Liebe Gitlab-Nutzerin, lieber Gitlab-Nutzer,
es ist nun möglich sich mittels des ZIH-Logins/LDAP an unserem Dienst anzumelden. Die Konten der externen Nutzer:innen sind über den Reiter "Standard" erreichbar.
Die Administratoren

Dear Gitlab user,
it is now possible to log in to our service using the ZIH login/LDAP. The accounts of external users can be accessed via the "Standard" tab.
The administrators

Commit 18651aa8 authored by Thomas Witkowski's avatar Thomas Witkowski
Browse files

Added document with parallel parameters.

parent debb7eaf
\usepackage[a4paper, left=1.5cm, right=2cm, top=2cm, bottom=2cm]{geometry}
The label \texttt{[name]} represents the string that is passed to the
constructor of MeshDistributor. {\small
\textrm{\textbf{keyword}} & \textrm{\textbf{data type}} & \textrm{\textbf{default}} & \textrm{\textbf{description}} \\
\basis repartitioning & \textbf{bool} & [0] & En/disables
repartitioning of the macro mesh, when derivation of mean number of
DOFs exceeds a threshold value. \\\hline
\basis repartition ith change & \textbf{int} & [20] & Sets number of
mesh changes to wait before threshold check for repartitioning will
be performed.\\\hline
\basis partitioner & \textbf{enum}\{parmetis, zoltan, simple\} &
[parmetis] & Defines the external tool that performs the
partitioning of the, e.g.~by graph-partitioning. {\em simple} does
not change the initial partitioning, i.e., its a random distribution
of the macro elements to the processors.\\\hline
\basis box partitioning & \textbf{bool} & [0] & If the macro mesh is
globally refined from macro.stand.2d or macro.stand.3d, than the
partitioner may compute the partitioning not based on triangled or
tetrahedron, but on (composed) rectangles or boxes. Till now this is
implemented only for 3D and Zoltan partitioner. \\\hline
\basis log main rank & \textbf{bool} & [0] & If set to {\em true},
stdout output will be printed only by the main rank 0. Otherwise,
all ranks stdout output will be created. \\\hline
\basis pre refine & \textbf{int} & [-1] & If set to -1, the number
of pre refinements for the macro mesh will be calculated for the
given number of processors. This value can be overwritten by setting
a value $>= 0$. \\\hline
\basis output->serialization filename & \textbf{string} & [] & Name
of the parallel serialization file. If at least one stationary
problem is serialized, this parameter must be set. \\\hline
\basis input->serialization filename & \textbf{string} & [] & Name
of the parallel deserialization file. If at least one stationary
problem is deserialized, this parameter must be set. \\\hline
\basis debug output dir & \textbf{string} & [] & Path name where
debug data should be written to. \\\hline
\basis write parallel debug file & \textbf{bool} & [0] & If set to
{\em true}, the parallelization will create for each rank a file
with the name ``mpi-dbg-\{rank-no\}.dat''. This files contain
information about all DOF indices in ranks domain. They may be
useful for debugging or some postprocessing steps. \\\hline
\ No newline at end of file
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment