EXC_ARITHMETIC when writing a floating point variable

Dominique Dhumieres dominiq@lps.ens.fr
Sun Oct 31 16:38:00 GMT 2010


Richard,

Usual comment: give a nonworking reduced test case!

You can also use -Wall and -Wuninitialized (with -O
for the later). Try to find the minimal set of options
tha trigger the problem: do you really need
-fdefault-double-8 -fdefault-integer-8 -fdefault-real-8?
If yes, are not you passing REAL variables to procs
expecting a REAL(4) (old timing procs are good candidates)?

You can also use valgrind on Mac OS 10.6.4 (I build it
with fink, but it may exist on macports).

If none of the above works, you can mail me the code
and I can have a look.

Cheers,

Dominique



More information about the Fortran mailing list