CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Question about ghost cells in parallel processing (http://www.cfd-online.com/Forums/main/10927-question-about-ghost-cells-parallel-processing.html)

vishwas March 5, 2006 14:17

Question about ghost cells in parallel processing
 
I have a 3-D grid in which there are 98 blocks. This grid was created using Gridgen. It will be processed using the INCA code prepared by Amtec engineering. It will be processed on one of the linux clusters at Sandia labs. I do not know any thing about the cluster and parallel processing. All I know is there a two ghost cells to be created at the boundaries of the blocks. But my question is are the ghost cells created at all the boundaries for all the blocks for parallel processing?

or

Ghost cells are created only at the boundary of the grid?

or

Ghost cells are created at all the boundaries of the blocks except for the boundaries of the grid?

Thank you

vasanth March 7, 2006 13:33

Re: Question about ghost cells in parallel process
 
Hello friend . There are ghost cells where ever there is a boundary, weather block or grid. Solves two purposes:

1)get data from neighbouring blocks

2)impose boundary conditions


Mani March 9, 2006 22:57

Re: Question about ghost cells in parallel process
 
that should not be understood as a general answer, though. it very much depends on the code, and maybe vasanth's answer is true for your particular code. other codes may use ghostcells only for multiblock communication and specify boundary conditions otherwise. although a very common technique, ghostcells are not the only way to impose boundary conditions and deal with multiblock interfaces.

vasanth March 12, 2006 22:46

Re: Question about ghost cells in parallel process
 
So mani what are the other ways of dealing with interfaces?

Thanks


All times are GMT -4. The time now is 11:07.