float_sig.c running, may die 
type float has about six significant digits 
eps is the smallest positive number added to 1.0f that changes 1.0f
type float eps= 5.960464478E-08 
type float small= 1.401298464E-45 
this could be about 1.0E-38, above shows good IEEE floating point
type float large= 1.701411835E+38 

type double has about fifteen significant digits 
type double eps= 1.11022302462515654E-16 
type double small=4.940656458E-324 
this could be about 1.0E-304, above shows good IEEE floating point
type double large=8.988465674E+307