Real, double precision, and quadruple precision number data elements are represented according to the IEEE standard by the following form, where f is the bits in the fraction. Quad is SPARC only.
(1)^{sign} * 2^{exponentbias} *1.f
Table C1 Floatingpoint Representation

Single 
Double 
Quadruple 

Sign 
Bit 31 
Bit 63 
Bit 127 
Exponent 
Bits 3023 Bias 127 
Bits 6252 Bias 1023 
Bits 126112 Bias 16583 
Fraction 
Bits 220 
Bits 510 
Bits 1110 
Range approx. 
3.402823e+38 1.175494e38 
1.797693e+308 2.225074e308 
3.362E4932 1.20E+4932 