I don't really understand why the code would be spending a significant
amount of time in "isinf", but if it is then probably you want to
inline it by moving the definition to nlopt-util.h (as "static inline
nlopt_ininf(...) { ... }" or a macro).
On Jul 2, 2012, at 6:21 PM, Benoit Scherrer wrote:
Sorry that should work:
#ifdef _MSC_VER
#include <direct.h>
#endif
int nlopt_isinf(double x) {
#ifdef HAVE_ISINF
return isinf(x);
#elif defined(_MSC_VER)
return _finite(x)==0;
#else
return fabs(x) >= HUGE_VAL * 0.99 ;
#endif
}
On Mon, Jul 2, 2012 at 6:03 PM, Benoit Scherrer <[email protected]
> wrote:
Hi,
I'm using nlopt in a project, using the BOBYQA optimization.
I did a runtime profiling of my code with Visual Studio (so under
Windows) and
surprisingly one of the bottlenecks (after my objective function)
was nlopt_isinf
(see screenshots)
A fix might be :
#ifdef MSVC
#include <direct.h>
#endif
int nlopt_isinf(double x) {
#ifdef HAVE_ISINF
return isinf(x);
#elif MSVC
return _finite(x)==0;
#else
return fabs(x) >= HUGE_VAL * 0.99 ;
#endif
}
I Don't know if _finite exists for all WIN32 or only MSVC
B
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
