Multitasking for Local Parallelism in Applications To Chemically Reacting Supersonic Flows On Cray Y-Mp

This article is concerned with research on parallel pro cessing for chemically reactive supersonic flows. Effi ciency of autotasking and microtasking is examined for the problems of linear algebra and two-dimensional Euler equations with the finite element method. For microtasking, satisfactory improvements in speedup (2.63 for 4 CPUs and 1.67 for 2 CPUs) are achieved, with the load balance and the amount of parallelism being the dominant factors for the speedup. Autotask ing has advantages in providing automatic paralleliza tion, but it has been shown that in the solution of finite element Euler equations based on conventional algo rithms, parallelization by means of user's intervention is required for significant speedups (3.35 for 4 CPUs and 1.79 for 2 CPUs). A much better performance may be achieved with a combination of autotasking and microtasking. These results are verified from various example problems of linear algebra and supersonic diffuser flow field calculations with hydrogen-oxygen chemical reactions.

[1]  Ahmed K. Noor,et al.  Finite element dynamic analysis on CDC STAR-100 computer , 1979 .

[2]  Douglas J. Theis.  Vector supercomputers , 1974, Computer.

[3]  T. Chung,et al.  Shock waves/turbulence interactions in reacting flows , 1991 .

[4]  Dennis C. Jespersen,et al.  Computational fluid dynamics on a massively parallel computer , 1989 .

[5]  David J. Silvester.  Optimising finite element matrix calculations using the general technique of element vectorisation , 1988, Parallel Comput..

[6]  Julie M. Swisshelm.  A parallel finite-difference method for computational aerodynamics , 1989 .

[7]  R. C. Rogers,et al.  Using a global hydrogen-air combustion model in turbulent reacting flow calculations , 1983 .

[8]  Dennis C. Jespersen,et al.  A Computational Fluid Dynamics Algorithm On a Massively Parallel Computer , 1989, Int. J. High Perform. Comput. Appl..

[9]  Matthias Kratz.  Vectorized finite-element stiffness generation: tuning the Noor - Lambiotte algorithm , 1984, Parallel Comput..

[10]  Friedel Hossfeld,et al.  Multitasking: experience with applications on a CRAY X-MP , 1989, Parallel Comput..

[11]  Jack J. Dongarra,et al.  Unrolling loops in fortran , 1979, Softw. Pract. Exp..

[12]  F. Gustavson,et al.  Implementing Linear Algebra Algorithms for Dense Matrices on a Vector Pipeline Machine , 1984 .

[13]  Mark Bromley,et al.  Extremely fast finite difference techniques for the Connection Machine , 1991 .

[14]  Dennis C. Jespersen,et al.  Explicit and implicit solution of the Navier-Stokes equations on a massively parallel computer , 1988 .

[15]  R. Janßen.  A note on superlinear speedup , 1987, Parallel Comput..

[16]  Jack Dongarra,et al.  Implementing dense linear algebra algorithms using multitasking on the CRAY X-MP-4 (or approaching the Gigaflop) , 1986 .

[17]  K. Fong,et al.  Some linear algebraic algorithms and their performance on CRAY-1 , 1977 .

[18]  Charbel Farhat,et al.  Solving Navier- Stokes Equations On a Massively Parallel Processor: Beyond the 1 Gflop Performance , 1990, Int. J. High Perform. Comput. Appl..

[19]  Wolfgang E. Nagel.  Using multiple CPUs for problem solving: experiences in multitasking on the CRAY X-MP/48 , 1988, Parallel Comput..

[20]  T. J. Chung,et al.  FINITE ELEMENT ANALYSIS IN COMBUSTION PHENOMENA , 1987 .

[21]  Vance Faber,et al.  Comments on the paper "parallel efficiency can be greater than unity" , 1987, Parallel Comput..

[22]  Wolfgang E. Nagel.  Exploiting autotasking on a CRAY Y-MP: an improved software interface to multitasking , 1990, Parallel Comput..

[23]  J. M. Ortega,et al.  The ijk forms of factorization methods I. Vector computers , 1988, Parallel Comput..