__builtin_ctz (ctzl, ctzll) and __builtin_clz (clzl, clzll) for Visual Studio

ctz_clz.cpp

// Note, bsf/bsr are used by default.
// Enable /arch:AVX2 compilation for better optimizations

#if defined(_MSC_VER) && !defined(__clang__)
#include <intrin.h>

static __forceinline int __builtin_ctz(unsigned x)
{
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
    return (int)_CountTrailingZeros(x);
#elif defined(__AVX2__) || defined(__BMI__)
    return (int)_tzcnt_u32(x);
#else
    unsigned long r;
    _BitScanForward(&r, x);
    return (int)r;
#endif
}

static __forceinline int __builtin_ctzll(unsigned long long x)
{
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
    return (int)_CountTrailingZeros64(x);
#elif defined(_WIN64)
#if defined(__AVX2__) || defined(__BMI__)
    return (int)_tzcnt_u64(x);
#else
    unsigned long r;
    _BitScanForward64(&r, x);
    return (int)r;
#endif
#else
    int l = __builtin_ctz((unsigned)x);
    int h = __builtin_ctz((unsigned)(x >> 32)) + 32;
    return !!((unsigned)x) ? l : h;
#endif
}

static __forceinline int __builtin_ctzl(unsigned long x)
{
    return sizeof(x) == 8 ? __builtin_ctzll(x) : __builtin_ctz((unsigned)x);
}

static __forceinline int __builtin_clz(unsigned x)
{
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
    return (int)_CountLeadingZeros(x);
#elif defined(__AVX2__) || defined(__LZCNT__)
    return (int)_lzcnt_u32(x);
#else
    unsigned long r;
    _BitScanReverse(&r, x);
    return (int)(r ^ 31);
#endif
}

static __forceinline int __builtin_clzll(unsigned long long x)
{
#if defined(_M_ARM) || defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC)
    return (int)_CountLeadingZeros64(x);
#elif defined(_WIN64)
#if defined(__AVX2__) || defined(__LZCNT__)
    return (int)_lzcnt_u64(x);
#else
    unsigned long r;
    _BitScanReverse64(&r, x);
    return (int)(r ^ 63);
#endif
#else
    int l = __builtin_clz((unsigned)x) + 32;
    int h = __builtin_clz((unsigned)(x >> 32));
    return !!((unsigned)(x >> 32)) ? h : l;
#endif
}

static __forceinline int __builtin_clzl(unsigned long x)
{
    return sizeof(x) == 8 ? __builtin_clzll(x) : __builtin_clz((unsigned)x);
}
#endif // defined(_MSC_VER) && !defined(__clang__)

Thanks very much, Pavel, that's perfect, you're a star! :-)

https://github.com/nickpelling/C_STD/blob/develop/THANKS.md

Thanks very much, Pavel, that's perfect, you're a star! :-)

Note, I updated it to:

• check for __AVX2__ builds to use lzcnt/tzcnt vs bsf/bsr
• add code to handle arm/arm64
• add guard for clang-cl

The code was tested to verify that it's bit exact with the code that clang/gcc emits for these builtins.