/external/XNNPACK/src/f32-sigmoid/gen/ |
D | neonfma-rr1-lut2048-p1-div-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4() 152 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4() local 154 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4() 156 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x4()
|
D | neonfma-rr1-lut64-p2-div-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4() 154 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4() local 156 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4() 158 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x4()
|
D | neonfma-rr1-lut2048-p1-nr1recps1fma-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4() 161 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4() local 163 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4() 165 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x4()
|
D | neonfma-rr1-lut2048-p1-nr2recps-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4() 161 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4() local 163 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4() 165 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x4()
|
D | neonfma-rr1-lut64-p2-nr1recps1fma-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4() 163 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4() local 165 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4() 167 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x4()
|
D | neonfma-rr1-lut2048-p1-nr2fma-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4() 161 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4() local 163 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4() 165 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2fma_x4()
|
D | neon-rr2-lut2048-p1-nr2recps-x4.c | 77 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4() local 79 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4() 81 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4() 165 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4() local 167 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4() 169 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neon_rr2_lut2048_p1_nr2recps_x4()
|
D | neonfma-rr1-lut64-p2-nr2recps-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4() 163 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4() local 165 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4() 167 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2recps_x4()
|
D | neon-rr2-lut64-p2-nr2recps-x4.c | 77 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4() local 79 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4() 81 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4() 167 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4() local 169 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4() 171 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neon_rr2_lut64_p2_nr2recps_x4()
|
D | neonfma-rr1-lut64-p2-nr2fma-x4.c | 75 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4() local 77 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4() 79 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4() 163 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4() local 165 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4() 167 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr2fma_x4()
|
D | neonfma-rr1-lut64-p2-div-x8.c | 183 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8() local 185 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8() 187 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8() 262 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8() local 264 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8() 266 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_div_x8()
|
D | neonfma-rr1-lut2048-p1-div-x8.c | 179 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8() local 181 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8() 183 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8() 256 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8() local 258 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8() 260 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_div_x8()
|
D | neonfma-rr1-lut2048-p1-nr1recps1fma-x8.c | 191 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8() local 193 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8() 195 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8() 277 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8() local 279 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8() 281 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr1recps1fma_x8()
|
D | neonfma-rr1-lut2048-p1-nr2recps-x8.c | 191 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8() local 193 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8() 195 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8() 277 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8() local 279 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_2048[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8() 281 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_2048[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut2048_p1_nr2recps_x8()
|
D | neonfma-rr1-lut64-p2-nr1recps1fma-x8.c | 195 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8() local 197 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8() 199 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8() 283 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8() local 285 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8() 287 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_sigmoid_ukernel__neonfma_rr1_lut64_p2_nr1recps1fma_x8()
|
/external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
D | neon-lut64-p2-x4.c | 74 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4() local 76 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4() 78 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4() 150 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4() local 152 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4() 154 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x4()
|
D | neonfma-lut64-p2-x4.c | 73 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4() local 75 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4() 77 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4() 149 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4() local 151 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4() 153 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x4()
|
D | neon-lut64-p2-x8-acc2.c | 175 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2() local 177 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2() 179 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2() 251 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2() local 253 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2() 255 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8_acc2()
|
D | neonfma-lut64-p2-x8-acc2.c | 174 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2() local 176 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2() 178 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2() 250 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2() local 252 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2() 254 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8_acc2()
|
D | neon-lut64-p2-x8.c | 172 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8() local 174 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8() 176 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8() 248 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8() local 250 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8() 252 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neon_lut64_p2_x8()
|
D | neonfma-lut64-p2-x8.c | 171 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8() local 173 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8() 175 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8() 247 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8() local 249 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8() 251 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_f32_raddstoreexpminusmax_ukernel__neonfma_lut64_p2_x8()
|
/external/XNNPACK/src/math/ |
D | sigmoid-neonfma-rr1-lut64-p2-div.c | 72 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_div() local 74 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_div() 76 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_div()
|
D | sigmoid-neonfma-rr2-lut64-p2-div.c | 73 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_math_f32_sigmoid__neonfma_rr2_lut64_p2_div() local 75 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_math_f32_sigmoid__neonfma_rr2_lut64_p2_div() 77 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_math_f32_sigmoid__neonfma_rr2_lut64_p2_div()
|
D | sigmoid-neon-rr1-lut64-p2-nr2recps.c | 72 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_math_f32_sigmoid__neon_rr1_lut64_p2_nr2recps() local 74 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_math_f32_sigmoid__neon_rr1_lut64_p2_nr2recps() 76 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_math_f32_sigmoid__neon_rr1_lut64_p2_nr2recps()
|
D | sigmoid-neonfma-rr1-lut64-p2-nr2recps.c | 72 const uint64_t vidx_lo = vgetq_lane_u64(vidx, 0); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_nr2recps() local 74 float32x2_t vl_lo = vld1_dup_f32(&xnn_table_exp2_k_over_64[(uint32_t) vidx_lo]); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_nr2recps() 76 vl_lo = vld1_lane_f32(&xnn_table_exp2_k_over_64[(uint32_t) (vidx_lo >> 32)], vl_lo, 1); in xnn_math_f32_sigmoid__neonfma_rr1_lut64_p2_nr2recps()
|