We are having difficulty debugging parallel computations based on OpenMP using Intel Fortran 16,17,18 and Visual Studio 2013 +.
We don't have these problems using Intel Fortran 14 and Visual Studio 2010.
Here are the compiler options we are using
/nologo /debug:full /MP /Od /fpscomp:ioformat /warn:declarations /real_size:64 /Qauto /module:"x64\Debug/" /object:"x64\Debug/" /Fd"x64\Debug\vc120.pdb" /traceback /check:bounds /libs:static /threads /dbglibs /c
/Qopenmp /Z7 /MP
We don't have any problems using the Release compilation and the parallel computing works fine.
What happens is that in debug the program never really goes into the second thread - instead all the parallel processes are working as process 1 and of course that causes the same variables to be used by all the processes and from there it crashes.
Any ideas what we are missing?
Thank you in advance