|
|
(C++20) | ||||
(C++23) | ||||
Integral powers of2 | ||||
(C++20) | ||||
(C++20) | ||||
(C++20) | ||||
bit_width (C++20) | ||||
Rotating | ||||
(C++20) | ||||
(C++20) | ||||
Counting | ||||
(C++20) | ||||
(C++20) | ||||
(C++20) | ||||
(C++20) | ||||
(C++20) | ||||
Endian | ||||
(C++20) |
Defined in header <bit> | ||
template<class T> constexprint bit_width( T x)noexcept; | (since C++20) | |
Ifx is not zero, calculates the number of bits needed to store the valuex, that is,\(1 + \lfloor \log_2(x) \rfloor\)1 + floor(log2(x)). Ifx is zero, returns zero.
This overload participates in overload resolution only ifT
is an unsigned integer type (that is,unsignedchar,unsignedshort,unsignedint,unsignedlong,unsignedlonglong, or an extended unsigned integer type).
Contents |
x | - | unsigned integer value |
Zero ifx is zero; otherwise, one plus the base-2 logarithm ofx, with any fractional part discarded.
This function is equivalent toreturnstd::numeric_limits<T>::digits-std::countl_zero(x);.
Feature-test macro | Value | Std | Feature |
---|---|---|---|
__cpp_lib_int_pow2 | 202002L | (C++20) | Integral power-of-2 operations |
#include <bit>#include <bitset>#include <iostream> int main(){for(unsigned x{}; x!=010;++x)std::cout<<"bit_width( "<<std::bitset<4>{x}<<" ) = "<< std::bit_width(x)<<'\n';}
Output:
bit_width( 0000 ) = 0bit_width( 0001 ) = 1bit_width( 0010 ) = 2bit_width( 0011 ) = 2bit_width( 0100 ) = 3bit_width( 0101 ) = 3bit_width( 0110 ) = 3bit_width( 0111 ) = 3
The following behavior-changing defect reports were applied retroactively to previously published C++ standards.
DR | Applied to | Behavior as published | Correct behavior |
---|---|---|---|
LWG 3656 | C++20 | the return type ofbit_width is the same as the type of its function argument | made itint |
(C++20) | counts the number of consecutive0 bits, starting from the most significant bit (function template)[edit] |