Created
September 27, 2023 21:02
-
-
Save hushell/23e68eef236673655db4635b80297c40 to your computer and use it in GitHub Desktop.
qnn-net-run profiling
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This is the detailed profiling: | |
``` | |
Execute Stats: | |
Execute Stat 1 | |
Number of HVX threads used : 4 count | |
RPC (execute) time : 2717302 us | |
QNN accelerator (execute) time : 2714698 us | |
Num times yield occured : 0 count | |
Time for initial VTCM acquire : 0 us | |
Time for HVX + HMX power on and acquire : 0 us | |
Accelerator (execute) time (cycles) : 454266408 cycles | |
Input OpId_2 (cycles) : 0 cycles | |
module_normalize_l2norm:OpId_16 (cycles) : 720336 cycles | |
layers_0_input_layernorm_module_mul:OpId_21 (cycles) : 1995579 cycles | |
t_8_converted_UFIXED_POINT_8:OpId_22 (cycles) : 129474 cycles | |
layers_0_self_attn_q_proj_fc:OpId_27 (cycles) : 0 cycles | |
layers_0_self_attn_q_proj_fc_post_reshape:OpId_29 (cycles) : 0 cycles | |
layers_0_self_attn_k_proj_fc:OpId_32 (cycles) : 0 cycles | |
layers_0_self_attn_k_proj_fc_post_reshape:OpId_34 (cycles) : 0 cycles | |
layers_0_self_attn_v_proj_fc:OpId_37 (cycles) : 0 cycles | |
layers_0_self_attn_v_proj_fc_post_reshape:OpId_39 (cycles) : 0 cycles | |
layers_0_self_attn_module_transpose:OpId_41 (cycles) : 299784 cycles | |
layers_0_self_attn_module_transpose_1:OpId_44 (cycles) : 286246 cycles | |
layers_0_self_attn_module_transpose_2:OpId_47 (cycles) : 0 cycles | |
Slice_96:OpId_50 (cycles) : 29021 cycles | |
Slice_97:OpId_56 (cycles) : 32867 cycles | |
module_cat:OpId_61 (cycles) : 0 cycles | |
module_mul_1:OpId_65 (cycles) : 400492 cycles | |
module_mul_2:OpId_67 (cycles) : 333347 cycles | |
module_add:OpId_68 (cycles) : 54992 cycles | |
Slice_114:OpId_70 (cycles) : 32122 cycles | |
Slice_115:OpId_76 (cycles) : 29130 cycles | |
module_cat_1:OpId_81 (cycles) : 0 cycles | |
module_mul_3:OpId_85 (cycles) : 448773 cycles | |
module_mul_4:OpId_87 (cycles) : 362554 cycles | |
module_add_1:OpId_88 (cycles) : 0 cycles | |
layers_0_self_attn_module_transpose_3:OpId_90 (cycles) : 0 cycles | |
module_matmul:OpId_92 (cycles) : 536031 cycles | |
module_add_2:OpId_97 (cycles) : 276870 cycles | |
layers_0_self_attn_softmax:OpId_98 (cycles) : 516530 cycles | |
module_matmul_1:OpId_102 (cycles) : 623121 cycles | |
layers_0_self_attn_module_transpose_4:OpId_107 (cycles) : 23827 cycles | |
t_208_converted_UFIXED_POINT_16:OpId_109 (cycles) : 625587 cycles | |
layers_0_self_attn_o_proj_fc:OpId_114 (cycles) : 489391 cycles | |
module_add_3:OpId_116 (cycles) : 690396 cycles | |
module_normalize_1_l2norm:OpId_117 (cycles) : 809801 cycles | |
layers_0_post_attention_layernorm_module_mul_5:OpId_122 (cycles) : 2734894 cycles | |
