Hello Andreas,
AR> It is possible (though require specific testing) that the resulting
AR> floating point values cause underflow or similar fp exceptions and may
AR> be handled specially. Only a well-defined benchmark could tell. But just
AR> out of curiosity, what processor are you running?
Pentium M 1.86GHz under XP SP2. Latest VM by you but similar results
on the old 3.7 VM
Would you look into it if I try to cook up some benchmarks or at least
criticize the benchmarks to get them better defined?
Sorry, this went private.
Cheers,
Herbert mailto:
[hidden email]
Locked
