May 6th, 2009

Much ado about almost nothing

OK, can anyone explain why FLT_MIN in Microsoft Visual C++'s float.h is 1.175494351e-38 which is not actually the smallest normalised positive IEEE single precision floating point number which, as far as I can see, is (roughly) 1.401298e-45?

EDIT: never mind, my interpretation of the web page I was looking at was wrong, FLT_MIN IS the smallest normalised positive value, 1.401298e-45 is the smallest positive denormalised value.