layers_0_mlp_gate_proj_fc:OpId_125 (cycles) : 9345088 cycles | |
layers_0_mlp_act_sigmoid:OpId_127 (cycles) : 0 cycles | |
layers_0_mlp_act_mul:OpId_128 (cycles) : 5520656 cycles | |
layers_0_mlp_up_proj_fc:OpId_131 (cycles) : 8584551 cycles | |
module_mul_6:OpId_133 (cycles) : 7618632 cycles | |
layers_0_mlp_down_proj_linear_fc:OpId_136 (cycles) : 13868851 cycles | |
module_mul_7:OpId_139 (cycles) : 2679924 cycles | |
module_add_4:OpId_140 (cycles) : 842321 cycles | |
module_normalize_2_l2norm:OpId_141 (cycles) : 1140908 cycles | |
layers_1_input_layernorm_module_mul_8:OpId_146 (cycles) : 2014922 cycles | |
t_332_converted_UFIXED_POINT_8:OpId_147 (cycles) : 153509 cycles | |
layers_1_self_attn_q_proj_fc:OpId_152 (cycles) : 0 cycles | |
layers_1_self_attn_q_proj_fc_post_reshape:OpId_154 (cycles) : 0 cycles | |
layers_1_self_attn_k_proj_fc:OpId_157 (cycles) : 0 cycles | |
layers_1_self_attn_k_proj_fc_post_reshape:OpId_159 (cycles) : 0 cycles | |
layers_1_self_attn_v_proj_fc:OpId_162 (cycles) : 0 cycles | |
layers_1_self_attn_v_proj_fc_post_reshape:OpId_164 (cycles) : 0 cycles | |
layers_1_self_attn_module_transpose_5:OpId_166 (cycles) : 392017 cycles | |
layers_1_self_attn_module_transpose_6:OpId_169 (cycles) : 420522 cycles | |
layers_1_self_attn_module_transpose_7:OpId_172 (cycles) : 0 cycles | |
Slice_242:OpId_175 (cycles) : 27514 cycles | |
Slice_243:OpId_181 (cycles) : 33508 cycles | |
module_cat_2:OpId_186 (cycles) : 0 cycles | |
module_mul_9:OpId_190 (cycles) : 486166 cycles | |
module_mul_10:OpId_192 (cycles) : 387203 cycles | |
module_add_5:OpId_193 (cycles) : 63153 cycles | |
Slice_260:OpId_195 (cycles) : 26937 cycles | |
Slice_261:OpId_201 (cycles) : 42007 cycles | |
module_cat_3:OpId_206 (cycles) : 0 cycles | |
module_mul_11:OpId_210 (cycles) : 621816 cycles | |
module_mul_12:OpId_212 (cycles) : 475887 cycles | |
module_add_6:OpId_213 (cycles) : 65669 cycles | |
layers_1_self_attn_module_transpose_8:OpId_215 (cycles) : 0 cycles | |
module_matmul_2:OpId_217 (cycles) : 956360 cycles | |
module_add_7:OpId_222 (cycles) : 81690 cycles | |
layers_1_self_attn_softmax:OpId_223 (cycles) : 679749 cycles | |
module_matmul_3:OpId_227 (cycles) : 956211 cycles | |
layers_1_self_attn_module_transpose_9:OpId_232 (cycles) : 29593 cycles | |
t_532_converted_UFIXED_POINT_16:OpId_234 (cycles) : 685186 cycles | |
layers_1_self_attn_o_proj_fc:OpId_239 (cycles) : 1050808 cycles | |
module_add_8:OpId_241 (cycles) : 668301 cycles | |
module_normalize_3_l2norm:OpId_242 (cycles) : 1645449 cycles | |
layers_1_post_attention_layernorm_module_mul_13:OpId_247 (cycles) : 3111318 cycles | |
layers_1_mlp_gate_proj_fc:OpId_250 (cycles) : 9352471 cycles | |
layers_1_mlp_act_sigmoid:OpId_252 (cycles) : 0 cycles | |
