NAME

std::bit_ceil - std::bit_ceil

Synopsis

Defined in header <bit>
template< class T > (since C++20)
constexpr T bit_ceil( T x );

Calculates the smallest integral power of two that is not smaller than x.

If that value is not representable in T, the behavior is undefined. Call to this
function is permitted in constant evaluation only if the undefined behavior does not
occur.

This overload participates in overload resolution only if T is an unsigned integer
type (that is, unsigned char, unsigned short, unsigned int, unsigned long, unsigned
long long, or an extended unsigned integer type).

Parameters

x - value of unsigned integer type

Return value

The smallest integral power of two that is not smaller than x.

Exceptions

Throws nothing.

Notes

Feature-test macro: __cpp_lib_int_pow2

Possible implementation

See possible implementations in libstdc++ (gcc) and libc++ (clang)

template <std::unsigned_integral T>
requires !std::same_as<T, bool> && !std::same_as<T, char> &&
!std::same_as<T, char8_t> && !std::same_as<T, char16_t> &&
!std::same_as<T, char32_t> && !std::same_as<T, wchar_t>
constexpr T bit_ceil(T x) noexcept
{
if (x <= 1u)
return T(1);
if constexpr (std::same_as<T, decltype(+x)>)
return T(1) << std::bit_width(T(x - 1));
else { // for types subject to integral promotion
constexpr int offset_for_ub =
std::numeric_limits<unsigned>::digits - std::numeric_limits<T>::digits;
return T(1u << (std::bit_width(T(x - 1)) + offset_for_ub) >> offset_for_ub);
}
}

Example

// Run this code

#include <bit>
#include <bitset>
#include <iostream>

auto main() -> signed int // :()
{
using bin = std::bitset<8>;

for (unsigned x{0}; x != 10; ++x)
{
unsigned const z = std::bit_ceil(x); // `ceil2` before P1956R1

std::cout << "bit_ceil( " << bin(x) << " ) = " << bin(z) << '\n';
}
}

Output:

bit_ceil( 00000000 ) = 00000001
bit_ceil( 00000001 ) = 00000001
bit_ceil( 00000010 ) = 00000010
bit_ceil( 00000011 ) = 00000100
bit_ceil( 00000100 ) = 00000100
bit_ceil( 00000101 ) = 00001000
bit_ceil( 00000110 ) = 00001000
bit_ceil( 00000111 ) = 00001000
bit_ceil( 00001000 ) = 00001000
bit_ceil( 00001001 ) = 00010000