Digital Library
Close Browse articles from a journal
 
<< previous    next >>
     Journal description
       All volumes of the corresponding journal
         All issues of the corresponding volume
           All articles of the corresponding issues
                                       Details for article 100 of 105 found articles
 
 
  Synchronization-Free Parallel Implementation of Factored-Implicit and Explicit Techniques
 
 
Title: Synchronization-Free Parallel Implementation of Factored-Implicit and Explicit Techniques
Author: Varghese, A. N.
Raad, P. E.
Appeared in: International journal of computational fluid dynamics
Paging: Volume 6 (1996) nr. 2 pages 137-155
Year: 1996-04-01
Contents: This work is motivated by the desire to assess the adaptability of alternating direction implicit and explicit finite difference techniques, used in computationl fluid dynamics, to parallelization on tightly-coupled, shared-memory parallel computers. The aim of the work is to achieve synchronization-free parallel implementations by the introduction and use of load balancing and processor routing schemes. Two representative and popular finite difference techniques, the implicit Beam and Warming and the explicit MacCormack techniques, are parallelized and tested on a 20-processor computer. Several parallel implementations of increasing complexity are investigated in the context of the simple heat and the more computationally intensive curvilinear grid mapping equations. In attaining near-theoretical speedups, the need for processor synchronization has to be eliminated completely. To achieve a synchronization-free parallel implementation, three stratagems are introduced and evaluated. First, each finite difference technique is split into multiple computational phases. Second, the computational loads of the working processors are carefully balanced. Finally, efficient processor routing is applied in order to avoid creating regions of processor interference. The result of combining these three novel but intuitive approaches is the gain of a general parallelization philosophy that should be applicable to finite difference and volume techniques in general. Speedup values obtained for the two specific finite difference techniques chosen for this work are significant with the largest ones resulting from more complex equations, larger size grids, and more stringent convergence criteria.
Publisher: Taylor & Francis
Source file: Elektronische Wetenschappelijke Tijdschriften
 
 

                             Details for article 100 of 105 found articles
 
<< previous    next >>
 
 Koninklijke Bibliotheek - National Library of the Netherlands