Home
last modified time | relevance | path

Searched refs:sum (Results 1 – 25 of 122) sorted by relevance

12345

/hal_nxp-latest/mcux/mcux-sdk/CMSIS/NN/Source/BasicMathFunctions/
Darm_elementwise_add_s8.c128 int32_t sum; in arm_elementwise_add_s8()
163 sum = input_1 + input_2; in arm_elementwise_add_s8()
164 SAT_INPUT(sum, out_mult, out_shift); in arm_elementwise_add_s8()
165 sum += out_offset; in arm_elementwise_add_s8()
166 sum = MAX(sum, out_activation_min); in arm_elementwise_add_s8()
167 sum = MIN(sum, out_activation_max); in arm_elementwise_add_s8()
168 r1 = (q7_t)sum; in arm_elementwise_add_s8()
177 sum = input_1 + input_2; in arm_elementwise_add_s8()
178 SAT_INPUT(sum, out_mult, out_shift); in arm_elementwise_add_s8()
179 sum += out_offset; in arm_elementwise_add_s8()
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/DSP/Source/FilteringFunctions/
Darm_conv_fast_q31.c76 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_conv_fast_q31() local
150 sum = 0; in arm_conv_fast_q31()
160 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
164 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
168 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
172 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
186 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
194 *pOut++ = sum << 1; in arm_conv_fast_q31()
379 sum = 0; in arm_conv_fast_q31()
389 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_fast_q31()
[all …]
Darm_correlate_fast_q31.c78 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_correlate_fast_q31() local
174 sum = 0; in arm_correlate_fast_q31()
184 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
188 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
192 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
196 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
211 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
219 *pOut = sum << 1; in arm_correlate_fast_q31()
413 sum = 0; in arm_correlate_fast_q31()
423 sum = (q31_t) ((((q63_t) sum << 32) + in arm_correlate_fast_q31()
[all …]
Darm_conv_partial_fast_q31.c72 q31_t sum; /* Accumulators */ in arm_conv_partial_fast_q31() local
169 sum = 0; in arm_conv_partial_fast_q31()
179 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
183 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
187 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
191 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
211 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
219 *pOut++ = sum << 1; in arm_conv_partial_fast_q31()
416 sum = 0; in arm_conv_partial_fast_q31()
426 sum = (q31_t) ((((q63_t) sum << 32) + in arm_conv_partial_fast_q31()
[all …]
Darm_conv_partial_q15.c74 q63_t sum, acc0, acc1, acc2, acc3; /* Accumulator */ in arm_conv_partial_q15() local
174 sum = 0; in arm_conv_partial_q15()
183 sum = __SMLALD(*px++, *py--, sum); in arm_conv_partial_q15()
190 *pOut++ = (q15_t) (__SSAT((sum >> 15), 16)); in arm_conv_partial_q15()
212 sum = 0; in arm_conv_partial_q15()
223 sum = __SMLALDX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_partial_q15()
225 sum = __SMLALDX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_partial_q15()
