On Release x64 on VC++ 2015 update 3 this doesn't work, on Debug x64/x86 this works just fine... >:|

ssebug.cpp

#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "emmintrin.h"

struct Color4
{
    uint8_t red;
    uint8_t green;
    uint8_t blue;
    uint8_t alpha;

    Color4(uint8_t red, uint8_t green, uint8_t blue, uint8_t alpha = 255)
        : red(red), green(green), blue(blue), alpha(alpha) {}

    explicit Color4(uint32_t rgba)
    {
        red = (uint8_t)(rgba & 0xFF);
        green = (uint8_t)((rgba >> 8)&0xFF);
        blue = (uint8_t)((rgba >> 16) & 0xFF);
        alpha = (uint8_t)((rgba >> 24) & 0xFF);
    }
};

Color4 PackUnpack(Color4 col)
{
    uint32_t tmp;

    memcpy(&tmp, &col, sizeof(tmp));
    
    __m128 aFloat = _mm_cvtepi32_ps(
        _mm_unpacklo_epi16(
            _mm_unpacklo_epi8(
                _mm_set1_epi32(tmp),
                _mm_setzero_si128()
            ),
            _mm_setzero_si128()
        )
    );

    __m128i ret = _mm_packus_epi16(
        _mm_packs_epi32(
            _mm_cvtps_epi32(aFloat),
            _mm_setzero_si128()
        ),
        _mm_setzero_si128()
    );

    return Color4((uint32_t)_mm_cvtsi128_si32(ret));
}

int main()
{
#ifdef _DEBUG
    printf("DEBUG\n");
#else
    printf("RELEASE\n");
#endif

    Color4 c = PackUnpack(Color4(32, 64, 128, 255));

    // Debug x64 or Debug/Release x86: Prints "32 64 128 255"
    // Release x64: Prints "255 0 0 0"
    printf("%d %d %d %d\n",  c.red, c.green, c.blue, c.alpha);

    return 0;
}