__m64 EdgeDetector::populateRun(unsigned int x, unsigned int y, unsigned int channelOffset, RunOrientation orientation)
{
    unsigned int index = (y * 2560) + (x << 2) + channelOffset;

    if(orientation == HORIZONTAL)
    {
        return _mm_setr_pi8(yuvArray[index],
                            yuvArray[index + 4],
                            yuvArray[index + 8],
                            yuvArray[index + 12],
                            yuvArray[index + 16],
                            yuvArray[index + 20],
                            yuvArray[index + 24],
                            yuvArray[index + 28]);
    }
    else if(orientation == VERTICAL)
    {
        return _mm_setr_pi8(yuvArray[index],
                            yuvArray[index + 2560],
                            yuvArray[index + 5120],
                            yuvArray[index + 7680],
                            yuvArray[index + 10240],
                            yuvArray[index + 12800],
                            yuvArray[index + 15360],
                            yuvArray[index + 17920]);
    }

    return _m_from_int(0);
}

long long EdgeDetector::differenceCombined(unsigned int startX, unsigned int startY, RunOrientation orientation, unsigned int channelOffset)
{
    __m64 pixelValues = populateRun(startX, startY, channelOffset, orientation);

    __m64 shiftedValuesRight = _mm_srli_si64(pixelValues, 8);
    __m64 shiftedValuesLeft = _mm_slli_si64(pixelValues, 8);

    __m64 resultRight = _mm_subs_pu8(pixelValues, shiftedValuesRight);
    __m64 resultLeft = _mm_subs_pu8(pixelValues, shiftedValuesLeft);

    __m64 result = _mm_adds_pu8(resultRight, resultLeft);

    long long retVal = _m_to_int64(result);

    _mm_empty();

    return retVal;
}