The float can be converted to well known single-precision IEEE 754 number, why 754? It's the standard representation of single-precision numbers, used in net transmission.
Compares to send the float as text or send/receive without re-formatting/un-formatting, it's much faster & safer.
What confuses me is that nowadays floats are stored just what IEEE 754 says to be. Or maybe windows only?
But the most important thing is, when sending/receiving data/numbers, use standards.
See this page for more info.
I'll explain this format a little. It uses the first bit as a sign representation(1 if negative), the following 8bit as its exponential part, the last 23bit as its fraction part.
See this example:
1 10000001 10110011001100110011010
sign expo mantissa
It is divided as stated above.
The Microsoft's floats is following this standard. See the following definition of float in Windows.
Where float uses 8 bits of exponent and 23 bits of mantissa, compares to double's 11 bits of exponent and 52 bits of mantissa.
About the bias refered to in the page above:
Because exponents are stored in an unsigned form, the exponent is biased by half its possible value.For type float, the bias is 127; for type double, it is 1023.You can compute the actual exponent value by subtracting the bias value from the exponent value.
/QUOTE
Note that this exponent's base is not 10, of course it's 2! So 2E127 is approximately 10E38, that's why float's range is about 1E-38~3E38!
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
