Created
August 9, 2021 10:42
-
-
Save PhanDuc/5fcaf6a2f62e2fe90642c559c50699a1 to your computer and use it in GitHub Desktop.
horror_triton_console_output
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
| I0809 10:27:23.454255 1 logging.cc:52] Tactic: 861694390046228376 time 0.401024 | |
| I0809 10:27:23.457287 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.468172 1 logging.cc:52] Tactic: 5258189349241541167 time 0.214656 | |
| I0809 10:27:23.468615 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.479728 1 logging.cc:52] Tactic: 5821621277990374316 time 0.399456 | |
| I0809 10:27:23.480157 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.490417 1 logging.cc:52] Tactic: 5863767799113001648 time 0.117984 | |
| I0809 10:27:23.490856 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.501746 1 logging.cc:52] Tactic: -9147980667639709536 time 0.399328 | |
| I0809 10:27:23.502250 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.513258 1 logging.cc:52] Tactic: -8892196987859366827 time 0.399616 | |
| I0809 10:27:23.513700 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.529109 1 logging.cc:52] Tactic: -8850904373104590857 time 0.216416 | |
| I0809 10:27:23.529603 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.540780 1 logging.cc:52] Tactic: -8010679767156598961 time 0.1168 | |
| I0809 10:27:23.541303 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.552073 1 logging.cc:52] Tactic: -7751035352149795660 time 0.399584 | |
| I0809 10:27:23.552536 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.563815 1 logging.cc:52] Tactic: -5115676123557684531 time 0.39888 | |
| I0809 10:27:23.564290 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.574690 1 logging.cc:52] Tactic: -493597327599791285 time 0.208128 | |
| I0809 10:27:23.575140 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.585428 1 logging.cc:52] Tactic: -423878181466897819 time 0.118784 | |
| I0809 10:27:23.585870 1 logging.cc:52] Fastest Tactic: -8010679767156598961 Time: 0.1168 | |
| I0809 10:27:23.585930 1 logging.cc:52] --------------- Timing Runner: Conv_105 + Relu_106 (CudaConvolution) | |
| I0809 10:27:23.585944 1 logging.cc:52] CudaConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:23.585964 1 logging.cc:52] --------------- Timing Runner: Conv_105 + Relu_106 (CudaDepthwiseConvolution) | |
| I0809 10:27:23.585976 1 logging.cc:52] CudaDepthwiseConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:23.585993 1 logging.cc:52] --------------- Timing Runner: Conv_105 + Relu_106 (CublasConvolution) | |
| I0809 10:27:23.586010 1 logging.cc:52] CublasConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:23.586023 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: CaskConvolution Tactic: -8010679767156598961 | |
| I0809 10:27:23.586047 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.586062 1 logging.cc:52] | |
| I0809 10:27:23.601796 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.601926 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.601980 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.602024 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.602066 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.602127 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.602181 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.602239 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.602295 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:23.602356 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.602401 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.602456 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:23.602500 1 logging.cc:52] Conv_105 + Relu_106 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:23.607560 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,25088) -> Float(1,7,49,25088) *************** | |
| I0809 10:27:23.637932 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_medium_nn_v1 | |
| I0809 10:27:23.638088 1 logging.cc:52] Conv_107 + Relu_108 (scudnn_winograd) Set Tactic Name: volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | |
| I0809 10:27:23.638155 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_xregs_large_nn_v1 | |
| I0809 10:27:23.638212 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_small_nn_v1 | |
| I0809 10:27:23.638262 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_xregs_large_nn_v1 | |
| I0809 10:27:23.638325 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_small_nn_v1 | |
| I0809 10:27:23.638370 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_medium_nn_v1 | |
| I0809 10:27:23.638428 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:23.638470 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_small_nn_v1 | |
| I0809 10:27:23.640643 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (FusedConvActConvolution) | |
| I0809 10:27:23.708891 1 logging.cc:52] Tactic: 524287 time 0.413248 | |
| I0809 10:27:23.767828 1 logging.cc:52] Tactic: 720895 time 0.32384 | |
| I0809 10:27:23.826866 1 logging.cc:52] Tactic: 983039 time 0.159392 | |
| I0809 10:27:23.883327 1 logging.cc:52] Tactic: 1048575 time 0.260384 | |
| I0809 10:27:23.944499 1 logging.cc:52] Tactic: 1703935 time 0.13856 | |
| I0809 10:27:24.006803 1 logging.cc:52] Tactic: 1769471 time 0.141568 | |
| I0809 10:27:24.065768 1 logging.cc:52] Tactic: 1966079 time 0.665856 | |
| I0809 10:27:24.130570 1 logging.cc:52] Tactic: 2031615 time 0.563456 | |
| I0809 10:27:24.188978 1 logging.cc:52] Tactic: 2228223 time 0.274688 | |
| I0809 10:27:24.251749 1 logging.cc:52] Tactic: 2424831 time 0.101856 | |
| I0809 10:27:24.309429 1 logging.cc:52] Tactic: 2621439 time 0.101504 | |
| I0809 10:27:24.370296 1 logging.cc:52] Tactic: 2752511 time 0.358688 | |
| I0809 10:27:24.427243 1 logging.cc:52] Tactic: 2818047 time 0.327168 | |
| I0809 10:27:24.487980 1 logging.cc:52] Tactic: 2883583 time 0.782368 | |
| I0809 10:27:24.543535 1 logging.cc:52] Tactic: 3014655 time 0.165344 | |
| I0809 10:27:24.603222 1 logging.cc:52] Tactic: 3145727 time 0.186624 | |
| I0809 10:27:24.660489 1 logging.cc:52] Tactic: 3473407 time 0.395264 | |
| I0809 10:27:24.716376 1 logging.cc:52] Tactic: 3604479 time 0.165408 | |
| I0809 10:27:24.767513 1 logging.cc:52] Tactic: 3735551 time 0.33968 | |
| I0809 10:27:24.813256 1 logging.cc:52] Tactic: 4390911 time 0.702624 | |
| I0809 10:27:24.856154 1 logging.cc:52] Tactic: 5046271 time 0.217056 | |
| I0809 10:27:24.925326 1 logging.cc:52] Tactic: 5963775 time 0.604416 | |
| I0809 10:27:24.983589 1 logging.cc:52] Tactic: 6160383 time 0.339616 | |
| I0809 10:27:25.042201 1 logging.cc:52] Tactic: 6488063 time 0.290816 | |
| I0809 10:27:25.101418 1 logging.cc:52] Tactic: 6881279 time 0.51024 | |
| I0809 10:27:25.158918 1 logging.cc:52] Tactic: 7274495 time 0.1016 | |
| I0809 10:27:25.214957 1 logging.cc:52] Tactic: 7864319 time 0.105472 | |
| I0809 10:27:25.274265 1 logging.cc:52] Tactic: 7995391 time 0.33984 | |
| I0809 10:27:25.333222 1 logging.cc:52] Tactic: 8585215 time 0.464832 | |
| I0809 10:27:25.389143 1 logging.cc:52] Tactic: 8847359 time 0.10672 | |
