如何最好地模拟_mm_slli_si128（128位位移）的逻辑含义，而不是_mm_bslli_si128

1 这不是疏忽。该指令确实按字节移动，即 8 位的倍数。

2 无关紧要，_mm_slli_si128并且_mm_bslli_si128是等价的，两者都编译为pslldqSSE2 指令。

至于仿真，假设您有 C++/17，我会这样做。如果您正在编写 C++/14，请替换if constexpr为 normal if，同时在static_assert.

template<int i>
inline __m128i shiftLeftBits( __m128i vec )
{
    static_assert( i >= 0 && i < 128 );
    // Handle couple trivial cases
    if constexpr( 0 == i )
        return vec;
    if constexpr( 0 == ( i % 8 ) )
        return _mm_slli_si128( vec, i / 8 );

    if constexpr( i > 64 )
    {
        // Shifting by more than 8 bytes, the lowest half will be all zeros
        vec = _mm_slli_si128( vec, 8 );
        return _mm_slli_epi64( vec, i - 64 );
    }
    else
    {
        // Shifting by less than 8 bytes.
        // Need to propagate a few bits across 64-bit lanes.
        __m128i low = _mm_slli_si128( vec, 8 );
        __m128i high = _mm_slli_epi64( vec, i );
        low = _mm_srli_epi64( low, 64 - i );
        return _mm_or_si128( low, high );
    }
}

• I'd recommend `_mm_bslli_si128` - the newer name more clearly implies that it's a byte shift. Note that you don't really need `if constexpr`. That does maybe help your compiler make more efficient code in debug mode, if your compiler doesn't remove `if(false)` blocks in that case (*cough* MSVC), but that's all. With `i` as a template parameter, it's definitely a compile-time constant and even MSVC does dead-code removal with optimization enabled.