242 sum = __SMLALD(*px++, *py--, sum); in arm_conv_partial_q15()
249 *pOut++ = (q15_t) (__SSAT((sum >> 15), 16)); in arm_conv_partial_q15()
482 sum = 0; in arm_conv_partial_q15()
[all …]
Darm_conv_fast_q15.c73 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_conv_fast_q15() local
154 sum = 0; in arm_conv_fast_q15()
163 sum = __SMLAD(*px++, *py--, sum); in arm_conv_fast_q15()
170 *pOut++ = (q15_t) (sum >> 15); in arm_conv_fast_q15()
192 sum = 0; in arm_conv_fast_q15()
203 sum = __SMLADX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_fast_q15()
205 sum = __SMLADX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_fast_q15()
222 sum = __SMLAD(*px++, *py--, sum); in arm_conv_fast_q15()
229 *pOut++ = (q15_t) (sum >> 15); in arm_conv_fast_q15()
454 sum = 0; in arm_conv_fast_q15()
[all …]
Darm_conv_partial_fast_q15.c68 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulator */ in arm_conv_partial_fast_q15() local
168 sum = 0; in arm_conv_partial_fast_q15()
177 sum = __SMLAD(*px++, *py--, sum); in arm_conv_partial_fast_q15()
184 *pOut++ = (q15_t) (sum >> 15); in arm_conv_partial_fast_q15()
206 sum = 0; in arm_conv_partial_fast_q15()
217 sum = __SMLADX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_partial_fast_q15()
219 sum = __SMLADX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_partial_fast_q15()
236 sum = __SMLAD(*px++, *py--, sum); in arm_conv_partial_fast_q15()
243 *pOut++ = (q15_t) (sum >> 15); in arm_conv_partial_fast_q15()
476 sum = 0; in arm_conv_partial_fast_q15()
[all …]
Darm_conv_q15.c236 q63_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_conv_q15() local
315 sum = 0; in arm_conv_q15()
324 sum = __SMLALD(*px++, *py--, sum); in arm_conv_q15()
331 *pOut++ = (q15_t) (__SSAT((sum >> 15), 16)); in arm_conv_q15()
353 sum = 0; in arm_conv_q15()
364 sum = __SMLALDX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_q15()
366 sum = __SMLALDX(read_q15x2_ia ((q15_t **) &px), read_q15x2_da ((q15_t **) &py), sum); in arm_conv_q15()
383 sum = __SMLALD(*px++, *py--, sum); in arm_conv_q15()
390 *pOut++ = (q15_t) (__SSAT((sum >> 15), 16)); in arm_conv_q15()
615 sum = 0; in arm_conv_q15()
[all …]
Darm_conv_partial_f32.c96 float32_t sum; /* Accumulator */ in arm_conv_partial_f32() local
193 sum = 0.0f; in arm_conv_partial_f32()
203 sum += *px++ * *py--; in arm_conv_partial_f32()
206 sum += *px++ * *py--; in arm_conv_partial_f32()
209 sum += *px++ * *py--; in arm_conv_partial_f32()
212 sum += *px++ * *py--; in arm_conv_partial_f32()
231 sum += *px++ * *py--; in arm_conv_partial_f32()
238 *pOut++ = sum; in arm_conv_partial_f32()
435 sum = 0.0f; in arm_conv_partial_f32()
445 sum += *px++ * *py--; in arm_conv_partial_f32()
[all …]
Darm_conv_partial_q31.c75 q63_t sum; /* Accumulator */ in arm_conv_partial_q31() local
172 sum = 0; in arm_conv_partial_q31()
182 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
185 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
188 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
191 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
210 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
217 *pOut++ = (q31_t) (sum >> 31); in arm_conv_partial_q31()
395 sum = 0; in arm_conv_partial_q31()
