/cmsis-nn-latest/Source/NNSupportFunctions/ |
D | arm_nn_mat_mult_nt_t_s8_s32.c | 68 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_mat_mult_nt_t_s8_s32() local 82 int8x16_t v_rhs0 = vldrbq_s8(rhs_ptr); in arm_nn_mat_mult_nt_t_s8_s32() 102 rhs_ptr += rhs_rows; in arm_nn_mat_mult_nt_t_s8_s32() 109 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_mat_mult_nt_t_s8_s32() local 114 int8x16_t v_rhs0 = vldrbq_s8(rhs_ptr); in arm_nn_mat_mult_nt_t_s8_s32() 122 rhs_ptr += rhs_rows; in arm_nn_mat_mult_nt_t_s8_s32() 139 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_mat_mult_nt_t_s8_s32() local 152 int8x16_t v_rhs0 = vldrbq_z_s8(rhs_ptr, rmdr); in arm_nn_mat_mult_nt_t_s8_s32() 173 rhs_ptr += rhs_rows; in arm_nn_mat_mult_nt_t_s8_s32() 180 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_mat_mult_nt_t_s8_s32() local [all …]
|
D | arm_nn_vec_mat_mult_t_s4.c | 68 const int8_t *rhs_ptr = &packed_rhs[0]; in arm_nn_vec_mat_mult_t_s4() local 112 const int8x16_t ker_0 = vldrbq_s8(rhs_ptr); in arm_nn_vec_mat_mult_t_s4() 123 const int8x16_t ker_1 = vldrbq_s8(&rhs_ptr[rhs_offset]); in arm_nn_vec_mat_mult_t_s4() 134 rhs_ptr += 16; in arm_nn_vec_mat_mult_t_s4() 141 int8x16_t ker_0 = vldrbq_gather_offset_s8(rhs_ptr, gather_offset); in arm_nn_vec_mat_mult_t_s4() 148 int8x16_t ker_1 = vldrbq_gather_offset_s8(&rhs_ptr[rhs_offset], gather_offset); in arm_nn_vec_mat_mult_t_s4() 156 rhs_ptr += 8; in arm_nn_vec_mat_mult_t_s4() 163 int8x16_t ker_0 = vldrbq_gather_offset_z_s8(rhs_ptr, gather_offset, rmdr_mask); in arm_nn_vec_mat_mult_t_s4() 170 … int8x16_t ker_1 = vldrbq_gather_offset_z_s8(&rhs_ptr[rhs_offset], gather_offset, rmdr_mask); in arm_nn_vec_mat_mult_t_s4() 177 rhs_ptr += rhs_cols_offset >> 1; in arm_nn_vec_mat_mult_t_s4() [all …]
|
D | arm_nn_vec_mat_mul_result_acc_s8_s16.c | 57 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_vec_mat_mul_result_acc_s8_s16() local 67 const int8_t *rhs_0 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s8_s16() 68 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s8_s16() 69 const int8_t *rhs_1 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s8_s16() 70 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s8_s16() 71 const int8_t *rhs_2 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s8_s16() 72 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s8_s16() 73 const int8_t *rhs_3 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s8_s16() 74 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s8_s16() 127 const int8_t *rhs_0 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s8_s16() [all …]
|
D | arm_nn_transpose_conv_row_s8_s32.c | 72 const int8_t *rhs_ptr = rhs; in arm_nn_transpose_conv_row_s8_s32() local 76 rhs_ptr += skip_pre_rows; in arm_nn_transpose_conv_row_s8_s32() 116 int32_t rhs0 = arm_nn_read_s8x4(rhs_ptr); in arm_nn_transpose_conv_row_s8_s32() 119 rhs_ptr += 4; in arm_nn_transpose_conv_row_s8_s32() 143 const int32_t rhs_val0 = *rhs_ptr++; in arm_nn_transpose_conv_row_s8_s32() 181 int8x16_t v_rhs0 = vldrbq_z_s8(rhs_ptr, p0); in arm_nn_transpose_conv_row_s8_s32() 182 rhs_ptr += 16; in arm_nn_transpose_conv_row_s8_s32() 192 rhs_ptr += channel_count; in arm_nn_transpose_conv_row_s8_s32() 217 const int32_t rhs_val0 = *rhs_ptr++; in arm_nn_transpose_conv_row_s8_s32() 218 const int32_t rhs_val1 = *rhs_ptr++; in arm_nn_transpose_conv_row_s8_s32() [all …]
|
