Home
last modified time | relevance | path

Searched refs:vb0123x89AB (Results 1 – 8 of 8) sorted by relevance

/external/XNNPACK/src/qs8-igemm/gen/
D8x16c4-minmax-neondot.c161 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot() local
171 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
175 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
179 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
183 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
187 vacc4x89AB = vdotq_lane_s32(vacc4x89AB, vb0123x89AB, va4x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
191 vacc5x89AB = vdotq_lane_s32(vacc5x89AB, vb0123x89AB, va5x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
195 vacc6x89AB = vdotq_lane_s32(vacc6x89AB, vb0123x89AB, va6x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
199 vacc7x89AB = vdotq_lane_s32(vacc7x89AB, vb0123x89AB, va7x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot()
251 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_8x16c4__neondot() local
[all …]
D6x16c4-minmax-neondot.c135 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot() local
145 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
149 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
153 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
157 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
161 vacc4x89AB = vdotq_lane_s32(vacc4x89AB, vb0123x89AB, va4x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
165 vacc5x89AB = vdotq_lane_s32(vacc5x89AB, vb0123x89AB, va5x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
207 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot() local
213 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
217 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_6x16c4__neondot()
[all …]
D4x16c4-minmax-neondot.c109 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot() local
119 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
123 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
127 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
131 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
163 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot() local
169 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
173 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
177 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
181 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_4x16c4__neondot()
D1x16c4-minmax-neondot.c70 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_1x16c4__neondot() local
80 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_1x16c4__neondot()
97 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_igemm_minmax_ukernel_1x16c4__neondot() local
103 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_igemm_minmax_ukernel_1x16c4__neondot()
/external/XNNPACK/src/qs8-gemm/gen/
D8x16c4-minmax-neondot.c138 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot() local
148 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
152 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
156 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
160 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
164 vacc4x89AB = vdotq_lane_s32(vacc4x89AB, vb0123x89AB, va4x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
168 vacc5x89AB = vdotq_lane_s32(vacc5x89AB, vb0123x89AB, va5x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
172 vacc6x89AB = vdotq_lane_s32(vacc6x89AB, vb0123x89AB, va6x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
176 vacc7x89AB = vdotq_lane_s32(vacc7x89AB, vb0123x89AB, va7x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot()
228 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_8x16c4__neondot() local
[all …]
D6x16c4-minmax-neondot.c116 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot() local
126 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
130 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
134 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
138 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
142 vacc4x89AB = vdotq_lane_s32(vacc4x89AB, vb0123x89AB, va4x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
146 vacc5x89AB = vdotq_lane_s32(vacc5x89AB, vb0123x89AB, va5x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
188 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot() local
194 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
198 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_6x16c4__neondot()
[all …]
D4x16c4-minmax-neondot.c94 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot() local
104 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
108 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
112 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
116 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
148 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot() local
154 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
158 vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
162 vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
166 vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_4x16c4__neondot()
D1x16c4-minmax-neondot.c61 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_1x16c4__neondot() local
71 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_1x16c4__neondot()
88 const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const void*) ((const int8_t*) w + 16); in xnn_qs8_gemm_minmax_ukernel_1x16c4__neondot() local
94 vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); in xnn_qs8_gemm_minmax_ukernel_1x16c4__neondot()