| I0809 10:27:25.449480 1 logging.cc:52] Tactic: 8978431 time 0.611488 | |
| I0809 10:27:25.505218 1 logging.cc:52] Tactic: 9043967 time 0.149504 | |
| I0809 10:27:25.561690 1 logging.cc:52] Tactic: 9175039 time 0.165248 | |
| I0809 10:27:25.621336 1 logging.cc:52] Tactic: 9502719 time 0.712064 | |
| I0809 10:27:25.677178 1 logging.cc:52] Tactic: 9830399 time 0.313632 | |
| I0809 10:27:25.733041 1 logging.cc:52] Tactic: 9961471 time 0.116832 | |
| I0809 10:27:25.781205 1 logging.cc:52] Tactic: 10027007 time 0.2392 | |
| I0809 10:27:25.828585 1 logging.cc:52] Tactic: 10092543 time 0.704256 | |
| I0809 10:27:25.875895 1 logging.cc:52] Tactic: 10289151 time 0.666976 | |
| I0809 10:27:25.931271 1 logging.cc:52] Tactic: 10485759 time 0.123168 | |
| I0809 10:27:25.987500 1 logging.cc:52] Tactic: 10682367 time 0.096288 | |
| I0809 10:27:26.045030 1 logging.cc:52] Tactic: 10813439 time 0.175392 | |
| I0809 10:27:26.045630 1 logging.cc:52] Fastest Tactic: 10682367 Time: 0.096288 | |
| I0809 10:27:26.053137 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CaskConvolution) | |
| I0809 10:27:26.053184 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_medium_nn_v1 | |
| I0809 10:27:26.071160 1 logging.cc:52] Tactic: 1825138533642645384 time 0.909312 | |
| I0809 10:27:26.071572 1 logging.cc:52] Conv_107 + Relu_108 (scudnn_winograd) Set Tactic Name: volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | |
| I0809 10:27:26.087235 1 logging.cc:52] Tactic: 2775507031594384867 time 0.12928 | |
| I0809 10:27:26.087888 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_xregs_large_nn_v1 | |
| I0809 10:27:26.116113 1 logging.cc:52] Tactic: 2842488832350522458 time 0.60032 | |
| I0809 10:27:26.116647 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_small_nn_v1 | |
| I0809 10:27:26.147238 1 logging.cc:52] Tactic: 3915320020053085238 time 0.884128 | |
| I0809 10:27:26.147918 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_xregs_large_nn_v1 | |
| I0809 10:27:26.177574 1 logging.cc:52] Tactic: 6448355332020552203 time 0.925696 | |
| I0809 10:27:26.178117 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_small_nn_v1 | |
| I0809 10:27:26.205890 1 logging.cc:52] Tactic: 6808617066150061604 time 0.52656 | |
| I0809 10:27:26.206399 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_medium_nn_v1 | |
| I0809 10:27:26.234066 1 logging.cc:52] Tactic: -8060443123034038864 time 0.622592 | |
| I0809 10:27:26.234605 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:26.269166 1 logging.cc:52] Tactic: -4420849921117327522 time 0.647168 | |
| I0809 10:27:26.269728 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_small_nn_v1 | |
| I0809 10:27:26.301005 1 logging.cc:52] Tactic: -3946921629105938337 time 0.68608 | |
| I0809 10:27:26.301725 1 logging.cc:52] Fastest Tactic: 2775507031594384867 Time: 0.12928 | |
| I0809 10:27:26.302974 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CudaConvolution) | |
| I0809 10:27:26.310724 1 logging.cc:52] Tactic: 0 time 0.319104 | |
| I0809 10:27:26.319074 1 logging.cc:52] Tactic: 1 time 0.319488 | |
| I0809 10:27:26.326734 1 logging.cc:52] Tactic: 2 time 0.307456 | |
| I0809 10:27:26.327613 1 logging.cc:52] Tactic: 5 skipped. Scratch requested: 1145307136, available: 1073741824 | |
| I0809 10:27:26.333828 1 logging.cc:52] Tactic: 6 time 0.188416 | |
| I0809 10:27:26.343913 1 logging.cc:52] Tactic: 56 time 0.320576 | |
| I0809 10:27:26.352095 1 logging.cc:52] Tactic: 57 time 0.319104 | |
| I0809 10:27:26.359655 1 logging.cc:52] Tactic: 58 time 0.305152 | |
| I0809 10:27:26.360452 1 logging.cc:52] Tactic: 61 skipped. Scratch requested: 1145307136, available: 1073741824 | |
| I0809 10:27:26.366513 1 logging.cc:52] Tactic: 62 time 0.189952 | |
| I0809 10:27:26.367100 1 logging.cc:52] Fastest Tactic: 6 Time: 0.188416 | |
| I0809 10:27:26.367192 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CudaDepthwiseConvolution) | |
| I0809 10:27:26.367216 1 logging.cc:52] CudaDepthwiseConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.367231 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CublasConvolution) | |
| I0809 10:27:26.367240 1 logging.cc:52] CublasConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.367251 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: FusedConvActConvolution Tactic: 10682367 | |
| I0809 10:27:26.367261 1 logging.cc:52] | |
| I0809 10:27:26.368079 1 logging.cc:52] *************** Autotuning format combination: Float(512,3584,1,25088) -> Float(512,3584,1,25088) *************** | |
| I0809 10:27:26.391178 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.391325 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.391420 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.391553 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.391603 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.391651 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.391719 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.391783 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.391862 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.391981 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.392069 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.392921 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (FusedConvActConvolution) | |
| I0809 10:27:26.392975 1 logging.cc:52] FusedConvActConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.408369 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CaskConvolution) | |
| I0809 10:27:26.408443 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.425208 1 logging.cc:52] Tactic: 861694390046228376 time 0.89088 | |
| I0809 10:27:26.425740 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.442106 1 logging.cc:52] Tactic: 1017870653102653567 time 0.905216 | |
| I0809 10:27:26.446409 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.463490 1 logging.cc:52] Tactic: 5258189349241541167 time 0.469184 | |
| I0809 10:27:26.464016 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.479961 1 logging.cc:52] Tactic: 5821621277990374316 time 0.887072 | |
| I0809 10:27:26.480464 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.495147 1 logging.cc:52] Tactic: 5863767799113001648 time 0.251904 | |
| I0809 10:27:26.495853 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.514679 1 logging.cc:52] Tactic: -9147980667639709536 time 0.91744 | |
| I0809 10:27:26.515313 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.532912 1 logging.cc:52] Tactic: -8850904373104590857 time 0.480096 | |
| I0809 10:27:26.533537 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.552770 1 logging.cc:52] Tactic: -7751035352149795660 time 0.888608 | |
| I0809 10:27:26.553430 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.572848 1 logging.cc:52] Tactic: -3853827649136781465 time 0.907488 | |
| I0809 10:27:26.573511 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.588366 1 logging.cc:52] Tactic: -3263369460438823196 time 0.472896 | |
| I0809 10:27:26.588885 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.602011 1 logging.cc:52] Tactic: -423878181466897819 time 0.255328 | |