405 sum += (q63_t) *px++ * (*py--); in arm_conv_partial_q31()
[all …]
Darm_correlate_fast_q15.c73 q31_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_correlate_fast_q15() local
183 sum = 0; in arm_correlate_fast_q15()
193 sum = __SMLAD(read_q15x2_ia ((q15_t **) &px), read_q15x2_ia ((q15_t **) &py), sum); in arm_correlate_fast_q15()
195 sum = __SMLAD(read_q15x2_ia ((q15_t **) &px), read_q15x2_ia ((q15_t **) &py), sum); in arm_correlate_fast_q15()
209 sum = __SMLAD(*px++, *py++, sum); in arm_correlate_fast_q15()
216 *pOut = (q15_t) (sum >> 15); in arm_correlate_fast_q15()
445 sum = 0; in arm_correlate_fast_q15()
455 sum += ((q31_t) *px++ * *py++); in arm_correlate_fast_q15()
456 sum += ((q31_t) *px++ * *py++); in arm_correlate_fast_q15()
457 sum += ((q31_t) *px++ * *py++); in arm_correlate_fast_q15()
[all …]
Darm_conv_q31.c243 q63_t sum; /* Accumulators */ in arm_conv_q31() local
321 sum = 0; in arm_conv_q31()
331 sum += (q63_t) *px++ * (*py--); in arm_conv_q31()
334 sum += (q63_t) *px++ * (*py--); in arm_conv_q31()
337 sum += (q63_t) *px++ * (*py--); in arm_conv_q31()
340 sum += (q63_t) *px++ * (*py--); in arm_conv_q31()
359 sum += (q63_t) *px++ * *py--; in arm_conv_q31()
366 *pOut++ = (q31_t) (sum >> 31); in arm_conv_q31()
532 sum = 0; in arm_conv_q31()
542 sum += (q63_t) *px++ * *py--; in arm_conv_q31()
[all …]
Darm_correlate_q31.c276 q63_t sum; /* Accumulators */ in arm_correlate_q31() local
385 sum = 0; in arm_correlate_q31()
395 sum += (q63_t) *px++ * (*py++); in arm_correlate_q31()
398 sum += (q63_t) *px++ * (*py++); in arm_correlate_q31()
401 sum += (q63_t) *px++ * (*py++); in arm_correlate_q31()
404 sum += (q63_t) *px++ * (*py++); in arm_correlate_q31()
424 sum += (q63_t) *px++ * (*py++); in arm_correlate_q31()
431 *pOut = (q31_t) (sum >> 31); in arm_correlate_q31()
605 sum = 0; in arm_correlate_q31()
615 sum += (q63_t) *px++ * *py++; in arm_correlate_q31()
[all …]
Darm_correlate_q15.c281 q63_t sum, acc0, acc1, acc2, acc3; /* Accumulators */ in arm_correlate_q15() local
388 sum = 0; in arm_correlate_q15()
399 sum = __SMLALD(read_q15x2_ia ((q15_t **) &px), read_q15x2_ia ((q15_t **) &py), sum); in arm_correlate_q15()
401 sum = __SMLALD(read_q15x2_ia ((q15_t **) &px), read_q15x2_ia ((q15_t **) &py), sum); in arm_correlate_q15()
415 sum = __SMLALD(*px++, *py++, sum); in arm_correlate_q15()
422 *pOut = (q15_t) (__SSAT((sum >> 15), 16)); in arm_correlate_q15()
649 sum = 0; in arm_correlate_q15()
659 sum += ((q63_t) *px++ * *py++); in arm_correlate_q15()
660 sum += ((q63_t) *px++ * *py++); in arm_correlate_q15()
661 sum += ((q63_t) *px++ * *py++); in arm_correlate_q15()
[all …]
Darm_conv_f32.c262 float32_t sum; /* Accumulators */ in arm_conv_f32() local
349 sum = 0.0f; in arm_conv_f32()
383 sum += accum[0] + accum[1]; in arm_conv_f32()
392 sum += *px++ * *py--; in arm_conv_f32()
395 sum += *px++ * *py--; in arm_conv_f32()
398 sum += *px++ * *py--; in arm_conv_f32()
401 sum += *px++ * *py--; in arm_conv_f32()
421 sum += *px++ * *py--; in arm_conv_f32()
428 *pOut++ = sum; in arm_conv_f32()
694 sum = 0.0f; in arm_conv_f32()
[all …]
Darm_conv_partial_q7.c75 q31_t sum; /* Accumulator */ in arm_conv_partial_q7() local
174 sum = 0; in arm_conv_partial_q7()