layers_1_mlp_act_mul:OpId_253 (cycles) : 5599450 cycles | |
layers_1_mlp_up_proj_fc:OpId_256 (cycles) : 9988779 cycles | |
module_mul_14:OpId_258 (cycles) : 7638426 cycles | |
layers_1_mlp_down_proj_linear_fc:OpId_261 (cycles) : 13863116 cycles | |
module_mul_15:OpId_264 (cycles) : 2667951 cycles | |
module_add_9:OpId_265 (cycles) : 891754 cycles | |
module_normalize_4_l2norm:OpId_266 (cycles) : 1149609 cycles | |
layers_2_input_layernorm_module_mul_16:OpId_271 (cycles) : 1997625 cycles | |
t_656_converted_UFIXED_POINT_8:OpId_272 (cycles) : 147053 cycles | |
layers_2_self_attn_q_proj_fc:OpId_277 (cycles) : 0 cycles | |
layers_2_self_attn_q_proj_fc_post_reshape:OpId_279 (cycles) : 0 cycles | |
layers_2_self_attn_k_proj_fc:OpId_282 (cycles) : 0 cycles | |
layers_2_self_attn_k_proj_fc_post_reshape:OpId_284 (cycles) : 0 cycles | |
layers_2_self_attn_v_proj_fc:OpId_287 (cycles) : 0 cycles | |
layers_2_self_attn_v_proj_fc_post_reshape:OpId_289 (cycles) : 0 cycles | |
layers_2_self_attn_module_transpose_10:OpId_291 (cycles) : 422096 cycles | |
layers_2_self_attn_module_transpose_11:OpId_294 (cycles) : 426632 cycles | |
layers_2_self_attn_module_transpose_12:OpId_297 (cycles) : 0 cycles | |
Slice_388:OpId_300 (cycles) : 28197 cycles | |
Slice_389:OpId_306 (cycles) : 31769 cycles | |
module_cat_4:OpId_311 (cycles) : 0 cycles | |
module_mul_17:OpId_315 (cycles) : 452789 cycles | |
module_mul_18:OpId_317 (cycles) : 349301 cycles | |
module_add_10:OpId_318 (cycles) : 56255 cycles | |
Slice_406:OpId_320 (cycles) : 26844 cycles | |
Slice_407:OpId_326 (cycles) : 35801 cycles | |
module_cat_5:OpId_331 (cycles) : 0 cycles | |
module_mul_19:OpId_335 (cycles) : 659527 cycles | |
module_mul_20:OpId_337 (cycles) : 476845 cycles | |
module_add_11:OpId_338 (cycles) : 0 cycles | |
layers_2_self_attn_module_transpose_13:OpId_340 (cycles) : 0 cycles | |
module_matmul_4:OpId_342 (cycles) : 712224 cycles | |
module_add_12:OpId_347 (cycles) : 93879 cycles | |
layers_2_self_attn_softmax:OpId_348 (cycles) : 630206 cycles | |
module_matmul_5:OpId_352 (cycles) : 873587 cycles | |
layers_2_self_attn_module_transpose_14:OpId_357 (cycles) : 26956 cycles | |
t_856_converted_UFIXED_POINT_16:OpId_359 (cycles) : 681626 cycles | |
layers_2_self_attn_o_proj_fc:OpId_364 (cycles) : 1044670 cycles | |
module_add_13:OpId_366 (cycles) : 614637 cycles | |
module_normalize_5_l2norm:OpId_367 (cycles) : 1676008 cycles | |
layers_2_post_attention_layernorm_module_mul_21:OpId_372 (cycles) : 3091763 cycles | |
layers_2_mlp_gate_proj_fc:OpId_375 (cycles) : 9341540 cycles | |
layers_2_mlp_act_sigmoid:OpId_377 (cycles) : 0 cycles | |
layers_2_mlp_act_mul:OpId_378 (cycles) : 5529062 cycles | |
layers_2_mlp_up_proj_fc:OpId_381 (cycles) : 9967503 cycles | |
