78       __m128 va2 = _mm_loadu_ps(a2);  in xnn_f32_gemminc_ukernel_4x8s4__sse()  local
89       vacc2x0123 = _mm_add_ps(vacc2x0123, _mm_mul_ps(va2, vb0123c0));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
93       vacc2x4567 = _mm_add_ps(vacc2x4567, _mm_mul_ps(va2, vb4567c0));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
98       va2 = _mm_shuffle_ps(va2, va2, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
106       vacc2x0123 = _mm_add_ps(vacc2x0123, _mm_mul_ps(va2, vb0123c1));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
110       vacc2x4567 = _mm_add_ps(vacc2x4567, _mm_mul_ps(va2, vb4567c1));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
115       va2 = _mm_shuffle_ps(va2, va2, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
123       vacc2x0123 = _mm_add_ps(vacc2x0123, _mm_mul_ps(va2, vb0123c2));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
127       vacc2x4567 = _mm_add_ps(vacc2x4567, _mm_mul_ps(va2, vb4567c2));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
132       va2 = _mm_shuffle_ps(va2, va2, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
140       vacc2x0123 = _mm_add_ps(vacc2x0123, _mm_mul_ps(va2, vb0123c3));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
144       vacc2x4567 = _mm_add_ps(vacc2x4567, _mm_mul_ps(va2, vb4567c3));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
157         const __m128 va2 = _mm_load1_ps(a2);  in xnn_f32_gemminc_ukernel_4x8s4__sse()  local
168         vacc2x0123 = _mm_add_ps(vacc2x0123, _mm_mul_ps(va2, vb0123));  in xnn_f32_gemminc_ukernel_4x8s4__sse()
172         vacc2x4567 = _mm_add_ps(vacc2x4567, _mm_mul_ps(va2, vb4567));  in xnn_f32_gemminc_ukernel_4x8s4__sse()