TauBench.
========

This is the alpha release of a unstructured grid benchmark. TauBench
is a pseudo benchmark. The respective kernels are derived from Tau --
a Navier Stokes solver which has been developed at the german aerospace   
research DLR in germany. The original flow solver is a three-dimensional
parallel hybrid multigrid solver, which uses a finite volume scheme in
order to solve the Reynolds-averaged Navier-Stokes equations. 
The benchmark itself doesn't do anything useful in terms of producing 
meaningful results - rather it is meant to mimic the run-time
performance of the Tau solver. 

INSTALL
=======

configure; make; make run; make gr might do the trick. 
The mpi version is extracted from the path of mpicc. 
Setting the MPICC/CFLAGS environment variable also works.

make run; make gr currently only supports PBS. The
first make file target submits the jobs to the batch system.
If all runs complete the second makefile target evaluates 
the results.

If you have a different batch system you need to adapt
the corresponding files. 

Since Tau supports both vector and cache colored
grids, you might also need to adapt the compiler directives in nodep.h and
expand.h.

The benchmark itself can be used with two flags, the gridsize per process
and the number of pseudo steps:

mpirun -np 2 ./taubench -n 100000 -s 10


