B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, University of Manchester Institute of Science and Technology, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....P is the number of processing nodes. 2.4 Benchmark Applications The applications are summarised in Table III, along with the problem sizes used in this work. GE implements a Gaussian Elimination algorithm with pivoting. CFD is a computational uid dynamics application modelling laminar ow [43]. The remaining six applications were taken from Stanford s Splash 2 suite [48] and were selected to give a representative cross section of scienti c and engineering shared memory applications. The uid dynamics situation modelled in CFD is laminar ow in a 2D cavity with a sliding lid. The lid ....

....the cavity, which introduces a zone of re circulatory uid. If the analysis grid of the cavity is suciently ne, then small counter rotating eddies should be observed in the corners of the cavity. The CFD algorithm was originally implemented on a distributed memory message passing architecture [43], but the version used in this paper had already been adapted to run as a shared memory application and has been optimised to reduce false sharing [39] GE is a simple Gaussian elimination application, similar to that used by Bianchini and LeBlanc in their study of eager combining [8] At the end ....

B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, University of Manchester Institute of Science and Technology, 1993.

....The caches are kept coherent using an invalidation based, distributed directory protocol using singlylinked lists [9] The benchmark applications are summarised in Table 1. GE implements a Gaussian Elimination algorithm [2] CFD is a computational uid dynamics application modelling laminar ow [8]. The remaining six applications were taken from Stanford s SPLASH 2 suite [10] The adaptive proxies scheme adjusts the proxying period according to the level of congestion at individual home nodes. However it has the storage overheads of holding the LB (x;y) PP (x;y) PP unit , PPmax , and ....

B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, UMIST, 1993.

....elimination program, similar to that used by Bianchini and LeBlanc in their study of eager combining [2] We chose this benchmark because it is an example of widely shared data. cfd is a computational fluid dynamics application, modelling laminar flow in a square cavity with a lid causing friction [13]. We selected six applications from the splash 2 suite, to give a cross section of scientific shared memory applications [16] We used both Ocean benchmark applications, in order to study the effect of proxies on the tuned for data locality and easy to understand variants. Other work which ....

B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, University of Manchester Institute of Science and Technology, 1993.

....grid of the square cavity is sufficiently fine, then small counter rotating eddies should be observed in the corners of the cavity. 76 5. Tuning substantial applications 5.2. 2 Structure of the parallel application The cfd program was originally implemented on a distributed memory architecture [32], but the version used in this project had already been adapted to run as a shared memory application. The program is written in fortran, so it is converted to C as part of the process of instrumenting it to run with the alite simulator (see Appendix A) The structure of the program is illustrated ....

B. A. Tanyi and R. W. Thatcher. Iterative solution of the incompressible navier-stokes equations on the meiko computing surface.

....is a laminar flow in a square cavity with a lid which slides across the cavity introducing a zone of re circulatory fluid. cfd uses barriers to ensure that CPU 1 has updated global variables before all the processors move on to the next stage, but there are long periods without a barrier [6]. Time slot analysis was therefore appropriate, and time slot length was chosen to give around 100 slots over the execution time, i.e. to give a reasonably detailed profile without being swamped by too much information. The phase level graphs generated by clarissa are shown in Fig. 4. The graphs, ....

B. A. Tanyi. Iterative Solution of the Incompressible Navier-Stokes Equations on a Distributed Memory Parallel Computer. PhD thesis, UMIST, 1993. This article was processed using the L a T E X macro package with LLNCS style

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC