Lines Matching refs:vxa0
70 const int16x8_t vxa0 = vmovl_s8(va0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane() local
77 vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c0), vget_low_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
78 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c0), vget_low_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
84 vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c1), vget_low_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
85 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c1), vget_low_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
91 vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c2), vget_low_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
92 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c2), vget_low_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
98 vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c3), vget_low_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
99 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c3), vget_low_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
106 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c4), vget_high_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
107 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c4), vget_high_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
113 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c5), vget_high_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
114 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c5), vget_high_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
120 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c6), vget_high_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
121 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c6), vget_high_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
127 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c7), vget_high_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
128 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c7), vget_high_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
136 const int16x8_t vxa0 = vmovl_s8(va0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane() local
143 vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c0), vget_low_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
144 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c0), vget_low_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
152 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c1), vget_low_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
153 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c1), vget_low_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
161 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c2), vget_low_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
162 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c2), vget_low_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
170 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c3), vget_low_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
171 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c3), vget_low_s16(vxa0), 3); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
179 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c4), vget_high_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
180 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c4), vget_high_s16(vxa0), 0); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
188 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c5), vget_high_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
189 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c5), vget_high_s16(vxa0), 1); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
197 … vacc0x0123 = vmlal_lane_s16(vacc0x0123, vget_low_s16(vxb01234567c6), vget_high_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()
198 … vacc0x4567 = vmlal_lane_s16(vacc0x4567, vget_high_s16(vxb01234567c6), vget_high_s16(vxa0), 2); in xnn_qs8_igemm_minmax_ukernel_2x8__neon_mlal_lane()