[comp.lang.fortran] Fortran, 386, Inconsistent Results

comeau@hpuxa.ircc.ohio-state.edu (Donald C. Comeau) (07/18/90)

A friend of mine has written a program using the Lahey Fortran 32-bit
compiler and their DOS-extender.  When run on a Gateway 386 machine,
the program runs fine.  When run on either of two AST machines or a
Compaq it sometimes dies.

These deaths are inconsistent both by error and by time.  Sometimes it
is bad subscript.  Sometimes it is stack overflow.  Sometimes it is
after 10 or 20 minutes.  Sometimes after 8 hours.  When the program
does run to completion, the answers are usually correct, but once were
wrong.

So, my friend is looking for any ideas on how to investigate this
problem.  (I use Fortran, but on Suns and Crays.)  I would assume an
ordinary bug of some sort, except for the fact that it works fine on
the Gateway machine.

Any sugestions welcome, reasonable or not.  I prefer E-mail, but will
monitor these newsgroups.  Summaries (and final result) available by
request.

Thanks,

Don Comeau                      comeau@hydrogen.mps.ohio-state.edu
Dept of Chemistry               comeau@ohstpy.bitnet
Ohio State University           614-292-9431
-=-
Don Comeau                      comeau@hydrogen.mps.ohio-state.edu
Dept of Chemistry               comeau@ohstpy.bitnet
Ohio State University           614-292-9431