Embedding a 'Treecode' on a MIMD parallel computer using a domain decomposition paradigm
نویسنده
چکیده
\Tree-codes" and \Multigrid" methods are both hierarchical, recursive algorithms for solving vector eld equations, faster than traditional methods. If the methods are non-adaptive, then both Tree-codes and Multigrid methods require identical parallel communication strategies since they share the same internal structure. This paper contains a description of Tree-codes followed by an overview of one Tree-code in particular, namely the Greengard-Rokhlin Fast Multipole Method in 2 dimensions. The parallel version of this method is then discussed at length with the presentation of two communication strategies; An \All-to-all" broadcast and a routine which enables neighbouring transputers to swap information in a predetermined sequence. The computer used is the Meiko Computing Surface, a MIMD parallel computer employing T800 transputers running the CSTools communication harness using a SPMD (Single Program, Multiple Data), domain decomposition paradigm. The 3D versions of the communication algorithms have also been implemented and are simple extensions of the 2D routines.
منابع مشابه
Embedding a \treecode" on a Mimd Parallel Computer Using a Domain Decomposition Paradigm. 2. Introduction
1. Abstract \Treecodes" are hierarchical, recursive algorithms which reduce the computational time for the N-body problem from O(N 2), by the direct summation method, to O(N log N) or O(N). This paper contains a description of Treecodes followed by an overview of one Treecode in particular, namely the Greengard-Rokhlin Fast Multipole Method in 2 dimensions. The parallel version of this method i...
متن کاملParallel Community Climate Model 32 Ncar
This paper is a brief overview of a parallel version of the NCAR Community Climate Model, CCM2, implemented for MIMD massively parallel computers using a message-passing programming paradigm. The parallel implementation was developed on an Intel iPSC/860 with 128 processors and on the Intel Delta with 512 processors, and the initial target platform for the production version of the code is the ...
متن کاملNumerical Study of Three - Dimensional Flowusing Fast Parallel Particle Algorithms
Numerical studies of turbulent ows have always been prone to crude approximations due to the limitations in computing power. With the advent of supercomputers, new turbulence models and fast particle algorithms, more highly resolved models can now be computed. Vortex Methods are grid-free and so avoid a number of shortcomings of gridbased methods for solving turbulent uid ow equations; these in...
متن کاملParallel Semiconductor Device Simulation: from Power to ‘Atomistic’ Devices
This paper discusses various aspects of the parallel simulation of semiconductor devices on mesh connected MIMD platforms with distributed memory and a message passing programming paradigm. We describe the spatial domain decomposition approach adopted in the simulation of various devices, the generation of structured topologically rectangular 2D and 3D finite element grids and the optimisation ...
متن کاملSingular Value Decomposition based Steganography Technique for JPEG2000 Compressed Images
In this paper, a steganography technique for JPEG2000 compressed images using singular value decomposition in wavelet transform domain is proposed. In this technique, DWT is applied on the cover image to get wavelet coefficients and SVD is applied on these wavelet coefficients to get the singular values. Then secret data is embedded into these singular values using scaling factor. Different com...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Future Generation Comp. Syst.
دوره 11 شماره
صفحات -
تاریخ انتشار 1995