D | arm_nn_mat_mult_nt_t_s8.c | 256 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_mat_mult_nt_t_s8() 269 val1 = arm_nn_read_s8x4_ia((const int8_t **)&rhs_ptr); in arm_nn_mat_mult_nt_t_s8() 273 val4 = arm_nn_read_s8x4((const int8_t *)&rhs_ptr[rhs_off0]); in arm_nn_mat_mult_nt_t_s8() 292 val1 = arm_nn_read_s8x4_ia((const int8_t **)&rhs_ptr); in arm_nn_mat_mult_nt_t_s8() 295 val4 = arm_nn_read_s8x4((const int8_t *)&rhs_ptr[rhs_off0]); in arm_nn_mat_mult_nt_t_s8() 317 val1 = arm_nn_read_s8x4_ia((const int8_t **)&rhs_ptr); in arm_nn_mat_mult_nt_t_s8() 320 val4 = arm_nn_read_s8x4((const int8_t *)&rhs_ptr[rhs_off0]); in arm_nn_mat_mult_nt_t_s8() 342 val1 = arm_nn_read_s8x4_ia((const int8_t **)&rhs_ptr); in arm_nn_mat_mult_nt_t_s8() 345 val4 = arm_nn_read_s8x4((const int8_t *)&rhs_ptr[rhs_off0]); in arm_nn_mat_mult_nt_t_s8() 372 val1 = arm_nn_read_s8x4_ia((const int8_t **)&rhs_ptr); in arm_nn_mat_mult_nt_t_s8() [all …]
|
D | arm_nn_vec_mat_mul_result_acc_s16.c | 65 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_vec_mat_mul_result_acc_s16() local 86 const int8_t *rhs_ptr_0 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s16() 87 const int8_t *rhs_ptr_1 = rhs_ptr + rhs_cols; in arm_nn_vec_mat_mul_result_acc_s16() 88 const int8_t *rhs_ptr_2 = rhs_ptr + rhs_cols * 2; in arm_nn_vec_mat_mul_result_acc_s16() 89 const int8_t *rhs_ptr_3 = rhs_ptr + rhs_cols * 3; in arm_nn_vec_mat_mul_result_acc_s16() 167 rhs_ptr += 4 * rhs_cols; in arm_nn_vec_mat_mul_result_acc_s16() 175 const int8_t *rhs_ptr0 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s16() 212 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s16() 230 const int8_t *rhs_0 = rhs_ptr; in arm_nn_vec_mat_mul_result_acc_s16() 231 rhs_ptr += rhs_cols; in arm_nn_vec_mat_mul_result_acc_s16() [all …]
|
D | arm_nn_vec_mat_mult_t_per_ch_s8.c | 150 const int8_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_per_ch_s8() local 161 const int8x16_t ker_0 = vldrbq_z_s8(rhs_ptr, p); in arm_nn_vec_mat_mult_t_per_ch_s8() 165 rhs_ptr += 16; in arm_nn_vec_mat_mult_t_per_ch_s8() 265 const int8_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_per_ch_s8() 273 int32_t ker_0 = arm_nn_read_s8x4_ia(&rhs_ptr); in arm_nn_vec_mat_mult_t_per_ch_s8() 285 acc_0 += lhs_temp * (*rhs_ptr + rhs_offset); in arm_nn_vec_mat_mult_t_per_ch_s8() 286 rhs_ptr++; in arm_nn_vec_mat_mult_t_per_ch_s8() 370 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_vec_mat_mult_t_per_ch_s8() 380 int32_t rhs_value0 = (int8_t)rhs_ptr[0] + rhs_offset; in arm_nn_vec_mat_mult_t_per_ch_s8() 385 ++rhs_ptr; in arm_nn_vec_mat_mult_t_per_ch_s8() [all …]
|
D | arm_nn_vec_mat_mult_t_s8.c | 149 const int8_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_s8() local 160 const int8x16_t ker_0 = vldrbq_z_s8(rhs_ptr, p); in arm_nn_vec_mat_mult_t_s8() 164 rhs_ptr += 16; in arm_nn_vec_mat_mult_t_s8() 264 const int8_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_s8() 272 int32_t ker_0 = arm_nn_read_s8x4_ia(&rhs_ptr); in arm_nn_vec_mat_mult_t_s8() 284 acc_0 += lhs_temp * (*rhs_ptr + rhs_offset); in arm_nn_vec_mat_mult_t_s8() 285 rhs_ptr++; in arm_nn_vec_mat_mult_t_s8() 368 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_vec_mat_mult_t_s8() 378 int32_t rhs_value0 = (int8_t)rhs_ptr[0] + rhs_offset; in arm_nn_vec_mat_mult_t_s8() 383 ++rhs_ptr; in arm_nn_vec_mat_mult_t_s8() [all …]
|
D | arm_nn_vec_mat_mult_t_s16.c | 172 const int8_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_s16() local 182 int16x8_t rhs_input = vldrbq_z_s16(rhs_ptr, pred); in arm_nn_vec_mat_mult_t_s16() 187 rhs_ptr += 8; in arm_nn_vec_mat_mult_t_s16() 203 result_64 += *rhs_ptr++ * lhs_temp; in arm_nn_vec_mat_mult_t_s16()
|
D | arm_nn_vec_mat_mult_t_s16_s16.c | 146 const int16_t *rhs_ptr = rhs; in arm_nn_vec_mat_mult_t_s16_s16() local 156 int16x8_t rhs_input = vldrhq_z_s16(rhs_ptr, pred); in arm_nn_vec_mat_mult_t_s16_s16() 161 rhs_ptr += 8; in arm_nn_vec_mat_mult_t_s16_s16()
|
D | arm_nn_vec_mat_mult_t_svdf_s8.c | 388 const int8_t *rhs_ptr = &rhs[0]; in arm_nn_vec_mat_mult_t_svdf_s8() local 394 int32_t rhs_value0 = (int8_t)rhs_ptr[0]; in arm_nn_vec_mat_mult_t_svdf_s8() 399 ++rhs_ptr; in arm_nn_vec_mat_mult_t_svdf_s8()
|