It's follow-up question to:

C23 wording adds stronger assertion

ISO/IEC 9899:202x (E)

Floating types shall be able to represent zero (all fk == 0) and all normalized floating-point numbers (f1 > 0 and all possible k digits and e exponents result in values representable in the type

CodePudding user response：

If you speak about "floating point types" in general (and not about the type named float) when using the word "floats":

Yes:

If you design a compiler, you might define the non-standard type unsigned float with 24 mantissa bits, 8 exponent bits, no sign bit and 256 possible exponent values (instead of 254).

Such a floating point format would only be able to store positive values (not equal to zero) and infinite or NaN would also not be supported.

The advantage of such a format would be a wider range and/or a higher precision than a float data type with the same number of bits.