This directory contains the OpenMP version of the LU benchmark.
The benchmark solves a lower-upper tridiagonal system of equations using 
the SSOR scheme. The directory includes three approaches in parallelizing
the SSOR scheme, selectable via the make option VERSION:

   VERSION=          the pipelined version (the default)
   VERSION=hp        the hyperplane version
   VERSION=doac      the hyperplane version using OpenMP DOACROSS

Special note on feature requirements:
   the pipelined version - ATOMIC read/write from OpenMP 3.0
   the "doac" version - DOACROSS feature from OpenMP 4.0.

To select different approaches, use the option VERSION for make:

   % make CLASS=<class>               # build the pipelined version
   % make CLASS=<class> VERSION=hp    # build the hyperplane version
   % make CLASS=<class> VERSION=doac  # build the doacross version

where <class> is one of [S,W,A,B,C,D,E,F].
