Re: [GD-General] FLT_MIN
Brought to you by:
vexxed72
From: Colin F. <cp...@ea...> - 2003-01-03 20:58:43
|
> Ok, I found this in the MSDN : > _control87( _EM_INVALID, _MCW_EM ); // DENORMAL is unmasked by this call Hey, that's neat! > As long as it works well, I think I'm going to > allow denormalized numbers. Any known problems > with this ? Loss of precision. 1.4E-45 has one bit of precision! So, in this extreme case you can have up to 50% error in precision, but at least the order of magnitude is still sane. You can have exact powers of two all the way down to (2^(-149)) without loss of precision of course. > That's a very subtle (and interesting IMHO) bug that happens only because of > a combination of bad things : Yes, this is very interesting. I didn't even know about denormalized floats until you mentioned this mystery and I did some research, but now I can see that it is something to be aware of. --- Colin |