106       int16x8_t vprod0x0 = vmull_s8(vget_low_s8(vb0), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
107       int16x8_t vprod1x0 = vmull_s8(vget_low_s8(vb0), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
112       int16x8_t vprod0x1 = vmull_s8(vget_low_s8(vb1), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
113       int16x8_t vprod1x1 = vmull_s8(vget_low_s8(vb1), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
118       int16x8_t vprod0x2 = vmull_s8(vget_low_s8(vb2), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
119       int16x8_t vprod1x2 = vmull_s8(vget_low_s8(vb2), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
124       int16x8_t vprod0x3 = vmull_s8(vget_low_s8(vb3), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
125       int16x8_t vprod1x3 = vmull_s8(vget_low_s8(vb3), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
130       int16x8_t vprod0x4 = vmull_s8(vget_low_s8(vb4), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
131       int16x8_t vprod1x4 = vmull_s8(vget_low_s8(vb4), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
136       int16x8_t vprod0x5 = vmull_s8(vget_low_s8(vb5), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
137       int16x8_t vprod1x5 = vmull_s8(vget_low_s8(vb5), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
142       int16x8_t vprod0x6 = vmull_s8(vget_low_s8(vb6), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
143       int16x8_t vprod1x6 = vmull_s8(vget_low_s8(vb6), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
148       int16x8_t vprod0x7 = vmull_s8(vget_low_s8(vb7), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
149       int16x8_t vprod1x7 = vmull_s8(vget_low_s8(vb7), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
154       int16x8_t vprod0x8 = vmull_s8(vget_low_s8(vb8), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
155       int16x8_t vprod1x8 = vmull_s8(vget_low_s8(vb8), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
160       int16x8_t vprod0x9 = vmull_s8(vget_low_s8(vb9), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
161       int16x8_t vprod1x9 = vmull_s8(vget_low_s8(vb9), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
166       int16x8_t vprod0x10 = vmull_s8(vget_low_s8(vb10), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
167       int16x8_t vprod1x10 = vmull_s8(vget_low_s8(vb10), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
172       int16x8_t vprod0x11 = vmull_s8(vget_low_s8(vb11), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
173       int16x8_t vprod1x11 = vmull_s8(vget_low_s8(vb11), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
178       int16x8_t vprod0x12 = vmull_s8(vget_low_s8(vb12), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
179       int16x8_t vprod1x12 = vmull_s8(vget_low_s8(vb12), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
184       int16x8_t vprod0x13 = vmull_s8(vget_low_s8(vb13), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
185       int16x8_t vprod1x13 = vmull_s8(vget_low_s8(vb13), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
190       int16x8_t vprod0x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
191       int16x8_t vprod1x14 = vmull_s8(vget_low_s8(vb14), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
196       int16x8_t vprod0x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va0));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
197       int16x8_t vprod1x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va1));  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
358 …int8x16_t vout0x01234567_1x01234567 = vcombine_s8(vget_low_s8(vout0x0123456789ABCDEF), vget_low_s8…  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()
360         vst1_s8(c0, vget_low_s8(vout0x01234567_1x01234567)); c0 += 8;  in xnn_qs8_gemm_minmax_ukernel_2x16c16__neon_mlal_padal()