/**
* Calculate scalar product of v1 and v2,
* and v1[i] += v3[i] * mul
- * @param len length of vectors, should be multiple of 16
+ * @param len length of vectors, should be multiple of 8
*/
int32_t (*scalarproduct_and_madd_int16)(int16_t *v1 /* align 16 */,
const int16_t *v2,