Build MCNPX 2.5.F with 64-bit integers:

The 64-bit integer capability is now supported on all platforms,
however it has not yet been tested with all compiler/OS
combinations. The Unix and Windows builds are a bit different:

(1) Unix systems. To date, this has been tested on SGI,
    Compaq/DEC, HP, and Linux. It is invoked with the following
    configure options:

    ../v25f/configure --with-FFLAGS="-i8" --with-NOCHEAP ...

    Note, on most systems the compiler flag is "-i8"; however
    on some systems it is different (e.g., on AIX the flag is
    -qintsize=8 and on SunOS it is -xtypemap=integer:64). On
    the SunOS you must also use the -dalign and -O0, and 64-bit
    object code is possible with the -xarch-v9. Although the
    NOCHEAP option should remove the CHEAP CPP directive, it
    is always best to check the src/mcnpx/mcnpf/Makefile.h file
    in the build tree to make sure the -DCHEAP=1 directive was
    indeed removed from the DEFS variable. Run "gmake" as usual.

(2) Windows systems. This requires the Intel compiler, version
    8.0 or later. Follow the instructsions in the README.W32 file
    (top directory of the source tree). Replace the configure
    command with:

    ..\v25f\configure I8

It should be noted that on all systems the binary cross-section
data must be reprocessed with MAKXS to generate new binary data
that contains 8-byte integer data. If you use Type 1 (ASCII) data,
no change is required. If you use Type 2 (binary) data, you must
reprocess the Type 1 data again, using the new MAKXS code that
will be created during the build process described above. If you
have deleted the Type 1 data, you can use your old MAKXS code
to convert the old binary data back to Type 1 data, then use the
new MAKXS code to reconvert the data back to Type 2 data. You must
also be sure to use the new bertin and phtlib files created from
this 8-byte integer build.