| I0809 10:27:26.602557 1 logging.cc:52] Fastest Tactic: 5863767799113001648 Time: 0.251904 | |
| I0809 10:27:26.602650 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CudaConvolution) | |
| I0809 10:27:26.602675 1 logging.cc:52] CudaConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.602707 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CudaDepthwiseConvolution) | |
| I0809 10:27:26.602736 1 logging.cc:52] CudaDepthwiseConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.602751 1 logging.cc:52] --------------- Timing Runner: Conv_107 + Relu_108 (CublasConvolution) | |
| I0809 10:27:26.602764 1 logging.cc:52] CublasConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.602776 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: CaskConvolution Tactic: 5863767799113001648 | |
| I0809 10:27:26.602804 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.602816 1 logging.cc:52] | |
| I0809 10:27:26.618542 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.618682 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.618743 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.618828 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.618943 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.619056 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.619176 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.619279 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.619447 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.619596 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_large_nhwc_tn_v1 | |
| I0809 10:27:26.619718 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.619845 1 logging.cc:52] Conv_107 + Relu_108 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.629042 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,25088), Float(1,7,49,100352) -> Float(1,7,49,100352) *************** | |
| I0809 10:27:26.649229 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_interior_nn_v1 | |
| I0809 10:27:26.649368 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_medium_nn_v1 | |
| I0809 10:27:26.649424 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_interior_nn_v1 | |
| I0809 10:27:26.649467 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_small_nn_v1 | |
| I0809 10:27:26.649513 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_small_nn_v1 | |
| I0809 10:27:26.649579 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_interior_nn_v1 | |
| I0809 10:27:26.649625 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_medium_nn_v1 | |
| I0809 10:27:26.649670 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:26.649740 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_small_nn_v1 | |
| I0809 10:27:26.663955 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CaskConvolution) | |
| I0809 10:27:26.664029 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_interior_nn_v1 | |
| I0809 10:27:26.686783 1 logging.cc:52] Tactic: 1754569683116234317 time 0.110784 | |
| I0809 10:27:26.687547 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_medium_nn_v1 | |
| I0809 10:27:26.709176 1 logging.cc:52] Tactic: 1825138533642645384 time 0.112 | |
| I0809 10:27:26.709796 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_interior_nn_v1 | |
| I0809 10:27:26.731671 1 logging.cc:52] Tactic: 2733356012094739613 time 0.081728 | |
| I0809 10:27:26.732233 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_small_nn_v1 | |
| I0809 10:27:26.754753 1 logging.cc:52] Tactic: 3915320020053085238 time 0.110784 | |
| I0809 10:27:26.755449 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_small_nn_v1 | |
| I0809 10:27:26.776832 1 logging.cc:52] Tactic: 6808617066150061604 time 0.069888 | |
| I0809 10:27:26.777422 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_interior_nn_v1 | |
| I0809 10:27:26.798760 1 logging.cc:52] Tactic: 9091006216302412844 time 0.067808 | |
| I0809 10:27:26.799419 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_medium_nn_v1 | |
| I0809 10:27:26.820636 1 logging.cc:52] Tactic: -8060443123034038864 time 0.073952 | |
| I0809 10:27:26.821209 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:26.842820 1 logging.cc:52] Tactic: -4420849921117327522 time 0.065568 | |
| I0809 10:27:26.843352 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_small_nn_v1 | |
| I0809 10:27:26.865498 1 logging.cc:52] Tactic: -3946921629105938337 time 0.08352 | |
| I0809 10:27:26.866036 1 logging.cc:52] Fastest Tactic: -4420849921117327522 Time: 0.065568 | |
| I0809 10:27:26.867133 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CudaConvolution) | |
| I0809 10:27:26.872469 1 logging.cc:52] Tactic: 0 time 0.06576 | |
| I0809 10:27:26.877997 1 logging.cc:52] Tactic: 1 time 0.057408 | |
| I0809 10:27:26.883929 1 logging.cc:52] Tactic: 2 time 0.137216 | |
| I0809 10:27:26.895501 1 logging.cc:52] Tactic: 5 time 1.07299 | |
| I0809 10:27:26.902602 1 logging.cc:52] Tactic: 56 time 0.065536 | |
| I0809 10:27:26.908581 1 logging.cc:52] Tactic: 57 time 0.057568 | |
| I0809 10:27:26.914895 1 logging.cc:52] Tactic: 58 time 0.138592 | |
| I0809 10:27:26.923777 1 logging.cc:52] Tactic: 61 time 1.07411 | |
| I0809 10:27:26.924414 1 logging.cc:52] Fastest Tactic: 1 Time: 0.057408 | |
| I0809 10:27:26.924516 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CublasConvolution) | |
| I0809 10:27:26.924539 1 logging.cc:52] CublasConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:26.924558 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: CudaConvolution Tactic: 1 | |
| I0809 10:27:26.924574 1 logging.cc:52] | |
| I0809 10:27:26.926790 1 logging.cc:52] *************** Autotuning format combination: Float(512,3584,1,25088), Float(2048,14336,1,100352) -> Float(2048,14336,1,100352) *************** | |
| I0809 10:27:26.952443 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.952603 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.952661 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.952703 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.952772 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.952817 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:26.952886 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.952931 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:26.952999 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.953044 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:26.953085 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:26.953126 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.967572 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CaskConvolution) | |
| I0809 10:27:26.967647 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:26.978461 1 logging.cc:52] Tactic: 861694390046228376 time 0.110112 | |
| I0809 10:27:26.978922 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:26.988926 1 logging.cc:52] Tactic: 5258189349241541167 time 0.062944 | |
| I0809 10:27:26.989462 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.020187 1 logging.cc:52] Tactic: 5821621277990374316 time 0.108928 | |
