em solver code
I am interested in this topic.
Anyone tried?
I know FDTD can be implemented
parallelly. But I donot have experience
to use it.
Div
There is web site on this...Search under parallel fdtd...
Other methods can also take advantage of parallel computing. I know FEKO (a MoM code) has this feature implemented.
Why don't you try a peak at what the QWED engineers are doing in this field. www.qwed.com.pl
That is for an FTDT software.
Things are easier for MOM software house take a look at the approach used by Zeland to use a computer network for one simulation project.
www.zeland.com
HFSS is also capable of using more then one CPU.
SuperNEC from Poynting can use the PVM library (MPI) to speed up computations. It isn't very difficult to paralelize MoM codes using MPI because the computation of the MoM matrix elements is a easily paralelized (we have done it in our code and I'm not talking about SuperNEC) Using a paralelized solver is as esay as goto netlib ang get one of the LAPACK relatives. Things aren't so easy when you have to face with the million dolar question
?Where can I store my F-16 MoM matrix?
Iterative/multilevel methods are the future, MoM is in a dead road when facing thousand-wavelengths problems.
Is it possible to share your available code?
I am interested....