module_mul_22:OpId_383 (cycles) : 7642190 cycles | |
layers_2_mlp_down_proj_linear_fc:OpId_386 (cycles) : 13905308 cycles | |
module_mul_23:OpId_389 (cycles) : 2663889 cycles | |
module_add_14:OpId_390 (cycles) : 876457 cycles | |
module_normalize_6_l2norm:OpId_391 (cycles) : 1170357 cycles | |
layers_3_input_layernorm_module_mul_24:OpId_396 (cycles) : 2024309 cycles | |
t_980_converted_UFIXED_POINT_8:OpId_397 (cycles) : 130417 cycles | |
layers_3_self_attn_q_proj_fc:OpId_402 (cycles) : 0 cycles | |
layers_3_self_attn_q_proj_fc_post_reshape:OpId_404 (cycles) : 0 cycles | |
layers_3_self_attn_k_proj_fc:OpId_407 (cycles) : 0 cycles | |
layers_3_self_attn_k_proj_fc_post_reshape:OpId_409 (cycles) : 0 cycles | |
layers_3_self_attn_v_proj_fc:OpId_412 (cycles) : 0 cycles | |
layers_3_self_attn_v_proj_fc_post_reshape:OpId_414 (cycles) : 0 cycles | |
layers_3_self_attn_module_transpose_15:OpId_416 (cycles) : 425592 cycles | |
layers_3_self_attn_module_transpose_16:OpId_419 (cycles) : 418683 cycles | |
layers_3_self_attn_module_transpose_17:OpId_422 (cycles) : 0 cycles | |
Slice_534:OpId_425 (cycles) : 38572 cycles | |
Slice_535:OpId_431 (cycles) : 27896 cycles | |
module_cat_6:OpId_436 (cycles) : 0 cycles | |
module_mul_25:OpId_440 (cycles) : 481049 cycles | |
module_mul_26:OpId_442 (cycles) : 360959 cycles | |
module_add_15:OpId_443 (cycles) : 58783 cycles | |
Slice_552:OpId_445 (cycles) : 28488 cycles | |
Slice_553:OpId_451 (cycles) : 45813 cycles | |
module_cat_7:OpId_456 (cycles) : 0 cycles | |
module_mul_27:OpId_460 (cycles) : 657318 cycles | |
module_mul_28:OpId_462 (cycles) : 495107 cycles | |
module_add_16:OpId_463 (cycles) : 0 cycles | |
layers_3_self_attn_module_transpose_18:OpId_465 (cycles) : 0 cycles | |
module_matmul_6:OpId_467 (cycles) : 741089 cycles | |
module_add_17:OpId_472 (cycles) : 77289 cycles | |
layers_3_self_attn_softmax:OpId_473 (cycles) : 632167 cycles | |
module_matmul_7:OpId_477 (cycles) : 929300 cycles | |
layers_3_self_attn_module_transpose_19:OpId_482 (cycles) : 31147 cycles | |
t_1180_converted_UFIXED_POINT_16:OpId_484 (cycles) : 680460 cycles | |
layers_3_self_attn_o_proj_fc:OpId_489 (cycles) : 1049739 cycles | |
module_add_18:OpId_491 (cycles) : 633146 cycles | |
module_normalize_7_l2norm:OpId_492 (cycles) : 1676759 cycles | |
layers_3_post_attention_layernorm_module_mul_29:OpId_497 (cycles) : 3071213 cycles | |
layers_3_mlp_gate_proj_fc:OpId_500 (cycles) : 9330336 cycles | |
layers_3_mlp_act_sigmoid:OpId_502 (cycles) : 0 cycles | |
layers_3_mlp_act_mul:OpId_503 (cycles) : 5526094 cycles | |
layers_3_mlp_up_proj_fc:OpId_506 (cycles) : 9995817 cycles | |
module_mul_30:OpId_508 (cycles) : 7287359 cycles | |
layers_3_mlp_down_proj_linear_fc:OpId_511 (cycles) : 12076258 cycles | |