| I0809 10:27:27.021931 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.032904 1 logging.cc:52] Tactic: 5863767799113001648 time 0.040224 | |
| I0809 10:27:27.033453 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.044500 1 logging.cc:52] Tactic: -9147980667639709536 time 0.109856 | |
| I0809 10:27:27.045142 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.067731 1 logging.cc:52] Tactic: -8892196987859366827 time 0.109728 | |
| I0809 10:27:27.068583 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.081705 1 logging.cc:52] Tactic: -8850904373104590857 time 0.063616 | |
| I0809 10:27:27.082376 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.095506 1 logging.cc:52] Tactic: -8010679767156598961 time 0.03968 | |
| I0809 10:27:27.095992 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.105947 1 logging.cc:52] Tactic: -7751035352149795660 time 0.110208 | |
| I0809 10:27:27.106400 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.116796 1 logging.cc:52] Tactic: -5115676123557684531 time 0.109984 | |
| I0809 10:27:27.117271 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.127206 1 logging.cc:52] Tactic: -493597327599791285 time 0.061312 | |
| I0809 10:27:27.127728 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.137679 1 logging.cc:52] Tactic: -423878181466897819 time 0.04048 | |
| I0809 10:27:27.138112 1 logging.cc:52] Fastest Tactic: -8010679767156598961 Time: 0.03968 | |
| I0809 10:27:27.138173 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CudaConvolution) | |
| I0809 10:27:27.138187 1 logging.cc:52] CudaConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:27.138202 1 logging.cc:52] --------------- Timing Runner: Conv_109 + Add_110 + Relu_111 (CublasConvolution) | |
| I0809 10:27:27.138212 1 logging.cc:52] CublasConvolution has no valid tactics for this config, skipping | |
| I0809 10:27:27.138224 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: CaskConvolution Tactic: -8010679767156598961 | |
| I0809 10:27:27.138248 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.138276 1 logging.cc:52] | |
| I0809 10:27:27.153440 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.153570 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.153628 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.153672 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.153717 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.153773 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.153828 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.153889 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.153934 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.153992 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x128_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.154036 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.154078 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_medium_nhwc_tn_v1 | |
| I0809 10:27:27.154125 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.163087 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,100352) -> Float(1,7,49,25088) *************** | |
| I0809 10:27:27.165156 1 logging.cc:52] *************** Autotuning format combination: Float(2048,14336,1,100352) -> Float(512,3584,1,25088) *************** | |
| I0809 10:27:27.171122 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,25088) -> Float(1,7,49,25088) *************** | |
| I0809 10:27:27.173161 1 logging.cc:52] *************** Autotuning format combination: Float(512,3584,1,25088) -> Float(512,3584,1,25088) *************** | |
| I0809 10:27:27.182996 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,25088), Float(1,7,49,100352) -> Float(1,7,49,100352) *************** | |
| I0809 10:27:27.185091 1 logging.cc:52] *************** Autotuning format combination: Float(512,3584,1,25088), Float(2048,14336,1,100352) -> Float(2048,14336,1,100352) *************** | |
| I0809 10:27:27.187234 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:27.192084 1 logging.cc:52] Tactic: 1002 time 0.006176 | |
| I0809 10:27:27.194548 1 logging.cc:52] Tactic: 0 time 0.0072 | |
| I0809 10:27:27.194642 1 logging.cc:52] Fastest Tactic: 1002 Time: 0.006176 | |
| I0809 10:27:27.194931 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:27.199550 1 logging.cc:52] Tactic: 1002 time 0.0064 | |
| I0809 10:27:27.202129 1 logging.cc:52] Tactic: 0 time 0.007808 | |
| I0809 10:27:27.202216 1 logging.cc:52] Fastest Tactic: 1002 Time: 0.0064 | |
| I0809 10:27:27.204883 1 logging.cc:52] Adding reformat layer: Conv_0 + Relu_1 reformatted input 0 (INPUT__0) from Half(1,224,50176,150528) to Float(1,224,50176,150528) | |
| I0809 10:27:27.204990 1 logging.cc:52] Adding reformat layer: Conv_22 + Add_23 + Relu_24 output to be reformatted 0 (357) from Float(256,14336,1,802816) to Float(1,56,3136,802816) | |
| I0809 10:27:27.205022 1 logging.cc:52] Adding reformat layer: Conv_29 + Add_31 + Relu_32 output to be reformatted 0 (369) from Float(1,28,784,401408) to Float(512,14336,1,401408) | |
| I0809 10:27:27.205108 1 logging.cc:52] Adding reformat layer: Conv_51 + Add_52 + Relu_53 output to be reformatted 0 (399) from Float(512,14336,1,401408) to Float(1,28,784,401408) | |
| I0809 10:27:27.205206 1 logging.cc:52] Adding reformat layer: Conv_62 + Relu_63 reformatted input 0 (411) from Float(1024,14336,1,200704) to Float(1,14,196,200704) | |
| I0809 10:27:27.205297 1 logging.cc:52] Adding reformat layer: Conv_66 + Add_67 + Relu_68 reformatted input 0 (417) from Float(1,14,196,50176) to Float(256,3584,1,50176) | |
| I0809 10:27:27.205362 1 logging.cc:52] Adding reformat layer: Conv_69 + Relu_70 reformatted input 0 (421) from Float(1024,14336,1,200704) to Float(1,14,196,200704) | |
| I0809 10:27:27.205387 1 logging.cc:52] Adding reformat layer: Conv_73 + Add_74 + Relu_75 reformatted input 0 (427) from Float(1,14,196,50176) to Float(256,3584,1,50176) | |
| I0809 10:27:27.205409 1 logging.cc:52] Adding reformat layer: Conv_76 + Relu_77 reformatted input 0 (431) from Float(1024,14336,1,200704) to Float(1,14,196,200704) | |
| I0809 10:27:27.205435 1 logging.cc:52] Adding reformat layer: Conv_80 + Add_81 + Relu_82 reformatted input 0 (437) from Float(1,14,196,50176) to Float(256,3584,1,50176) | |
| I0809 10:27:27.205463 1 logging.cc:52] Adding reformat layer: Conv_83 + Relu_84 reformatted input 0 (441) from Float(1024,14336,1,200704) to Float(1,14,196,200704) | |
| I0809 10:27:27.205482 1 logging.cc:52] Adding reformat layer: Conv_87 + Add_88 + Relu_89 reformatted input 0 (447) from Float(1,14,196,50176) to Float(256,3584,1,50176) | |
| I0809 10:27:27.205505 1 logging.cc:52] Adding reformat layer: Conv_90 + Relu_91 reformatted input 0 (451) from Float(1024,14336,1,200704) to Float(1,14,196,200704) | |
| I0809 10:27:27.205534 1 logging.cc:52] Adding reformat layer: Conv_94 + Add_95 + Relu_96 reformatted input 0 (457) from Float(1,14,196,50176) to Float(256,3584,1,50176) | |
| I0809 10:27:27.205557 1 logging.cc:52] Adding reformat layer: Conv_105 + Relu_106 reformatted input 0 (473) from Float(2048,14336,1,100352) to Float(1,7,49,100352) | |
