Problems at 120 GeV

I need to produce kaons and pions near 0 deg. for 120-GeV protons on a
platinum cylinder 6 cm long. I use a geometry similar to many successful
applications at lower energies. In running 40 cpu cores at once, with
different random number seeds, some make it through while others bomb out.

The input file for one of the failures is attached, along with the .err
file. The only code modification is in the attached mgdraw routine, which
is used for output to post-processing.

I can live with warning messages in a file that I will eventually delete.
But what do I need to do to prevent the failures? I went from 'Precision'
to 'New-defaults' and then turned EMF off. These prevented problems from
showing up sooner, but I still get failures.

Thank you,
Joe