template <bool align> SIMD_INLINE void BgrToYuv420p(const uint8_t * bgr0, size_t bgrStride, uint8_t * y0, size_t yStride, uint8_t * u, uint8_t * v)
{
const uint8_t * bgr1 = bgr0 + bgrStride;
uint8_t * y1 = y0 + yStride;
__m256i blue[2][2], green[2][2], red[2][2];
LoadBgr<align>((__m256i*)bgr0 + 0, blue[0][0], green[0][0], red[0][0]);
Store<align>((__m256i*)y0 + 0, BgrToY8(blue[0][0], green[0][0], red[0][0]));
LoadBgr<align>((__m256i*)bgr0 + 3, blue[0][1], green[0][1], red[0][1]);
Store<align>((__m256i*)y0 + 1, BgrToY8(blue[0][1], green[0][1], red[0][1]));
LoadBgr<align>((__m256i*)bgr1 + 0, blue[1][0], green[1][0], red[1][0]);
Store<align>((__m256i*)y1 + 0, BgrToY8(blue[1][0], green[1][0], red[1][0]));
LoadBgr<align>((__m256i*)bgr1 + 3, blue[1][1], green[1][1], red[1][1]);
Store<align>((__m256i*)y1 + 1, BgrToY8(blue[1][1], green[1][1], red[1][1]));
blue[0][0] = Average16(blue[0][0], blue[1][0]);
blue[0][1] = Average16(blue[0][1], blue[1][1]);
green[0][0] = Average16(green[0][0], green[1][0]);
green[0][1] = Average16(green[0][1], green[1][1]);
red[0][0] = Average16(red[0][0], red[1][0]);
red[0][1] = Average16(red[0][1], red[1][1]);
Store<align>((__m256i*)u, PackU16ToU8(BgrToU16(blue[0][0], green[0][0], red[0][0]), BgrToU16(blue[0][1], green[0][1], red[0][1])));
Store<align>((__m256i*)v, PackU16ToU8(BgrToV16(blue[0][0], green[0][0], red[0][0]), BgrToV16(blue[0][1], green[0][1], red[0][1])));
}
template <bool align> SIMD_INLINE void BgrToYuv444p(const uint8_t * bgr, uint8_t * y, uint8_t * u, uint8_t * v)
{
__m256i blue, green, red;
LoadBgr<align>((__m256i*)bgr, blue, green, red);
Store<align>((__m256i*)y, BgrToY8(blue, green, red));
Store<align>((__m256i*)u, BgrToU8(blue, green, red));
Store<align>((__m256i*)v, BgrToV8(blue, green, red));
}
template <bool align> SIMD_INLINE void BgrToYuv422p(const uint8_t * bgr, uint8_t * y, uint8_t * u, uint8_t * v)
{
__m128i blue[2], green[2], red[2];
LoadBgr<align>((__m128i*)bgr + 0, blue[0], green[0], red[0]);
Store<align>((__m128i*)y + 0, BgrToY8(blue[0], green[0], red[0]));
LoadBgr<align>((__m128i*)bgr + 3, blue[1], green[1], red[1]);
Store<align>((__m128i*)y + 1, BgrToY8(blue[1], green[1], red[1]));
Average16(blue[0]);
Average16(blue[1]);
Average16(green[0]);
Average16(green[1]);
Average16(red[0]);
Average16(red[1]);
Store<align>((__m128i*)u, _mm_packus_epi16(BgrToU16(blue[0], green[0], red[0]), BgrToU16(blue[1], green[1], red[1])));
Store<align>((__m128i*)v, _mm_packus_epi16(BgrToV16(blue[0], green[0], red[0]), BgrToV16(blue[1], green[1], red[1])));
}
