#############################################
How to run tests:
#############################################

 - go to ~/trunk/job/regression_tests/ directory

 - run RT.sh script

 - wait about three hours (and growing)
   progress should appear on the screen

 - look in RegressionTests.log for results

#############################################
How to create new baselines :
#############################################

 - use RT.sh script with one argument: nmm, gfs or all
   ./RT.sh nmm - will create baselines for NMMB
   ./RT.sh gfs - will create baselines for GFS
   ./RT.sh post- will create baselines for post
   ./RT.sh all - will create baselines for both GFS & NMMB

 - results will be stored on /stmp/${LOGIN}/REGRESSION_TEST directory
   (new /stmp/${LOGIN}/REGRESSION_TEST directory will be ready to
    replace current REGRESSION_TEST directory)

#############################################
Tests performed:
#############################################

    NMMB, global, control version
    NMMB, global, test NEMSIO input file
    NMMB, global, test restart
    NMMB, global, test restart with NEMSIO file
    NMMB, global, test different decompostition
    NMMB, global, test threads
    NMMB, global, test with GFS physics

    NMMB, regional, control version
    NMMB, regional, test NEMSIO input file
    NMMB, regional, test restart
    NMMB, regional, test restart with NEMSIO file
    NMMB, regional, test different decompostition
    NMMB, regional, test threads
    NMMB, regional, test with GFS physics
    NMMB, regional, test nesting
    NMMB, regional, test precipitation adjustment
    NMMB, regional, test writing time series

    GFS, control version (32 processors, 1 thread)
    GFS, 2-copy option
    GFS, different decompostition (48 processors, 1 thread)
    GFS, test threading (16 processors, 2 threads)
    GFS, test single processor version (1 processor)
    GFS,  1 proc, 1 thread, no quilt, nsout=1
    GFS, 16 proc, 2 threads,no quilt, output every 2 time steps
    GFS, 48 proc, 1 thread, no quilt
    GFS, 32 proc, 1 thread, no quilt, output every 4 timestep
    GFS, 32 proc, 1 thread,    quilt, digital filter on reduced grid
    GFS, 16 proc, 2 thread,    quilt, 2x2 wrt pe, digital filter on reduced grid
    GFS,  1 proc,           no quilt, digital filter on reduced grid
    GFS, use the OPAC climo scheme for SW and LW
    GFS, 32tasks, 2threads, quilt, dfi3hr, reduced grid, NDSL
    GFS, 16tasks, 2 thread, quilt,2x2 wrt pe, HYB 2loop digital filter on reduced grid
    GFS, 16tasks, 2 thread, quilt,2x2 wrt pe, HYB 2loop digital filter on reduced grid with nst
    GFS, 16tasks, 2 thread, quilt,2x2 wrt pe, HYB 1loop digital filter on reduced grid
    GFS, Concurrency GEFS, stochastic perturbations, 4 members, T190L28