module_mul_31:OpId_514 (cycles) : 2576820 cycles | |
module_add_19:OpId_515 (cycles) : 771181 cycles | |
module_normalize_8_l2norm:OpId_516 (cycles) : 1139915 cycles | |
layers_4_input_layernorm_module_mul_32:OpId_521 (cycles) : 2007767 cycles | |
t_1304_converted_UFIXED_POINT_8:OpId_522 (cycles) : 147629 cycles | |
layers_4_self_attn_q_proj_fc:OpId_527 (cycles) : 0 cycles | |
layers_4_self_attn_q_proj_fc_post_reshape:OpId_529 (cycles) : 0 cycles | |
layers_4_self_attn_k_proj_fc:OpId_532 (cycles) : 0 cycles | |
layers_4_self_attn_k_proj_fc_post_reshape:OpId_534 (cycles) : 0 cycles | |
layers_4_self_attn_v_proj_fc:OpId_537 (cycles) : 0 cycles | |
layers_4_self_attn_v_proj_fc_post_reshape:OpId_539 (cycles) : 0 cycles | |
layers_4_self_attn_module_transpose_20:OpId_541 (cycles) : 402315 cycles | |
layers_4_self_attn_module_transpose_21:OpId_544 (cycles) : 412260 cycles | |
layers_4_self_attn_module_transpose_22:OpId_547 (cycles) : 0 cycles | |
Slice_680:OpId_550 (cycles) : 27523 cycles | |
Slice_681:OpId_556 (cycles) : 32949 cycles | |
module_cat_8:OpId_561 (cycles) : 0 cycles | |
module_mul_33:OpId_565 (cycles) : 453289 cycles | |
module_mul_34:OpId_567 (cycles) : 361587 cycles | |
module_add_20:OpId_568 (cycles) : 60183 cycles | |
Slice_698:OpId_570 (cycles) : 26767 cycles | |
Slice_699:OpId_576 (cycles) : 37088 cycles | |
module_cat_9:OpId_581 (cycles) : 0 cycles | |
module_mul_35:OpId_585 (cycles) : 639908 cycles | |
module_mul_36:OpId_587 (cycles) : 478489 cycles | |
module_add_21:OpId_588 (cycles) : 0 cycles | |
layers_4_self_attn_module_transpose_23:OpId_590 (cycles) : 0 cycles | |
module_matmul_8:OpId_592 (cycles) : 724852 cycles | |
module_add_22:OpId_597 (cycles) : 73876 cycles | |
layers_4_self_attn_softmax:OpId_598 (cycles) : 639005 cycles | |
module_matmul_9:OpId_602 (cycles) : 885949 cycles | |
layers_4_self_attn_module_transpose_24:OpId_607 (cycles) : 29449 cycles | |
t_1504_converted_UFIXED_POINT_16:OpId_609 (cycles) : 681582 cycles | |
layers_4_self_attn_o_proj_fc:OpId_614 (cycles) : 1047058 cycles | |
module_add_23:OpId_616 (cycles) : 650599 cycles | |
module_normalize_9_l2norm:OpId_617 (cycles) : 1636957 cycles | |
layers_4_post_attention_layernorm_module_mul_37:OpId_622 (cycles) : 3093072 cycles | |
layers_4_mlp_gate_proj_fc:OpId_625 (cycles) : 9360564 cycles | |
layers_4_mlp_act_sigmoid:OpId_627 (cycles) : 0 cycles | |
layers_4_mlp_act_mul:OpId_628 (cycles) : 5542436 cycles | |
layers_4_mlp_up_proj_fc:OpId_631 (cycles) : 9999771 cycles | |
module_mul_38:OpId_633 (cycles) : 7600336 cycles | |
layers_4_mlp_down_proj_linear_fc:OpId_636 (cycles) : 14067964 cycles | |
module_mul_39:OpId_639 (cycles) : 2654440 cycles | |
module_add_24:OpId_640 (cycles) : 857210 cycles | |