| I0809 10:27:27.205577 1 logging.cc:52] Adding reformat layer: Conv_109 + Add_110 + Relu_111 reformatted input 0 (479) from Float(1,7,49,25088) to Float(512,3584,1,25088) | |
| I0809 10:27:27.205605 1 logging.cc:52] Adding reformat layer: Conv_112 + Relu_113 reformatted input 0 (483) from Float(2048,14336,1,100352) to Float(1,7,49,100352) | |
| I0809 10:27:27.205626 1 logging.cc:52] Adding reformat layer: Conv_116 + Add_117 + Relu_118 reformatted input 0 (489) from Float(1,7,49,25088) to Float(512,3584,1,25088) | |
| I0809 10:27:27.205650 1 logging.cc:52] Adding reformat layer: Conv_116 + Add_117 + Relu_118 output to be reformatted 0 (OUTPUT__2) from Half(1,7,49,100352) to Float(2048,14336,1,100352) | |
| I0809 10:27:27.210013 1 logging.cc:52] Formats and tactics selection completed in 20.5648 seconds. | |
| I0809 10:27:27.210081 1 logging.cc:52] After reformat layers: 73 layers | |
| I0809 10:27:27.210566 1 logging.cc:52] Block size 1073741824 | |
| I0809 10:27:27.210608 1 logging.cc:52] Block size 3211264 | |
| I0809 10:27:27.210619 1 logging.cc:52] Block size 3211264 | |
| I0809 10:27:27.210629 1 logging.cc:52] Block size 1605632 | |
| I0809 10:27:27.210639 1 logging.cc:52] Block size 802816 | |
| I0809 10:27:27.210649 1 logging.cc:52] Total Activation Memory: 1082572800 | |
| I0809 10:27:27.210782 1 logging.cc:49] Detected 1 inputs and 1 output network tensors. | |
| I0809 10:27:27.211102 1 logging.cc:52] Conv_0 + Relu_1 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.211217 1 logging.cc:52] Conv_3 + Relu_4 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.212004 1 logging.cc:52] Conv_5 + Relu_6 (scudnn_winograd) Set Tactic Name: volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | |
| I0809 10:27:27.212420 1 logging.cc:52] Conv_8 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.212826 1 logging.cc:52] Conv_7 + Add_9 + Relu_10 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.214078 1 logging.cc:52] Conv_13 + Relu_14 (scudnn_winograd) Set Tactic Name: volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | |
| I0809 10:27:27.214545 1 logging.cc:52] Conv_15 + Add_16 + Relu_17 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.215993 1 logging.cc:52] Conv_20 + Relu_21 (scudnn_winograd) Set Tactic Name: volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | |
| I0809 10:27:27.216472 1 logging.cc:52] Conv_22 + Add_23 + Relu_24 (scudnn) Set Tactic Name: volta_scudnn_128x32_relu_medium_nn_v1 | |
| I0809 10:27:27.217183 1 logging.cc:52] Conv_25 + Relu_26 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.217967 1 logging.cc:52] Conv_27 + Relu_28 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.218361 1 logging.cc:52] Conv_30 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.219060 1 logging.cc:52] Conv_29 + Add_31 + Relu_32 (scudnn) Set Tactic Name: volta_scudnn_128x64_sliced1x2_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.220597 1 logging.cc:52] Conv_37 + Add_38 + Relu_39 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_interior_nn_v1 | |
| I0809 10:27:27.222851 1 logging.cc:52] Conv_44 + Add_45 + Relu_46 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_interior_nn_v1 | |
| I0809 10:27:27.225148 1 logging.cc:52] Conv_51 + Add_52 + Relu_53 (scudnn) Set Tactic Name: volta_scudnn_128x64_relu_interior_nn_v1 | |
| I0809 10:27:27.225871 1 logging.cc:52] Conv_54 + Relu_55 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.228827 1 logging.cc:52] Conv_56 + Relu_57 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.230187 1 logging.cc:52] Conv_59 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.232886 1 logging.cc:52] Conv_58 + Add_60 + Relu_61 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.238418 1 logging.cc:52] Conv_66 + Add_67 + Relu_68 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.248528 1 logging.cc:52] Conv_73 + Add_74 + Relu_75 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.259636 1 logging.cc:52] Conv_80 + Add_81 + Relu_82 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.270487 1 logging.cc:52] Conv_87 + Add_88 + Relu_89 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.281097 1 logging.cc:52] Conv_94 + Add_95 + Relu_96 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.285025 1 logging.cc:52] Conv_97 + Relu_98 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.302202 1 logging.cc:52] Conv_99 + Relu_100 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_small_nhwc_tn_v1 | |
| I0809 10:27:27.309805 1 logging.cc:52] Conv_102 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.325675 1 logging.cc:52] Conv_101 + Add_103 + Relu_104 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.354851 1 logging.cc:52] Conv_109 + Add_110 + Relu_111 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.395029 1 logging.cc:52] Conv_116 + Add_117 + Relu_118 (scudnn) Set Tactic Name: volta_scudnn_128x32_sliced1x4_ldg4_relu_exp_interior_nhwc_tn_v1 | |
| I0809 10:27:27.708893 1 logging.cc:52] Layer: Conv_0 + Relu_1 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.709086 1 logging.cc:52] Layer: Conv_0 + Relu_1 Weights: 0 HostPersistent: 2176 DevicePersistent: 113664 | |
| I0809 10:27:27.709107 1 logging.cc:52] Layer: MaxPool_2 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.709153 1 logging.cc:52] Layer: Conv_3 + Relu_4 Weights: 0 HostPersistent: 2176 DevicePersistent: 35840 | |
| I0809 10:27:27.709205 1 logging.cc:52] Layer: Conv_5 + Relu_6 Weights: 0 HostPersistent: 512 DevicePersistent: 410112 | |
| I0809 10:27:27.709275 1 logging.cc:52] Layer: Conv_8 Weights: 0 HostPersistent: 2176 DevicePersistent: 85504 | |
| I0809 10:27:27.709319 1 logging.cc:52] Layer: Conv_7 + Add_9 + Relu_10 Weights: 0 HostPersistent: 2176 DevicePersistent: 85504 | |
| I0809 10:27:27.709363 1 logging.cc:52] Layer: Conv_11 + Relu_12 Weights: 65536 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.709405 1 logging.cc:52] Layer: Conv_13 + Relu_14 Weights: 0 HostPersistent: 512 DevicePersistent: 410112 | |
| I0809 10:27:27.709462 1 logging.cc:52] Layer: Conv_15 + Add_16 + Relu_17 Weights: 0 HostPersistent: 2176 DevicePersistent: 85504 | |
| I0809 10:27:27.709477 1 logging.cc:52] Layer: Conv_18 + Relu_19 Weights: 65536 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.709566 1 logging.cc:52] Layer: Conv_20 + Relu_21 Weights: 0 HostPersistent: 512 DevicePersistent: 410112 | |
| I0809 10:27:27.709672 1 logging.cc:52] Layer: Conv_22 + Add_23 + Relu_24 Weights: 0 HostPersistent: 2176 DevicePersistent: 85504 | |
| I0809 10:27:27.709784 1 logging.cc:52] Layer: Conv_22 + Add_23 + Relu_24 output reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.709905 1 logging.cc:52] Layer: Conv_25 + Relu_26 Weights: 0 HostPersistent: 3200 DevicePersistent: 150528 | |
| I0809 10:27:27.710017 1 logging.cc:52] Layer: Conv_27 + Relu_28 Weights: 0 HostPersistent: 1664 DevicePersistent: 595456 | |
| I0809 10:27:27.710133 1 logging.cc:52] Layer: Conv_30 Weights: 0 HostPersistent: 3200 DevicePersistent: 269312 | |
| I0809 10:27:27.710233 1 logging.cc:52] Layer: Conv_29 + Add_31 + Relu_32 Weights: 0 HostPersistent: 3200 DevicePersistent: 531456 | |
