`3d_turb_mixing` not working

[sbryngelson]

Example case 3d turbulent mixing https://github.com/MFlowCode/MFC/blob/master/examples/3D_turb_mixing/case.py is not working on GPU with error:

[C]wingtip-gpu3: sbryngelson3/MFC $ ./mfc.sh run --gpu -n 2 ./examples/3D_turb_mixing/case.py -t pre_process
mfc: Found CMake: /nethome/sbryngelson3/MFC/build/cmake/bin/cmake.
mfc: OK > (venv) Entered the Python virtual environment.
      ___            ___          ___
     /__/\          /  /\        /  /\       sbryngelson3@wingtip-gpu3 [Linux]
    |  |::\        /  /:/_      /  /:/       ---------------------------------
    |  |:|:\      /  /:/ /\    /  /:/        --jobs 1
  __|__|:|\:\    /  /:/ /:/   /  /:/  ___    --no-mpi
 /__/::::| \:\  /__/:/ /:/   /__/:/  /  /\   --gpu
 \  \:\~~\__\/  \  \:\/:/    \  \:\ /  /:/   --no-debug
  \  \:\         \  \::/      \  \:\  /:/    --targets pre_process
   \  \:\         \  \:\       \  \:\/:/     ---------------------------------------------------
    \  \:\         \  \:\       \  \::/      $ ./mfc.sh [build, run, test, clean, count] --help
     \__\/          \__\/        \__\/

Run
  Acquiring ./examples/3D_turb_mixing/case.py...
  Configuration:
    Input               ./examples/3D_turb_mixing/case.py
    Job Name      (-#)  MFC
    Engine        (-e)  interactive
    Nodes         (-N)  1
    Tasks (/node) (-n)  2
    MPI Binary    (-b)  mpirun
  Generating input files...
  Building pre_process:

    $ cmake --build /nethome/sbryngelson3/MFC/build/pre_process --target pre_process -j 1 --config Release

[100%] Built target pre_process

    $ cmake --install /nethome/sbryngelson3/MFC/build/pre_process

-- Install configuration: "Release"
-- Installing: /nethome/sbryngelson3/MFC/build/install/bin/pre_process

  Running pre_process:

    $ /nethome/sbryngelson3/MFC/build/install/bin/pre_process

 Pre-processing a 255x255x255 case on 1 rank(s)
 x_cb.dat in  contains non-positive cell-spacings. Exiting ...
Warning: ieee_inexact is signaling
    1

It presumably also fails on CPU since pre-process shouldn't be using the GPUs. Still working on testing that.

[sbryngelson]

update: this case does work on CPUs, it just uses more memory than was available on 2 GPUs I suppose. regardless, the variable weno_vars is no longer used so it needs to be removed from case.py @lee-hyeoksu .

[hyeoksu-lee]

I also had the memory issue even on CPUs when I use more than 320^3 grids. I think the subroutines to compute the instability waves take a lot of memory, so I am planning to check if I can reduce memory requirements in these subroutines.

I changed case.py to have coarser grids and weno_vars is removed from case.py. These changes are included in the pull request #151.