50       __m128 va0 = _mm_loadu_ps(a0);  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()  local
57       vacc0x0123 = _mm_add_ps(vacc0x0123, _mm_mul_ps(va0, vb0123c0));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
58       vacc0x4567 = _mm_add_ps(vacc0x4567, _mm_mul_ps(va0, vb4567c0));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
60       va0 = _mm_shuffle_ps(va0, va0, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
65       vacc0x0123 = _mm_add_ps(vacc0x0123, _mm_mul_ps(va0, vb0123c1));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
66       vacc0x4567 = _mm_add_ps(vacc0x4567, _mm_mul_ps(va0, vb4567c1));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
68       va0 = _mm_shuffle_ps(va0, va0, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
73       vacc0x0123 = _mm_add_ps(vacc0x0123, _mm_mul_ps(va0, vb0123c2));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
74       vacc0x4567 = _mm_add_ps(vacc0x4567, _mm_mul_ps(va0, vb4567c2));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
76       va0 = _mm_shuffle_ps(va0, va0, _MM_SHUFFLE(0, 3, 2, 1));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
81       vacc0x0123 = _mm_add_ps(vacc0x0123, _mm_mul_ps(va0, vb0123c3));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
82       vacc0x4567 = _mm_add_ps(vacc0x4567, _mm_mul_ps(va0, vb4567c3));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
90         const __m128 va0 = _mm_load1_ps(a0);  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()  local
97         vacc0x0123 = _mm_add_ps(vacc0x0123, _mm_mul_ps(va0, vb0123));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()
98         vacc0x4567 = _mm_add_ps(vacc0x4567, _mm_mul_ps(va0, vb4567));  in xnn_f32_gemminc_minmax_ukernel_1x8s4__sse()