| I0809 10:27:27.710250 1 logging.cc:52] Layer: Conv_29 + Add_31 + Relu_32 output reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710265 1 logging.cc:52] Layer: Conv_33 + Relu_34 Weights: 262144 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710277 1 logging.cc:52] Layer: Conv_35 + Relu_36 Weights: 589824 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710354 1 logging.cc:52] Layer: Conv_37 + Add_38 + Relu_39 Weights: 0 HostPersistent: 3200 DevicePersistent: 269312 | |
| I0809 10:27:27.710407 1 logging.cc:52] Layer: Conv_40 + Relu_41 Weights: 262144 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710432 1 logging.cc:52] Layer: Conv_42 + Relu_43 Weights: 589824 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710509 1 logging.cc:52] Layer: Conv_44 + Add_45 + Relu_46 Weights: 0 HostPersistent: 3200 DevicePersistent: 269312 | |
| I0809 10:27:27.710545 1 logging.cc:52] Layer: Conv_47 + Relu_48 Weights: 262144 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710558 1 logging.cc:52] Layer: Conv_49 + Relu_50 Weights: 589824 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710608 1 logging.cc:52] Layer: Conv_51 + Add_52 + Relu_53 Weights: 0 HostPersistent: 3200 DevicePersistent: 269312 | |
| I0809 10:27:27.710635 1 logging.cc:52] Layer: Conv_51 + Add_52 + Relu_53 output reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.710712 1 logging.cc:52] Layer: Conv_54 + Relu_55 Weights: 0 HostPersistent: 3200 DevicePersistent: 530432 | |
| I0809 10:27:27.710825 1 logging.cc:52] Layer: Conv_56 + Relu_57 Weights: 0 HostPersistent: 1664 DevicePersistent: 2361856 | |
| I0809 10:27:27.710904 1 logging.cc:52] Layer: Conv_59 Weights: 0 HostPersistent: 1664 DevicePersistent: 1054208 | |
| I0809 10:27:27.711010 1 logging.cc:52] Layer: Conv_58 + Add_60 + Relu_61 Weights: 0 HostPersistent: 3200 DevicePersistent: 2102784 | |
| I0809 10:27:27.711044 1 logging.cc:52] Layer: Conv_62 + Relu_63 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711069 1 logging.cc:52] Layer: Conv_62 + Relu_63 Weights: 1048576 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711098 1 logging.cc:52] Layer: Conv_64 + Relu_65 Weights: 2359296 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711122 1 logging.cc:52] Layer: Conv_66 + Add_67 + Relu_68 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711204 1 logging.cc:52] Layer: Conv_66 + Add_67 + Relu_68 Weights: 0 HostPersistent: 3200 DevicePersistent: 1054208 | |
| I0809 10:27:27.711232 1 logging.cc:52] Layer: Conv_69 + Relu_70 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711255 1 logging.cc:52] Layer: Conv_69 + Relu_70 Weights: 1048576 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711277 1 logging.cc:52] Layer: Conv_71 + Relu_72 Weights: 2359296 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711300 1 logging.cc:52] Layer: Conv_73 + Add_74 + Relu_75 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711433 1 logging.cc:52] Layer: Conv_73 + Add_74 + Relu_75 Weights: 0 HostPersistent: 3200 DevicePersistent: 1054208 | |
| I0809 10:27:27.711474 1 logging.cc:52] Layer: Conv_76 + Relu_77 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711528 1 logging.cc:52] Layer: Conv_76 + Relu_77 Weights: 1048576 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711549 1 logging.cc:52] Layer: Conv_78 + Relu_79 Weights: 2359296 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711573 1 logging.cc:52] Layer: Conv_80 + Add_81 + Relu_82 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711685 1 logging.cc:52] Layer: Conv_80 + Add_81 + Relu_82 Weights: 0 HostPersistent: 3200 DevicePersistent: 1054208 | |
| I0809 10:27:27.711705 1 logging.cc:52] Layer: Conv_83 + Relu_84 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711719 1 logging.cc:52] Layer: Conv_83 + Relu_84 Weights: 1048576 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711734 1 logging.cc:52] Layer: Conv_85 + Relu_86 Weights: 2359296 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711761 1 logging.cc:52] Layer: Conv_87 + Add_88 + Relu_89 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711835 1 logging.cc:52] Layer: Conv_87 + Add_88 + Relu_89 Weights: 0 HostPersistent: 3200 DevicePersistent: 1054208 | |
| I0809 10:27:27.711882 1 logging.cc:52] Layer: Conv_90 + Relu_91 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711907 1 logging.cc:52] Layer: Conv_90 + Relu_91 Weights: 1048576 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711943 1 logging.cc:52] Layer: Conv_92 + Relu_93 Weights: 2359296 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.711961 1 logging.cc:52] Layer: Conv_94 + Add_95 + Relu_96 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712045 1 logging.cc:52] Layer: Conv_94 + Add_95 + Relu_96 Weights: 0 HostPersistent: 3200 DevicePersistent: 1054208 | |
| I0809 10:27:27.712149 1 logging.cc:52] Layer: Conv_97 + Relu_98 Weights: 0 HostPersistent: 3200 DevicePersistent: 2100736 | |
| I0809 10:27:27.712252 1 logging.cc:52] Layer: Conv_99 + Relu_100 Weights: 0 HostPersistent: 1664 DevicePersistent: 9439744 | |
| I0809 10:27:27.712324 1 logging.cc:52] Layer: Conv_102 Weights: 0 HostPersistent: 3200 DevicePersistent: 4203008 | |
| I0809 10:27:27.712423 1 logging.cc:52] Layer: Conv_101 + Add_103 + Relu_104 Weights: 0 HostPersistent: 3200 DevicePersistent: 8397312 | |
| I0809 10:27:27.712473 1 logging.cc:52] Layer: Conv_105 + Relu_106 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712497 1 logging.cc:52] Layer: Conv_105 + Relu_106 Weights: 4194304 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712523 1 logging.cc:52] Layer: Conv_107 + Relu_108 Weights: 9437184 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712553 1 logging.cc:52] Layer: Conv_109 + Add_110 + Relu_111 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712611 1 logging.cc:52] Layer: Conv_109 + Add_110 + Relu_111 Weights: 0 HostPersistent: 3200 DevicePersistent: 4203008 | |
| I0809 10:27:27.712630 1 logging.cc:52] Layer: Conv_112 + Relu_113 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712643 1 logging.cc:52] Layer: Conv_112 + Relu_113 Weights: 4194304 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712655 1 logging.cc:52] Layer: Conv_114 + Relu_115 Weights: 9437184 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712671 1 logging.cc:52] Layer: Conv_116 + Add_117 + Relu_118 input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712743 1 logging.cc:52] Layer: Conv_116 + Add_117 + Relu_118 Weights: 0 HostPersistent: 3200 DevicePersistent: 4203008 | |
| I0809 10:27:27.712788 1 logging.cc:52] Layer: Conv_116 + Add_117 + Relu_118 output reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:27.712808 1 logging.cc:52] Total Host Persistent Memory: 78848 | |
| I0809 10:27:27.712826 1 logging.cc:52] Total Device Persistent Memory: 47943680 | |
| I0809 10:27:27.712863 1 logging.cc:52] Total Weight Memory: 46989312 | |
| I0809 10:27:27.723149 1 logging.cc:52] Engine generation completed in 21.0894 seconds. | |
| I0809 10:27:27.723217 1 logging.cc:52] Builder timing cache: created 91 entries, 188 hit(s) | |
