45 …float16x8_t vacc0x01234567 = vld1q_f16(w); w = (const void*) ((uintptr_t) w + sizeof(float16x8_t));  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()  local
54         vacc0x01234567 = vfmaq_lane_f16(vacc0x01234567, vb01234567c0, va0, 0);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
58         vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c0, vb01234567c0);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
63         vacc0x01234567 = vfmaq_lane_f16(vacc0x01234567, vb01234567c1, va0, 1);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
67         vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c1, vb01234567c1);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
72         vacc0x01234567 = vfmaq_lane_f16(vacc0x01234567, vb01234567c2, va0, 2);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
76         vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c2, vb01234567c2);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
81         vacc0x01234567 = vfmaq_lane_f16(vacc0x01234567, vb01234567c3, va0, 3);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
85         vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
96         vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0, vb01234567);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
103     vacc0x01234567 = vmulq_f16(vacc0x01234567, vscale);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
106     vacc0x01234567 = vminq_f16(vacc0x01234567, vmax);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
109     vacc0x01234567 = vmaxq_f16(vacc0x01234567, vmin);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
112       vst1q_f16(c0, vacc0x01234567);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
119       float16x4_t vacc0x0123 = vget_low_f16(vacc0x01234567);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
123         vacc0x0123 = vget_high_f16(vacc0x01234567);  in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()