Searched refs:sum_tmp (Results 1 – 6 of 6) sorted by relevance
| /cmsis-nn-latest/Source/NNSupportFunctions/ |
| D | arm_nn_mat_mul_core_4x_s8.c | 71 int32_t sum_tmp = 0; in arm_nn_mat_mul_core_4x_s8() local
77 sum_tmp += col; in arm_nn_mat_mul_core_4x_s8()
101 [sum] "+Te"(sum_tmp), in arm_nn_mat_mul_core_4x_s8()
115 sum_tmp *= conv_params->input_offset; in arm_nn_mat_mul_core_4x_s8()
118 sum_tmp += bias[i]; in arm_nn_mat_mul_core_4x_s8()
120 res = vaddq_n_s32(res, sum_tmp); in arm_nn_mat_mul_core_4x_s8()
|
| D | arm_nn_mat_mul_core_1x_s8.c | 71 int32_t sum_tmp = 0; in arm_nn_mat_mul_core_1x_s8() local
77 sum_tmp += col; in arm_nn_mat_mul_core_1x_s8()
91 … : [col] "+r"(col_base), [sum] "+Te"(sum_tmp), [row0] "+r"(row_base), [out0] "+Te"(acc_n0) in arm_nn_mat_mul_core_1x_s8()
96 sum_tmp *= conv_params->input_offset; in arm_nn_mat_mul_core_1x_s8()
97 acc_n0 += sum_tmp; in arm_nn_mat_mul_core_1x_s8()
|
| D | arm_nn_mat_mul_core_1x_s4.c | 74 int32_t sum_tmp = 0; in arm_nn_mat_mul_core_1x_s4() local
83 sum_tmp = vaddvaq_p_s8(sum_tmp, col_vec, rmdr_mask); in arm_nn_mat_mul_core_1x_s4()
91 sum_tmp *= conv_params->input_offset; in arm_nn_mat_mul_core_1x_s4()
92 acc_n0 += sum_tmp; in arm_nn_mat_mul_core_1x_s4()
|
| D | arm_nn_mat_mult_nt_interleaved_t_even_s4.c | 86 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_interleaved_t_even_s4() local
104 sum_tmp = vaddvaq_s8(sum_tmp, ker_low); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
105 sum_tmp = vaddvaq_s8(sum_tmp, ker_high); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
148 sum_tmp = vaddvaq_s8(sum_tmp, col_vec); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
174 sum_tmp = vaddvaq_p_s8(sum_tmp, col_vec, rmdr_mask); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
190 sum_tmp *= lhs_offset; in arm_nn_mat_mult_nt_interleaved_t_even_s4()
193 sum_tmp += bias[i]; in arm_nn_mat_mult_nt_interleaved_t_even_s4()
195 res = vaddq_n_s32(res, sum_tmp); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
223 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_interleaved_t_even_s4() local
236 sum_tmp = vaddvaq_s8(sum_tmp, ker_low); in arm_nn_mat_mult_nt_interleaved_t_even_s4()
[all …]
|
| D | arm_nn_mat_mult_nt_t_s4.c | 88 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_t_s4() local
96 sum_tmp = col; in arm_nn_mat_mult_nt_t_s4()
117 sum_tmp = vaddvaq_s8(sum_tmp, col_vec); in arm_nn_mat_mult_nt_t_s4()
143 sum_tmp = vaddvaq_p_s8(sum_tmp, col_vec, rmdr_mask); in arm_nn_mat_mult_nt_t_s4()
159 sum_tmp *= lhs_offset; in arm_nn_mat_mult_nt_t_s4()
162 sum_tmp += bias[i]; in arm_nn_mat_mult_nt_t_s4()
164 res = vaddq_n_s32(res, sum_tmp); in arm_nn_mat_mult_nt_t_s4()
194 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_t_s4() local
202 sum_tmp = col; in arm_nn_mat_mult_nt_t_s4()
221 sum_tmp = vaddvaq_s8(sum_tmp, col_vec); in arm_nn_mat_mult_nt_t_s4()
[all …]
|
| D | arm_nn_mat_mult_nt_t_s8.c | 81 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_t_s8() local
87 sum_tmp += col; in arm_nn_mat_mult_nt_t_s8()
118 [sum] "=Te"(sum_tmp), in arm_nn_mat_mult_nt_t_s8()
131 sum_tmp *= lhs_offset; in arm_nn_mat_mult_nt_t_s8()
134 sum_tmp += bias[i]; in arm_nn_mat_mult_nt_t_s8()
136 res = vaddq_n_s32(res, sum_tmp); in arm_nn_mat_mult_nt_t_s8()
163 int32_t sum_tmp = 0; in arm_nn_mat_mult_nt_t_s8() local
169 sum_tmp += col; in arm_nn_mat_mult_nt_t_s8()
185 … : [col] "+r"(col_base), [sum] "=Te"(sum_tmp), [row0] "+r"(lhs_vec), [out0] "=Te"(acc_n0) in arm_nn_mat_mult_nt_t_s8()
189 sum_tmp *= lhs_offset; in arm_nn_mat_mult_nt_t_s8()
[all …]
|