module_normalize_10_l2norm:OpId_641 (cycles) : 1136597 cycles | |
layers_5_input_layernorm_module_mul_40:OpId_646 (cycles) : 2014477 cycles | |
t_1628_converted_UFIXED_POINT_8:OpId_647 (cycles) : 146355 cycles | |
layers_5_self_attn_q_proj_fc:OpId_652 (cycles) : 0 cycles | |
layers_5_self_attn_q_proj_fc_post_reshape:OpId_654 (cycles) : 0 cycles | |
layers_5_self_attn_k_proj_fc:OpId_657 (cycles) : 0 cycles | |
layers_5_self_attn_k_proj_fc_post_reshape:OpId_659 (cycles) : 0 cycles | |
layers_5_self_attn_v_proj_fc:OpId_662 (cycles) : 0 cycles | |
layers_5_self_attn_v_proj_fc_post_reshape:OpId_664 (cycles) : 0 cycles | |
layers_5_self_attn_module_transpose_25:OpId_666 (cycles) : 417986 cycles | |
layers_5_self_attn_module_transpose_26:OpId_669 (cycles) : 423985 cycles | |
layers_5_self_attn_module_transpose_27:OpId_672 (cycles) : 0 cycles | |
Slice_826:OpId_675 (cycles) : 29064 cycles | |
Slice_827:OpId_681 (cycles) : 37924 cycles | |
module_cat_10:OpId_686 (cycles) : 0 cycles | |
module_mul_41:OpId_690 (cycles) : 476416 cycles | |
module_mul_42:OpId_692 (cycles) : 384719 cycles | |
module_add_25:OpId_693 (cycles) : 65684 cycles | |
Slice_844:OpId_695 (cycles) : 34105 cycles | |
Slice_845:OpId_701 (cycles) : 38918 cycles | |
module_cat_11:OpId_706 (cycles) : 0 cycles | |
module_mul_43:OpId_710 (cycles) : 628067 cycles | |
module_mul_44:OpId_712 (cycles) : 508313 cycles | |
module_add_26:OpId_713 (cycles) : 72065 cycles | |
layers_5_self_attn_module_transpose_28:OpId_715 (cycles) : 0 cycles | |
module_matmul_10:OpId_717 (cycles) : 951132 cycles | |
module_add_27:OpId_722 (cycles) : 77933 cycles | |
layers_5_self_attn_softmax:OpId_723 (cycles) : 671679 cycles | |
module_matmul_11:OpId_727 (cycles) : 944964 cycles | |
layers_5_self_attn_module_transpose_29:OpId_732 (cycles) : 30179 cycles | |
t_1828_converted_UFIXED_POINT_16:OpId_734 (cycles) : 676162 cycles | |
layers_5_self_attn_o_proj_fc:OpId_739 (cycles) : 1044943 cycles | |
module_add_28:OpId_741 (cycles) : 630352 cycles | |
module_normalize_11_l2norm:OpId_742 (cycles) : 1644031 cycles | |
layers_5_post_attention_layernorm_module_mul_45:OpId_747 (cycles) : 3077339 cycles | |
layers_5_mlp_gate_proj_fc:OpId_750 (cycles) : 9341592 cycles | |
layers_5_mlp_act_sigmoid:OpId_752 (cycles) : 0 cycles | |
layers_5_mlp_act_mul:OpId_753 (cycles) : 5538362 cycles | |
layers_5_mlp_up_proj_fc:OpId_756 (cycles) : 9962701 cycles | |
module_mul_46:OpId_758 (cycles) : 7628900 cycles | |
layers_5_mlp_down_proj_linear_fc:OpId_761 (cycles) : 14057564 cycles | |
module_mul_47:OpId_764 (cycles) : 2691020 cycles | |
module_add_29:OpId_765 (cycles) : 867446 cycles | |
module_normalize_12_l2norm:OpId_766 (cycles) : 1161095 cycles | |
