87       __m256 vacc01234567p0 = _mm256_load_ps(w);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()  local
94       vacc01234567p0 = _mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
100       vacc01234567p0 = _mm256_fmadd_ps(vi1x01234567, vk1x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
106       vacc01234567p0 = _mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
112       vacc01234567p0 = _mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
118       vacc01234567p0 = _mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
124       vacc01234567p0 = _mm256_fmadd_ps(vi5x01234567, vk5x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
130       vacc01234567p0 = _mm256_fmadd_ps(vi6x01234567, vk6x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
136       vacc01234567p0 = _mm256_fmadd_ps(vi7x01234567, vk7x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
142       vacc01234567p0 = _mm256_fmadd_ps(vi8x01234567, vk8x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
147       __m256 vacc01234567 = _mm256_max_ps(vacc01234567p0, vmin);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
158       __m256 vacc01234567p0 = _mm256_load_ps(w);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()  local
162       vacc01234567p0 = _mm256_fmadd_ps(vi0x01234567, vk0x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
166       vacc01234567p0 = _mm256_fmadd_ps(vi1x01234567, vk1x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
170       vacc01234567p0 = _mm256_fmadd_ps(vi2x01234567, vk2x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
174       vacc01234567p0 = _mm256_fmadd_ps(vi3x01234567, vk3x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
178       vacc01234567p0 = _mm256_fmadd_ps(vi4x01234567, vk4x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
182       vacc01234567p0 = _mm256_fmadd_ps(vi5x01234567, vk5x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
186       vacc01234567p0 = _mm256_fmadd_ps(vi6x01234567, vk6x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
190       vacc01234567p0 = _mm256_fmadd_ps(vi7x01234567, vk7x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
194       vacc01234567p0 = _mm256_fmadd_ps(vi8x01234567, vk8x01234567, vacc01234567p0);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()
197       __m256 vacc01234567 = _mm256_max_ps(vacc01234567p0, vmin);  in xnn_f32_dwconv_minmax_ukernel_up8x9__fma3()