195 sum = __SMLAD(input1, input2, sum); in arm_conv_partial_q7()
209 sum = __SMLAD(input1, input2, sum); in arm_conv_partial_q7()
228 sum += ((q31_t) * px++ * *py--); in arm_conv_partial_q7()
235 *pOut++ = (q7_t) (__SSAT(sum >> 7, 8)); in arm_conv_partial_q7()
478 sum = 0; in arm_conv_partial_q7()
498 sum = __SMLAD(input1, input2, sum); in arm_conv_partial_q7()
511 sum = __SMLAD(input1, input2, sum); in arm_conv_partial_q7()
530 sum += ((q31_t) * px++ * *py--); in arm_conv_partial_q7()
[all …]
Darm_correlate_f32.c317 float32_t sum; in arm_correlate_f32() local
427 sum = 0.0f; in arm_correlate_f32()
454 sum += accum[0] + accum[1]; in arm_correlate_f32()
463 sum += *px++ * *py++; in arm_correlate_f32()
466 sum += *px++ * *py++; in arm_correlate_f32()
469 sum += *px++ * *py++; in arm_correlate_f32()
472 sum += *px++ * *py++; in arm_correlate_f32()
493 sum += *px++ * *py++; in arm_correlate_f32()
500 *pOut = sum; in arm_correlate_f32()
771 sum = 0.0f; in arm_correlate_f32()
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/DSP/Source/MatrixFunctions/
Darm_mat_mult_fast_q15.c73 q31_t sum; /* Accumulator */ in arm_mat_mult_fast_q15() local
248 sum = 0; in arm_mat_mult_fast_q15()
281 sum = __SMLAD(inA1, inB1, sum); in arm_mat_mult_fast_q15()
290 sum += inA1 * inB1; in arm_mat_mult_fast_q15()
294 sum += inA2 * inB2; in arm_mat_mult_fast_q15()
298 sum += inA1 * inB1; in arm_mat_mult_fast_q15()
302 sum += inA2 * inB2; in arm_mat_mult_fast_q15()
316 sum += inA1 * inB1; in arm_mat_mult_fast_q15()
327 sum += (q31_t) *pInA++ * *pInB++; in arm_mat_mult_fast_q15()
335 *px++ = (q15_t) (sum >> 15); in arm_mat_mult_fast_q15()
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/NN/Source/PoolingFunctions/
Darm_avgpool_s8.c46 int32_t sum = buffer[i] > 0 ? (buffer[i] + half_count) : (buffer[i] - half_count); in scale_q31_to_q7_and_clamp() local
47 sum = sum / count; in scale_q31_to_q7_and_clamp()
48 sum = MAX(sum, act_min); in scale_q31_to_q7_and_clamp()
49 sum = MIN(sum, act_max); in scale_q31_to_q7_and_clamp()
51 target[i] = (q7_t)sum; in scale_q31_to_q7_and_clamp()
217 int32_t sum = 0; in arm_avgpool_s8() local
224 sum += pTmp[ch_src * (k_x + k_y * input_x)]; in arm_avgpool_s8()
228 sum = sum > 0 ? (sum + count / 2) / count : (sum - count / 2) / count; in arm_avgpool_s8()
229 sum = MAX(sum, act_min); in arm_avgpool_s8()
230 sum = MIN(sum, act_max); in arm_avgpool_s8()
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/DSP/Source/BasicMathFunctions/
Darm_dot_prod_q15.c68 q63_t sum = 0LL; in arm_dot_prod_q15() local
80 sum = vmlaldavaq(sum, vecA, vecB); in arm_dot_prod_q15()
100 sum = vmlaldavaq_p(sum, vecA, vecB, p0); in arm_dot_prod_q15()
103 *result = sum; in arm_dot_prod_q15()
114 q63_t sum = 0; /* Temporary return variable */ in arm_dot_prod_q15() local
127 sum = __SMLALD(read_q15x2_ia ((q15_t **) &pSrcA), read_q15x2_ia ((q15_t **) &pSrcB), sum); in arm_dot_prod_q15()
128 sum = __SMLALD(read_q15x2_ia ((q15_t **) &pSrcA), read_q15x2_ia ((q15_t **) &pSrcB), sum); in arm_dot_prod_q15()
130 sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q15()
131 sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q15()
132 sum += (q63_t)((q31_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q15()
