Re: NaN in intel fortran 9.0 for windows integrated with VS.Net
- From: "R. MacDonald" <scitec@xxxxxxxxxxxxxxx>
- Date: Tue, 30 May 2006 14:42:59 +0200
Hello, bei,
It may be simplest if you raise an error when a floating point exception occurs. Look under Project/Properties/Fortran/Floating Point for the "Exception Handling" setting. Setting this to "Underflow gives 0.0; Abort on other IEEE exceptions" should quickly reveal the source of the problem.
On the other hand, if you are calling a Fortran DLL from .Net (instead of from a Fortran Main program) I think that the above will not work. You may need to introduce a call to FOR_SET_FPE within the Fortran code to enable the exceptions. (I seem to recall that there was a way to set this from the .Net host, but I've forgotten the details.)
Cheers,
Randy
bei wrote:
I am solving some numerical functions under intel fortran 9.0 for.
windows integrated with VS.Net .However, my results have a lot of NaN
which I don't know how to handle.Most of my variable are very
small,like e-0003.I am wondering if it's because of my small
variables.Or I have something wrong with my algorithm.If I want to
track the variable which changes to NaN first,how can I do?
Any suggestion and advise about this are highly appreciated.
Best,
Bei
- References:
- Prev by Date: Re: Beginer needing help
- Next by Date: Re: NaN in intel fortran 9.0 for windows integrated with VS.Net
- Previous by thread: NaN in intel fortran 9.0 for windows integrated with VS.Net
- Next by thread: Re: NaN in intel fortran 9.0 for windows integrated with VS.Net
- Index(es):
Relevant Pages
|