Replace add and mul with fmadd

code/matrix_multiplication.h

@@ -145,7 +145,7 @@ void matrixMultiplyAVX2b(const std::complex<float>* a,
   __m256 b_4 = _mm256_permutevar8x32_ps(b_m, b_4_ind);
 
   __m256 c_m = _mm256_mul_ps(a_1, b_1);
-  c_m = _mm256_add_ps(_mm256_mul_ps(a_2, b_2), c_m);
+  c_m = _mm256_fmadd_ps(a_2, b_2, c_m);
   c_m = _mm256_addsub_ps(c_m, _mm256_mul_ps(a_3, b_3));
   c_m = _mm256_addsub_ps(c_m, _mm256_mul_ps(a_4, b_4));
   _mm256_store_ps(c_ptr, c_m);