299   __m256 vaccv = _mm256_mul_ps(vaccv0, vaccs0);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()  local
300   vaccv = _mm256_fmadd_ps(vaccv1, vaccs1, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
301   vaccv = _mm256_fmadd_ps(vaccv2, vaccs2, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
302   vaccv = _mm256_fmadd_ps(vaccv3, vaccs3, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
303   vaccv = _mm256_fmadd_ps(vaccv4, vaccs4, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
304   vaccv = _mm256_fmadd_ps(vaccv5, vaccs5, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
339     vaccv = _mm256_mul_ps(vaccv, vaccs);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
340     vaccv = _mm256_fmadd_ps(vp, vs, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
383     vaccv = _mm256_mul_ps(vaccv, vaccs);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
384     vaccv = _mm256_fmadd_ps(vp, vs, vaccv);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
396   vaccv = _mm256_mul_ps(vaccv, vaccs);  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()
397   __m128 vaccv_sum = _mm_add_ps(_mm256_castps256_ps128(vaccv), _mm256_extractf128_ps(vaccv, 1));  in xnn_f32_raddextexp_ukernel__avx2_p5_x96_acc6()