| I0809 10:27:27.725183 1 logging.cc:52] Engine Layer Information: | |
| I0809 10:27:27.725254 1 logging.cc:52] Layer(Reformat): Conv_0 + Relu_1 input reformatter 0, Tactic: 1002, INPUT__0[Half(3,224,224)] -> Conv_0 + Relu_1 reformatted input 0[Float(3,224,224)] | |
| I0809 10:27:27.725276 1 logging.cc:52] Layer(scudnn): Conv_0 + Relu_1, Tactic: -4420849921117327522, Conv_0 + Relu_1 reformatted input 0[Float(3,224,224)] -> 324[Float(64,112,112)] | |
| I0809 10:27:27.725293 1 logging.cc:52] Layer(PoolingTiled): MaxPool_2, Tactic: 6947073, 324[Float(64,112,112)] -> 325[Float(64,56,56)] | |
| I0809 10:27:27.725309 1 logging.cc:52] Layer(scudnn): Conv_3 + Relu_4, Tactic: -4420849921117327522, 325[Float(64,56,56)] -> 328[Float(64,56,56)] | |
| I0809 10:27:27.725325 1 logging.cc:52] Layer(scudnn_winograd): Conv_5 + Relu_6, Tactic: 2775507031594384867, 328[Float(64,56,56)] -> 331[Float(64,56,56)] | |
| I0809 10:27:27.725337 1 logging.cc:52] Layer(scudnn): Conv_8, Tactic: -4420849921117327522, 331[Float(64,56,56)] -> 538[Float(256,56,56)] | |
| I0809 10:27:27.725352 1 logging.cc:52] Layer(scudnn): Conv_7 + Add_9 + Relu_10, Tactic: -4420849921117327522, 325[Float(64,56,56)], 538[Float(256,56,56)] -> 337[Float(256,56,56)] | |
| I0809 10:27:27.725367 1 logging.cc:52] Layer(FusedConvActDirect): Conv_11 + Relu_12, Tactic: 1179647, 337[Float(256,56,56)] -> 340[Float(64,56,56)] | |
| I0809 10:27:27.725381 1 logging.cc:52] Layer(scudnn_winograd): Conv_13 + Relu_14, Tactic: 2775507031594384867, 340[Float(64,56,56)] -> 343[Float(64,56,56)] | |
| I0809 10:27:27.725397 1 logging.cc:52] Layer(scudnn): Conv_15 + Add_16 + Relu_17, Tactic: -4420849921117327522, 343[Float(64,56,56)], 337[Float(256,56,56)] -> 347[Float(256,56,56)] | |
| I0809 10:27:27.725416 1 logging.cc:52] Layer(FusedConvActDirect): Conv_18 + Relu_19, Tactic: 1179647, 347[Float(256,56,56)] -> 350[Float(64,56,56)] | |
| I0809 10:27:27.725429 1 logging.cc:52] Layer(scudnn_winograd): Conv_20 + Relu_21, Tactic: 2775507031594384867, 350[Float(64,56,56)] -> 353[Float(64,56,56)] | |
| I0809 10:27:27.725447 1 logging.cc:52] Layer(scudnn): Conv_22 + Add_23 + Relu_24, Tactic: -4420849921117327522, 353[Float(64,56,56)], 347[Float(256,56,56)] -> Conv_22 + Add_23 + Relu_24 output to be reformatted 0[Float(256,56,56)] | |
| I0809 10:27:27.725462 1 logging.cc:52] Layer(Reformat): Conv_22 + Add_23 + Relu_24 output reformatter 0, Tactic: 1002, Conv_22 + Add_23 + Relu_24 output to be reformatted 0[Float(256,56,56)] -> 357[Float(256,56,56)] | |
| I0809 10:27:27.725498 1 logging.cc:52] Layer(scudnn): Conv_25 + Relu_26, Tactic: -493597327599791285, 357[Float(256,56,56)] -> 360[Float(128,56,56)] | |
| I0809 10:27:27.725513 1 logging.cc:52] Layer(scudnn): Conv_27 + Relu_28, Tactic: 5863767799113001648, 360[Float(128,56,56)] -> 363[Float(128,28,28)] | |
| I0809 10:27:27.725526 1 logging.cc:52] Layer(scudnn): Conv_30, Tactic: -493597327599791285, 363[Float(128,28,28)] -> 568[Float(512,28,28)] | |
| I0809 10:27:27.725546 1 logging.cc:52] Layer(scudnn): Conv_29 + Add_31 + Relu_32, Tactic: -493597327599791285, 357[Float(256,56,56)], 568[Float(512,28,28)] -> Conv_29 + Add_31 + Relu_32 output to be reformatted 0[Float(512,28,28)] | |
| I0809 10:27:27.725559 1 logging.cc:52] Layer(Reformat): Conv_29 + Add_31 + Relu_32 output reformatter 0, Tactic: 1002, Conv_29 + Add_31 + Relu_32 output to be reformatted 0[Float(512,28,28)] -> 369[Float(512,28,28)] | |
| I0809 10:27:27.725574 1 logging.cc:52] Layer(FusedConvActDirect): Conv_33 + Relu_34, Tactic: 5898239, 369[Float(512,28,28)] -> 372[Float(128,28,28)] | |
| I0809 10:27:27.725587 1 logging.cc:52] Layer(FusedConvActDirect): Conv_35 + Relu_36, Tactic: 8847359, 372[Float(128,28,28)] -> 375[Float(128,28,28)] | |
| I0809 10:27:27.725604 1 logging.cc:52] Layer(scudnn): Conv_37 + Add_38 + Relu_39, Tactic: 9091006216302412844, 375[Float(128,28,28)], 369[Float(512,28,28)] -> 379[Float(512,28,28)] | |
| I0809 10:27:27.725617 1 logging.cc:52] Layer(FusedConvActDirect): Conv_40 + Relu_41, Tactic: 5898239, 379[Float(512,28,28)] -> 382[Float(128,28,28)] | |
| I0809 10:27:27.725632 1 logging.cc:52] Layer(FusedConvActDirect): Conv_42 + Relu_43, Tactic: 8847359, 382[Float(128,28,28)] -> 385[Float(128,28,28)] | |
| I0809 10:27:27.725651 1 logging.cc:52] Layer(scudnn): Conv_44 + Add_45 + Relu_46, Tactic: 9091006216302412844, 385[Float(128,28,28)], 379[Float(512,28,28)] -> 389[Float(512,28,28)] | |
| I0809 10:27:27.725666 1 logging.cc:52] Layer(FusedConvActDirect): Conv_47 + Relu_48, Tactic: 5898239, 389[Float(512,28,28)] -> 392[Float(128,28,28)] | |
| I0809 10:27:27.725679 1 logging.cc:52] Layer(FusedConvActDirect): Conv_49 + Relu_50, Tactic: 8847359, 392[Float(128,28,28)] -> 395[Float(128,28,28)] | |
| I0809 10:27:27.725698 1 logging.cc:52] Layer(scudnn): Conv_51 + Add_52 + Relu_53, Tactic: 9091006216302412844, 395[Float(128,28,28)], 389[Float(512,28,28)] -> Conv_51 + Add_52 + Relu_53 output to be reformatted 0[Float(512,28,28)] | |
| I0809 10:27:27.725715 1 logging.cc:52] Layer(Reformat): Conv_51 + Add_52 + Relu_53 output reformatter 0, Tactic: 1002, Conv_51 + Add_52 + Relu_53 output to be reformatted 0[Float(512,28,28)] -> 399[Float(512,28,28)] | |
| I0809 10:27:27.725731 1 logging.cc:52] Layer(scudnn): Conv_54 + Relu_55, Tactic: -8010679767156598961, 399[Float(512,28,28)] -> 402[Float(256,28,28)] | |
| I0809 10:27:27.725745 1 logging.cc:52] Layer(scudnn): Conv_56 + Relu_57, Tactic: 5863767799113001648, 402[Float(256,28,28)] -> 405[Float(256,14,14)] | |
| I0809 10:27:27.725759 1 logging.cc:52] Layer(scudnn): Conv_59, Tactic: 5863767799113001648, 405[Float(256,14,14)] -> 607[Float(1024,14,14)] | |
| I0809 10:27:27.725777 1 logging.cc:52] Layer(scudnn): Conv_58 + Add_60 + Relu_61, Tactic: -8010679767156598961, 399[Float(512,28,28)], 607[Float(1024,14,14)] -> 411[Float(1024,14,14)] | |
| I0809 10:27:27.725806 1 logging.cc:52] Layer(Reformat): Conv_62 + Relu_63 input reformatter 0, Tactic: 1002, 411[Float(1024,14,14)] -> Conv_62 + Relu_63 reformatted input 0[Float(1024,14,14)] | |
| I0809 10:27:27.725820 1 logging.cc:52] Layer(FusedConvActDirect): Conv_62 + Relu_63, Tactic: 7012351, Conv_62 + Relu_63 reformatted input 0[Float(1024,14,14)] -> 414[Float(256,14,14)] | |
| I0809 10:27:27.725833 1 logging.cc:52] Layer(FusedConvActDirect): Conv_64 + Relu_65, Tactic: 7274495, 414[Float(256,14,14)] -> 417[Float(256,14,14)] | |
| I0809 10:27:27.725853 1 logging.cc:52] Layer(Reformat): Conv_66 + Add_67 + Relu_68 input reformatter 0, Tactic: 0, 417[Float(256,14,14)] -> Conv_66 + Add_67 + Relu_68 reformatted input 0[Float(256,14,14)] | |
| I0809 10:27:27.725882 1 logging.cc:52] Layer(scudnn): Conv_66 + Add_67 + Relu_68, Tactic: -8010679767156598961, Conv_66 + Add_67 + Relu_68 reformatted input 0[Float(256,14,14)], 411[Float(1024,14,14)] -> 421[Float(1024,14,14)] | |
| I0809 10:27:27.725922 1 logging.cc:52] Layer(Reformat): Conv_69 + Relu_70 input reformatter 0, Tactic: 1002, 421[Float(1024,14,14)] -> Conv_69 + Relu_70 reformatted input 0[Float(1024,14,14)] | |