layers_6_input_layernorm_module_mul_48:OpId_771 (cycles) : 2007647 cycles | |
t_1952_converted_UFIXED_POINT_8:OpId_772 (cycles) : 150885 cycles | |
layers_6_self_attn_q_proj_fc:OpId_777 (cycles) : 0 cycles | |
layers_6_self_attn_q_proj_fc_post_reshape:OpId_779 (cycles) : 0 cycles | |
layers_6_self_attn_k_proj_fc:OpId_782 (cycles) : 0 cycles | |
layers_6_self_attn_k_proj_fc_post_reshape:OpId_784 (cycles) : 0 cycles | |
layers_6_self_attn_v_proj_fc:OpId_787 (cycles) : 0 cycles | |
layers_6_self_attn_v_proj_fc_post_reshape:OpId_789 (cycles) : 0 cycles | |
layers_6_self_attn_module_transpose_30:OpId_791 (cycles) : 424383 cycles | |
layers_6_self_attn_module_transpose_31:OpId_794 (cycles) : 450243 cycles | |
layers_6_self_attn_module_transpose_32:OpId_797 (cycles) : 0 cycles | |
Slice_972:OpId_800 (cycles) : 27781 cycles | |
Slice_973:OpId_806 (cycles) : 32502 cycles | |
module_cat_12:OpId_811 (cycles) : 0 cycles | |
module_mul_49:OpId_815 (cycles) : 446672 cycles | |
module_mul_50:OpId_817 (cycles) : 358457 cycles | |
module_add_30:OpId_818 (cycles) : 54450 cycles | |
Slice_990:OpId_820 (cycles) : 26513 cycles | |
Slice_991:OpId_826 (cycles) : 38472 cycles | |
module_cat_13:OpId_831 (cycles) : 0 cycles | |
module_mul_51:OpId_835 (cycles) : 688889 cycles | |
module_mul_52:OpId_837 (cycles) : 499602 cycles | |
module_add_31:OpId_838 (cycles) : 0 cycles | |
layers_6_self_attn_module_transpose_33:OpId_840 (cycles) : 0 cycles | |
module_matmul_12:OpId_842 (cycles) : 774614 cycles | |
module_add_32:OpId_847 (cycles) : 80632 cycles | |
layers_6_self_attn_softmax:OpId_848 (cycles) : 677180 cycles | |
module_matmul_13:OpId_852 (cycles) : 932013 cycles | |
layers_6_self_attn_module_transpose_34:OpId_857 (cycles) : 29286 cycles | |
t_2152_converted_UFIXED_POINT_16:OpId_859 (cycles) : 687697 cycles | |
layers_6_self_attn_o_proj_fc:OpId_864 (cycles) : 1042315 cycles | |
module_add_33:OpId_866 (cycles) : 645622 cycles | |
module_normalize_13_l2norm:OpId_867 (cycles) : 1636778 cycles | |
layers_6_post_attention_layernorm_module_mul_53:OpId_872 (cycles) : 3069745 cycles | |
layers_6_mlp_gate_proj_fc:OpId_875 (cycles) : 9344721 cycles | |
layers_6_mlp_act_sigmoid:OpId_877 (cycles) : 0 cycles | |
layers_6_mlp_act_mul:OpId_878 (cycles) : 5573093 cycles | |
layers_6_mlp_up_proj_fc:OpId_881 (cycles) : 10001903 cycles | |
module_mul_54:OpId_883 (cycles) : 7624913 cycles | |
layers_6_mlp_down_proj_linear_fc:OpId_886 (cycles) : 14041455 cycles | |
module_mul_55:OpId_889 (cycles) : 2671381 cycles | |
module_add_34:OpId_890 (cycles) : 721679 cycles | |
Output OpId_3 (cycles) : 79124 cycles | |
Accelerator (execute) time : 796071 us | |
QNN (execute) time : 2718790 us | |
``` |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment