetic problem was also successfully implemented on the SIMD CM-2. While the 1–5 ring hydrogenation problem efficiently took advantage of the MIMD architecture, its performance on the CM-2 was quite poor due to the the lack of a match between the data focus of complex model simulations and the CM-2 data parallel paradigm. The resid pyrolysis model performed extremely well on the MIMD machine.

Strategies for modelling kinetic interactions in complex mixtures: Monte Carlo algorithms for MIMD parallel architectures

The parallel implementation of Monte Carlo reaction models for kinetically coupled reaction systems is discussed in the context of a BBN TC2000 MIMD architecture. A comparatively simple Rice—Herzfeld co-pyrolysis reaction scheme is utilized to develop four modelling strategies encompassing “full system”, “partial system”, “fixed-time”, and “variable-time” simulation approaches. The parallel efficiency of the partial system approaches was good. The partial system variable-time algorithm was then applied to simulate the pyrolysis of a complex petroleum resid to demonstrate the scale-up characteristics of the parallel algorithms. The resulting parallel efficiency was excellent (> 90%). These results indicate that the Monte Carlo modelling approach is sufficiently robust to handle kinetically coupled systems, and that the approach is highly parallel.

Exact dynamic load balancing of MIMD architectures with linear programming algorithms

We discuss the global dynamic load balancing problem of parallel MIMD architectures with arbitrary topologies. The features of computational algorithms which must be taken into account applying our approach are indicated. The central part of this method is a linear optimization problem (linear programming). It is well known that for resolving such problems one has the polynomial algorithms discovered by Hachijan in 1979. Hence, we describe the exact polynomial load balancing algorithm for some computing models. A sample of applying our approach for an adaptive fluid dynamics algorithm is presented.

Parallel Block Matrix Factorizations on the Shared-Memory Multiprocessor Ibm 3090 VF/600J

Efficient parallel block algorithms for the LU factorization with partial pivoting, the Cholesky factorization, and the QR factorization transportable over a range of parallel MIMD architectures are presented. Parallel implementations of different block algorithms that utilize optimized uniprocessor level-3 BLAS are compared with corresponding routines of LAPACK (under development). Parallelism is mainly invoked implicitly in LAPACK by replacing calls to uniprocessor level-3 kernels by calls to parallel level-3 kernels and thereby maintaining portability. However, by parallelizing at the block level (explicitly) it is possible to overlap and pipeline different matrix-matrix operations and thereby gain some performance. Theoretical models give upper bounds on the best possible speedup of the explicitly and implicitly parallel block algorithms for the target machine.