static void VFilter8iSSE2(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
__m128i mask;
__m128i t1, t2, p1, p0, q0, q1;
// Load p3, p2, p1, p0
LOADUV_H_EDGES4(u, v, stride, t2, t1, p1, p0);
MAX_DIFF1(t2, t1, p1, p0, mask);
u += 4 * stride;
v += 4 * stride;
// Load q0, q1, q2, q3
LOADUV_H_EDGES4(u, v, stride, q0, q1, t1, t2);
MAX_DIFF2(t2, t1, q1, q0, mask);
COMPLEX_FL_MASK(p1, p0, q0, q1, thresh, ithresh, mask);
DoFilter4(&p1, &p0, &q0, &q1, &mask, hev_thresh);
// Store
STOREUV(p1, u, v, -2 * stride);
STOREUV(p0, u, v, -1 * stride);
STOREUV(q0, u, v, 0 * stride);
STOREUV(q1, u, v, 1 * stride);
}
// 8-pixels wide variant, for chroma filtering
static void VFilter8(uint8_t* u, uint8_t* v, int stride,
int thresh, int ithresh, int hev_thresh) {
__m128i mask;
__m128i t1, p2, p1, p0, q0, q1, q2;
// Load p3, p2, p1, p0
LOADUV_H_EDGES4(u - 4 * stride, v - 4 * stride, stride, t1, p2, p1, p0);
MAX_DIFF1(t1, p2, p1, p0, mask);
// Load q0, q1, q2, q3
LOADUV_H_EDGES4(u, v, stride, q0, q1, q2, t1);
MAX_DIFF2(t1, q2, q1, q0, mask);
ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
// Store
STOREUV(p2, u, v, -3 * stride);
STOREUV(p1, u, v, -2 * stride);
STOREUV(p0, u, v, -1 * stride);
STOREUV(q0, u, v, 0 * stride);
STOREUV(q1, u, v, 1 * stride);
STOREUV(q2, u, v, 2 * stride);
}
