int32_t WelsSampleSatd4x4_c (uint8_t* pSample1, int32_t iStride1, uint8_t* pSample2, int32_t iStride2) {
int32_t iSatdSum = 0;
int32_t pSampleMix[4][4] = { 0 };
int32_t iSample0, iSample1, iSample2, iSample3;
int32_t i = 0;
uint8_t* pSrc1 = pSample1;
uint8_t* pSrc2 = pSample2;
//step 1: get the difference
for (i = 0; i < 4; i++) {
pSampleMix[i][0] = pSrc1[0] - pSrc2[0];
pSampleMix[i][1] = pSrc1[1] - pSrc2[1];
pSampleMix[i][2] = pSrc1[2] - pSrc2[2];
pSampleMix[i][3] = pSrc1[3] - pSrc2[3];
pSrc1 += iStride1;
pSrc2 += iStride2;
}
//step 2: horizontal transform
for (i = 0; i < 4; i++) {
iSample0 = pSampleMix[i][0] + pSampleMix[i][2];
iSample1 = pSampleMix[i][1] + pSampleMix[i][3];
iSample2 = pSampleMix[i][0] - pSampleMix[i][2];
iSample3 = pSampleMix[i][1] - pSampleMix[i][3];
pSampleMix[i][0] = iSample0 + iSample1;
pSampleMix[i][1] = iSample2 + iSample3;
pSampleMix[i][2] = iSample2 - iSample3;
pSampleMix[i][3] = iSample0 - iSample1;
}
//step 3: vertical transform and get the sum of SATD
for (i = 0; i < 4; i++) {
iSample0 = pSampleMix[0][i] + pSampleMix[2][i];
iSample1 = pSampleMix[1][i] + pSampleMix[3][i];
iSample2 = pSampleMix[0][i] - pSampleMix[2][i];
iSample3 = pSampleMix[1][i] - pSampleMix[3][i];
pSampleMix[0][i] = iSample0 + iSample1;
pSampleMix[1][i] = iSample2 + iSample3;
pSampleMix[2][i] = iSample2 - iSample3;
pSampleMix[3][i] = iSample0 - iSample1;
iSatdSum += (WELS_ABS (pSampleMix[0][i]) + WELS_ABS (pSampleMix[1][i]) + WELS_ABS (pSampleMix[2][i]) + WELS_ABS (
pSampleMix[3][i]));
}
return ((iSatdSum + 1) >> 1);
}
int32_t WelsSampleSad4x4_c (uint8_t* pSample1, int32_t iStride1, uint8_t* pSample2, int32_t iStride2) {
int32_t iSadSum = 0;
int32_t i = 0;
uint8_t* pSrc1 = pSample1;
uint8_t* pSrc2 = pSample2;
for (i = 0; i < 4; i++) {
iSadSum += WELS_ABS ((pSrc1[0] - pSrc2[0]));
iSadSum += WELS_ABS ((pSrc1[1] - pSrc2[1]));
iSadSum += WELS_ABS ((pSrc1[2] - pSrc2[2]));
iSadSum += WELS_ABS ((pSrc1[3] - pSrc2[3]));
pSrc1 += iStride1;
pSrc2 += iStride2;
}
return iSadSum;
}
WELSVP_NAMESPACE_BEGIN
int32_t WelsSampleSad8x8_c (uint8_t* pSrcY, int32_t iSrcStrideY, uint8_t* pRefY, int32_t iRefStrideY) {
int32_t iSadSum = 0;
uint8_t* pSrcA = pSrcY;
uint8_t* pSrcB = pRefY;
for (int32_t i = 0; i < 8; i++) {
iSadSum += WELS_ABS ((pSrcA[0] - pSrcB[0]));
iSadSum += WELS_ABS ((pSrcA[1] - pSrcB[1]));
iSadSum += WELS_ABS ((pSrcA[2] - pSrcB[2]));
iSadSum += WELS_ABS ((pSrcA[3] - pSrcB[3]));
iSadSum += WELS_ABS ((pSrcA[4] - pSrcB[4]));
iSadSum += WELS_ABS ((pSrcA[5] - pSrcB[5]));
iSadSum += WELS_ABS ((pSrcA[6] - pSrcB[6]));
iSadSum += WELS_ABS ((pSrcA[7] - pSrcB[7]));
pSrcA += iSrcStrideY;
pSrcB += iRefStrideY;
}
return iSadSum;
}
