246 unsigned int num_points)
250 unsigned int quarter_points = num_points / 4;
251 float32x4x2_t a_val, b_val, c_val;
252 float32x4x2_t tmp_real, tmp_imag;
253 unsigned int number = 0;
255 for (number = 0; number < quarter_points; ++number) {
256 a_val = vld2q_f32((
float*)a_ptr);
257 b_val = vld2q_f32((
float*)b_ptr);
258 b_val.val[1] = vnegq_f32(b_val.val[1]);
264 tmp_real.val[0] = vmulq_f32(a_val.val[0], b_val.val[0]);
266 tmp_real.val[1] = vmulq_f32(a_val.val[1], b_val.val[1]);
270 tmp_imag.val[0] = vmulq_f32(a_val.val[0], b_val.val[1]);
272 tmp_imag.val[1] = vmulq_f32(a_val.val[1], b_val.val[0]);
275 c_val.val[0] = vsubq_f32(tmp_real.val[0], tmp_real.val[1]);
276 c_val.val[1] = vaddq_f32(tmp_imag.val[0], tmp_imag.val[1]);
277 vst2q_f32((
float*)cVector, c_val);
284 for (number = quarter_points * 4; number < num_points; number++) {
285 *cVector++ = (*a_ptr++) * conj(*b_ptr++);