[all …]
Darm_dot_prod_q7.c69 q31_t sum = 0; in arm_dot_prod_q7() local
81 sum = vmladavaq(sum, vecA, vecB); in arm_dot_prod_q7()
101 sum = vmladavaq_p(sum, vecA, vecB, p0); in arm_dot_prod_q7()
104 *result = sum; in arm_dot_prod_q7()
114 q31_t sum = 0; /* Temporary return variable */ in arm_dot_prod_q7() local
146 sum = __SMLAD(inA1, inB1, sum); in arm_dot_prod_q7()
147 sum = __SMLAD(inA2, inB2, sum); in arm_dot_prod_q7()
149 sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q7()
150 sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q7()
151 sum += (q31_t) ((q15_t) *pSrcA++ * *pSrcB++); in arm_dot_prod_q7()
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/NN/Source/FullyConnectedFunctions/
Darm_fully_connected_q7_opt.c153 q31_t sum = ((q31_t)(*pBias++) << bias_shift) + NN_ROUND(out_shift); in arm_fully_connected_q7_opt() local
174 sum = __SMLAD(inM11, inV, sum); in arm_fully_connected_q7_opt()
186 sum = __SMLAD(inM11, inV, sum); in arm_fully_connected_q7_opt()
205 sum = __SMLAD(inM12, inV, sum); in arm_fully_connected_q7_opt()
217 sum = __SMLAD(inM12, inV, sum); in arm_fully_connected_q7_opt()
269 : [ sum ] "+r"(sum), in arm_fully_connected_q7_opt()
307 : [ sum ] "+r"(sum), in arm_fully_connected_q7_opt()
328 sum += inV * inM; in arm_fully_connected_q7_opt()
334 *pO++ = (q7_t)(__SSAT((sum >> out_shift), 8)); in arm_fully_connected_q7_opt()
348 q31_t sum = ((q31_t)(*pBias++) << bias_shift) + NN_ROUND(out_shift); in arm_fully_connected_q7_opt() local
[all …]
/hal_nxp-latest/mcux/mcux-sdk/CMSIS/DSP/Source/StatisticsFunctions/
Darm_var_q15.c67 q63_t sum = 0LL; in arm_var_q15() local
85 sum = vaddvaq_s16(sum, vecSrc); in arm_var_q15()
106 sum += in; in arm_var_q15()
117 squareOfMean = arm_div_q63_to_q31((q63_t)sum * sum, (q31_t)(blockSize * (blockSize - 1U))); in arm_var_q15()
129 q31_t sum = 0; /* Accumulator */ in arm_var_q15() local
159 sum += ((in32 << 16U) >> 16U); in arm_var_q15()
160 sum += (in32 >> 16U); in arm_var_q15()
164 sum += ((in32 << 16U) >> 16U); in arm_var_q15()
165 sum += (in32 >> 16U); in arm_var_q15()
169 sum += in; in arm_var_q15()
[all …]
Darm_power_q15.c65 q63_t sum = 0LL; in arm_power_q15() local
76 sum = vmlaldavaq(sum, vecSrc, vecSrc); in arm_power_q15()
92 sum += ((q31_t) in * in); in arm_power_q15()
98 *pResult = sum; in arm_power_q15()
107 q63_t sum = 0; /* Temporary result storage */ in arm_power_q15() local
126 sum = __SMLALD(in32, in32, sum); in arm_power_q15()
129 sum = __SMLALD(in32, in32, sum); in arm_power_q15()
132 sum += ((q31_t) in * in); in arm_power_q15()
135 sum += ((q31_t) in * in); in arm_power_q15()
138 sum += ((q31_t) in * in); in arm_power_q15()
[all …]
Darm_power_q7.c64 q31_t sum = 0LL; in arm_power_q7() local
75 sum = vmladavaq(sum, vecSrc, vecSrc); in arm_power_q7()
91 sum += ((q15_t) in * in); in arm_power_q7()
97 *pResult = sum; in arm_power_q7()
106 q31_t sum = 0; /* Temporary result storage */ in arm_power_q7() local
131 sum = __SMLAD(in1, in1, sum); in arm_power_q7()
132 sum = __SMLAD(in2, in2, sum); in arm_power_q7()
135 sum += ((q15_t) in * in); in arm_power_q7()
138 sum += ((q15_t) in * in); in arm_power_q7()
141 sum += ((q15_t) in * in); in arm_power_q7()
[all …]

12345