摘要:

Monte Carlo simulations are a powerful tool in statistical mechanics and particle physics. In the latter field they are the leading technique for computing experimental properties of strongly interacting particles from first principles in quantum chromodynamics (QCD). However, due to their enormous demand on compute: power, these simulations remain a challenge to both physicists and computer scientists. To overcome the problems, intensive research on simulation algorithms and good implementations on powerful supercomputers are necessary. In this talk we report on the optimization of a Monte Carlo simulation program forSU(3) gauge theory to run on the currently most powerful vector and vectorparallel systems. We focus on a CRAY Y-MP/C90 implementation that achieves 8.3 GFLOPS for 16 processors.

ISSN:1063-7192    

DOI:10.1080/10637199408962535

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

The problem of scheduling task graphs on multiprocessor systems is known to be NP-complete in its general form as well as many restricted cases. Few polynomial algorithms have been developed for solving special cases of this problem when the communication cost is ignored. The complexity of the problem rises even further when communication among tasks is considered. Several different models have been used by researchers to compute the communication cost. The complexity of the problem changes based on which cost model is used to estimate the communication cost Several versions of the scheduling problem with communication were proven to be NP-complete using different models [10, 11]. In this paper, we first Survey the different communication models. Focusing on one of these models, we study the effect of considering communication on the relationship between the two-processor scheduling problem and the problem of finding maximum matching in the complement of the task graph We introduce the idea of augmented task graphs and show that in some cases scheduling an augmented graph without considering communication is equivalent to scheduling the original task graph with communication. We show that the problem of scheduling an arbitrary task graph with communication on two processors can be reduced to the problem of constructing an augmented graph. We introduce a polynomial algorithm to optimally schedule a tree-structured task graph with communication on two processors by the construction of its augmented graph e also prove the optimally of the introduced algorithm.

ISSN:1063-7192    

DOI:10.1080/10637199408962536

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

This paper present., the systolic implementation of Huang's direct method for the general solution of a system of linear equations. The method is extended to the solution of general matrix equations and generalised matrix inversion. Two alternative systolic implementations are described, one based on a systolic ring architecture and another based on a re-usable systolic array.

ISSN:1063-7192    

DOI:10.1080/10637199408962537

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In this work PFI (Perturbed Functional Iterations) has been extended to solve large-scale nonlinear models by applying parallel computations. PFI partially linearizes a given nonlinear system, and irrespective of the physical dimension of the model it solves in parallel a sequence of linear equations of significantly smaller order to compute perturbation parameters and adds them in parallel to nonlinear Jacobi iterations to compute new iterates. As convergence is approached all linearizations are damped out, restoring thereby nonlinear properties of the model near the root. This generates a high degree of accuracy. In comparison with other nonlinear algorithms, PFI has a simple algorithm which is easy to program. Some applications have produced encouraging results.

ISSN:1063-7192    

DOI:10.1080/10637199408962538

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

This paper is concerned with the solution of block tridiagonal linear systems by the preconditioned conjugate gradient (PCG) method. If we consider a block AGE splitting of the coefficient matrix, it is possible to derive an additive polynomial preconditioner and to give conditions for such preconditioner to be symmetric positive definite.

ISSN:1063-7192    

DOI:10.1080/10637199408962539

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In this paper the algorithms of the Block Vectorizable Preconditioned iterative Method for linear systems of the formA× = fare proposed, whenAis block tridiagonal matrix. The convergence of these iterative methods is analysed, whenAis anMmatrix orHmatrix- The resulting BVPI method has been tested on a YH-1 computer. Numerical examples indicates that the new methodisvery efficient, since the vectorial computational can be applied.

ISSN:1063-7192    

DOI:10.1080/10637199408962540

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

We present the parallel implementation on a transputer network of two preconditioned Lanczos-type schemes for the solution of unsymmetric sparse Linear systems: Bi-CGSTAB and TFQMR Solution of a number of sample tests shows that Bi-CGSTAB with ILU preconditioning is generally Faster than TFQMR, but do not achieve. a high degree of parallelize!ion. Better paraLlelization is obtained with the diagonal preconditioner, which uses less memory and allows therefore the solution of larger problems. With this preconditioner the TFQMR algorithm is found to be more robust than Bi-CGSTAB, but not always globally more efficient. The speed Ups obtained with 8 processors in this case reached a maximum of approximately 6.82, showing the high degree of parallelization achievable with the diagonal preconditioner. As in most massively parallel scenarios, however, the attainable speed ups are strongly dependent on the problem granularity.

ISSN:1063-7192    

DOI:10.1080/10637199408962541

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

France

ISSN:1063-7192    

DOI:10.1080/10637199408962542

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In this paper a parallel Romberg Integration procedure for the numerical evaluation of single and double integrals U described. The process is achieved by the use of a sub-rnanager to facilitate the data partitioning procedure.

ISSN:1063-7192    

DOI:10.1080/10637199408962543

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In this paper, we present a new and efficient parallel algorithm for the solution of linear equations based on Householder reductions, Unlike the existing methods, the new algorithm integrates both triangula-tion and back substitution phases, and replaces the back substitution phase by only one step division. Further, being based on the Householder method, it has unconditional numerical stability. We present a scheduling scheme for assigning the computational tasks in the algorithm onto processors for efficient implementation ( of the algorithm) on multiprocessor systems. Also, we discuss the error analysis and the performance of the algorithm on hypercube multiprocessor systems.

ISSN:1063-7192    

DOI:10.1080/10637199408962544

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor