A clz() implementation wrapping __builtin_clz() - see blog post http://foonathan.github.io/blog/2016/02/11/implementation-challenge-2.html

clz.cpp

#include <climits>
#include <cstdint>
#include <type_traits>

struct clzll_tag {};
struct clzl_tag : clzll_tag {};
struct clz_tag : clzl_tag {};

template <typename T, typename = typename std::enable_if<sizeof(T) <= sizeof(unsigned int)>::type>
unsigned clz_impl(clz_tag, T x)
{
	return __builtin_clz(x) - (sizeof(unsigned int) * CHAR_BIT - sizeof(T) * CHAR_BIT);
}

template <typename T, typename = typename std::enable_if<sizeof(T) <= sizeof(unsigned long)>::type>
unsigned clz_impl(clzl_tag, T x)
{
	return __builtin_clzl(x) - (sizeof(unsigned long) * CHAR_BIT - sizeof(T) * CHAR_BIT);
}

template <typename T, typename = typename std::enable_if<sizeof(T) <= sizeof(unsigned long long)>::type>
unsigned clz_impl(clzll_tag, T x)
{
	return __builtin_clzll(x) - (sizeof(unsigned long long) * CHAR_BIT - sizeof(T) * CHAR_BIT);
}

unsigned clz(std::uint8_t x)
{
    return clz_impl(clz_tag{}, x);
}

unsigned clz(std::uint16_t x)
{
    return clz_impl(clz_tag{}, x);
}

unsigned clz(std::uint32_t x)
{
    return clz_impl(clz_tag{}, x);
}

unsigned clz(std::uint64_t x)
{
    return clz_impl(clz_tag{}, x);
}