Are There Standardized Fixed-Size Floating Point Types in C and C ?

Fixed-Size Floating Point Types

In the quest for precise representation and efficient storage, the standardization of fixed-size floating point types has been a topic of discussion in the C and C communities. While C99, boost/cstdint.hpp, and C 0x provide fixed-size integer types like int32_t, the question arises: are there analogous fixed-size floating point types?

The Current Standard

The answer is a resounding no. Currently, neither the C nor C standards offer fixed-size floating point types. This is in contrast to integer types, where the presence of int32_t and similar types ensures a consistent representation across different platforms.

The Absence of float32_t

The lack of a type like float32_t is particularly noteworthy because float is not guaranteed to be a binary floating-point format in either C or C . Depending on the compiler, float might or might not adhere to the IEEE-754 32-bit binary format.

Compiler Guarantees and Caveats

In practice, float is often implemented as the IEEE-754 single type on many non-embedded platforms. However, it's crucial to be aware that some compilers may evaluate expressions in a wider format, potentially introducing discrepancies in numerical operations.

Future Developments

Efforts are underway to address this issue. A working group is exploring the addition of C language bindings for the 2008 revision of IEEE-754. If such a step were taken in C, it is likely that the C standard would eventually follow suit, paving the way for standardized fixed-size floating point types in the future.

The above is the detailed content of Are There Standardized Fixed-Size Floating Point Types in C and C ?. For more information, please follow other related articles on the PHP Chinese website!