fluent并行（Parallel fluent）.doc

fluent并行（Parallel fluent） Introduction to parallel computing Fluent parallel computing is the use of multiple computing nodes (processors) simultaneously computing. Parallel computing can segment the mesh Into multiple subdomains, the number of subdomains is an integer multiple of the computational node (e.g., 8 sub domains corresponding to 1, 4, and computed nodes). 2, 8 Each subdomain (or subset of subdomains) will reside on different computational nodes. It may be the computing node of the parallel machine, Or programs running on multiple CPU platforms, or running on different platforms connected by a network (UNIX flat) A program on a platform or Windows platform. The increase of computing information transmission rate will lead to the reduction of parallel computing efficiency In parallel computing, it is important to choose the solution. The recommended steps for running parallel Fluent are as follows: 1. open parallel solver, select the number of nodes calculated, see section 28.2 and 28.3. 2. reads into the case file, and Fluent automatically divides the mesh into several subdomains. Make it best to break up after building a problem, Because this segmentation is related to the computational model (like non conformal contact surfaces, sliding mesh, and shell-conduction) Adaptive encapsulation. If your case file contains a sliding mesh, or in the calculation process In order to modify the non - uniform contact surface, the serial solver is used to segment the non - contact surface. There are other ways to do segmentation, such as manual segmentation on a serial or parallel solver. 3. examine the segmentation area carefully and, if necessary, re segmentation, see section 28.4.5, how to check the segmentation area. 4. for calculations, see how to check and improve parallel computing in section 28.5. 28.2 open parallel solver The way to turn on the Fluent parallel solver depends on whether the operating platform is a dedicated parallel machine or a workstation. 2