| I0809 10:27:27.725945 1 logging.cc:52] Layer(FusedConvActDirect): Conv_69 + Relu_70, Tactic: 7012351, Conv_69 + Relu_70 reformatted input 0[Float(1024,14,14)] -> 424[Float(256,14,14)] | |
| I0809 10:27:27.725967 1 logging.cc:52] Layer(FusedConvActDirect): Conv_71 + Relu_72, Tactic: 7274495, 424[Float(256,14,14)] -> 427[Float(256,14,14)] | |
| I0809 10:27:27.725991 1 logging.cc:52] Layer(Reformat): Conv_73 + Add_74 + Relu_75 input reformatter 0, Tactic: 0, 427[Float(256,14,14)] -> Conv_73 + Add_74 + Relu_75 reformatted input 0[Float(256,14,14)] | |
| I0809 10:27:27.726019 1 logging.cc:52] Layer(scudnn): Conv_73 + Add_74 + Relu_75, Tactic: -8010679767156598961, Conv_73 + Add_74 + Relu_75 reformatted input 0[Float(256,14,14)], 421[Float(1024,14,14)] -> 431[Float(1024,14,14)] | |
| I0809 10:27:27.726055 1 logging.cc:52] Layer(Reformat): Conv_76 + Relu_77 input reformatter 0, Tactic: 1002, 431[Float(1024,14,14)] -> Conv_76 + Relu_77 reformatted input 0[Float(1024,14,14)] | |
| I0809 10:27:27.726074 1 logging.cc:52] Layer(FusedConvActDirect): Conv_76 + Relu_77, Tactic: 7012351, Conv_76 + Relu_77 reformatted input 0[Float(1024,14,14)] -> 434[Float(256,14,14)] | |
| I0809 10:27:27.726094 1 logging.cc:52] Layer(FusedConvActDirect): Conv_78 + Relu_79, Tactic: 7274495, 434[Float(256,14,14)] -> 437[Float(256,14,14)] | |
| I0809 10:27:27.726117 1 logging.cc:52] Layer(Reformat): Conv_80 + Add_81 + Relu_82 input reformatter 0, Tactic: 0, 437[Float(256,14,14)] -> Conv_80 + Add_81 + Relu_82 reformatted input 0[Float(256,14,14)] | |
| I0809 10:27:27.726140 1 logging.cc:52] Layer(scudnn): Conv_80 + Add_81 + Relu_82, Tactic: -8010679767156598961, Conv_80 + Add_81 + Relu_82 reformatted input 0[Float(256,14,14)], 431[Float(1024,14,14)] -> 441[Float(1024,14,14)] | |
| I0809 10:27:27.726174 1 logging.cc:52] Layer(Reformat): Conv_83 + Relu_84 input reformatter 0, Tactic: 1002, 441[Float(1024,14,14)] -> Conv_83 + Relu_84 reformatted input 0[Float(1024,14,14)] | |
| I0809 10:27:27.726196 1 logging.cc:52] Layer(FusedConvActDirect): Conv_83 + Relu_84, Tactic: 7012351, Conv_83 + Relu_84 reformatted input 0[Float(1024,14,14)] -> 444[Float(256,14,14)] | |
| I0809 10:27:27.726218 1 logging.cc:52] Layer(FusedConvActDirect): Conv_85 + Relu_86, Tactic: 7274495, 444[Float(256,14,14)] -> 447[Float(256,14,14)] | |
| I0809 10:27:27.726252 1 logging.cc:52] Layer(Reformat): Conv_87 + Add_88 + Relu_89 input reformatter 0, Tactic: 0, 447[Float(256,14,14)] -> Conv_87 + Add_88 + Relu_89 reformatted input 0[Float(256,14,14)] | |
| I0809 10:27:27.726277 1 logging.cc:52] Layer(scudnn): Conv_87 + Add_88 + Relu_89, Tactic: -8010679767156598961, Conv_87 + Add_88 + Relu_89 reformatted input 0[Float(256,14,14)], 441[Float(1024,14,14)] -> 451[Float(1024,14,14)] | |
| I0809 10:27:27.726300 1 logging.cc:52] Layer(Reformat): Conv_90 + Relu_91 input reformatter 0, Tactic: 1002, 451[Float(1024,14,14)] -> Conv_90 + Relu_91 reformatted input 0[Float(1024,14,14)] | |
| I0809 10:27:27.726323 1 logging.cc:52] Layer(FusedConvActDirect): Conv_90 + Relu_91, Tactic: 7012351, Conv_90 + Relu_91 reformatted input 0[Float(1024,14,14)] -> 454[Float(256,14,14)] | |
| I0809 10:27:27.726359 1 logging.cc:52] Layer(FusedConvActDirect): Conv_92 + Relu_93, Tactic: 7274495, 454[Float(256,14,14)] -> 457[Float(256,14,14)] | |
| I0809 10:27:27.726380 1 logging.cc:52] Layer(Reformat): Conv_94 + Add_95 + Relu_96 input reformatter 0, Tactic: 0, 457[Float(256,14,14)] -> Conv_94 + Add_95 + Relu_96 reformatted input 0[Float(256,14,14)] | |
| I0809 10:27:27.726402 1 logging.cc:52] Layer(scudnn): Conv_94 + Add_95 + Relu_96, Tactic: -8010679767156598961, Conv_94 + Add_95 + Relu_96 reformatted input 0[Float(256,14,14)], 451[Float(1024,14,14)] -> 461[Float(1024,14,14)] | |
| I0809 10:27:27.726441 1 logging.cc:52] Layer(scudnn): Conv_97 + Relu_98, Tactic: -8010679767156598961, 461[Float(1024,14,14)] -> 464[Float(512,14,14)] | |
| I0809 10:27:27.726464 1 logging.cc:52] Layer(scudnn): Conv_99 + Relu_100, Tactic: 5863767799113001648, 464[Float(512,14,14)] -> 467[Float(512,7,7)] | |
| I0809 10:27:27.726492 1 logging.cc:52] Layer(scudnn): Conv_102, Tactic: -8010679767156598961, 467[Float(512,7,7)] -> 664[Float(2048,7,7)] | |
| I0809 10:27:27.726517 1 logging.cc:52] Layer(scudnn): Conv_101 + Add_103 + Relu_104, Tactic: -8010679767156598961, 461[Float(1024,14,14)], 664[Float(2048,7,7)] -> 473[Float(2048,7,7)] | |
| I0809 10:27:27.726558 1 logging.cc:52] Layer(Reformat): Conv_105 + Relu_106 input reformatter 0, Tactic: 1002, 473[Float(2048,7,7)] -> Conv_105 + Relu_106 reformatted input 0[Float(2048,7,7)] | |
| I0809 10:27:27.726585 1 logging.cc:52] Layer(FusedConvActDirect): Conv_105 + Relu_106, Tactic: 7012351, Conv_105 + Relu_106 reformatted input 0[Float(2048,7,7)] -> 476[Float(512,7,7)] | |
| I0809 10:27:27.726626 1 logging.cc:52] Layer(FusedConvActDirect): Conv_107 + Relu_108, Tactic: 10682367, 476[Float(512,7,7)] -> 479[Float(512,7,7)] | |
| I0809 10:27:27.726643 1 logging.cc:52] Layer(Reformat): Conv_109 + Add_110 + Relu_111 input reformatter 0, Tactic: 0, 479[Float(512,7,7)] -> Conv_109 + Add_110 + Relu_111 reformatted input 0[Float(512,7,7)] | |
| I0809 10:27:27.726659 1 logging.cc:52] Layer(scudnn): Conv_109 + Add_110 + Relu_111, Tactic: -8010679767156598961, Conv_109 + Add_110 + Relu_111 reformatted input 0[Float(512,7,7)], 473[Float(2048,7,7)] -> 483[Float(2048,7,7)] | |
| I0809 10:27:27.726681 1 logging.cc:52] Layer(Reformat): Conv_112 + Relu_113 input reformatter 0, Tactic: 1002, 483[Float(2048,7,7)] -> Conv_112 + Relu_113 reformatted input 0[Float(2048,7,7)] | |
| I0809 10:27:27.726706 1 logging.cc:52] Layer(FusedConvActDirect): Conv_112 + Relu_113, Tactic: 7012351, Conv_112 + Relu_113 reformatted input 0[Float(2048,7,7)] -> 486[Float(512,7,7)] | |
| I0809 10:27:27.726745 1 logging.cc:52] Layer(FusedConvActDirect): Conv_114 + Relu_115, Tactic: 10682367, 486[Float(512,7,7)] -> 489[Float(512,7,7)] | |
| I0809 10:27:27.726770 1 logging.cc:52] Layer(Reformat): Conv_116 + Add_117 + Relu_118 input reformatter 0, Tactic: 0, 489[Float(512,7,7)] -> Conv_116 + Add_117 + Relu_118 reformatted input 0[Float(512,7,7)] | |
| I0809 10:27:27.726796 1 logging.cc:52] Layer(scudnn): Conv_116 + Add_117 + Relu_118, Tactic: -8010679767156598961, Conv_116 + Add_117 + Relu_118 reformatted input 0[Float(512,7,7)], 483[Float(2048,7,7)] -> Conv_116 + Add_117 + Relu_118 output to be reformatted 0[Float(2048,7,7)] | |
| I0809 10:27:27.726835 1 logging.cc:52] Layer(Reformat): Conv_116 + Add_117 + Relu_118 output reformatter 0, Tactic: 1002, Conv_116 + Add_117 + Relu_118 output to be reformatted 0[Float(2048,7,7)] -> OUTPUT__2[Half(2048,7,7)] | |
| I0809 10:27:27.733127 1 logging.cc:52] Allocated persistent device memory of size 47943680 | |
| I0809 10:27:27.734676 1 logging.cc:52] Allocated activation device memory of size 10436608 | |
| I0809 10:27:27.734942 1 logging.cc:52] Assigning persistent memory blocks for various profiles | |
| 2021-08-09 10:27:27.735548759 [I:onnxruntime:log, bfc_arena.cc:273 AllocateRawInternal] Extending BFCArena for Tensorrt. bin_num:9 rounded_bytes:200704 | |
| 2021-08-09 10:27:27.735666831 [I:onnxruntime:log, bfc_arena.cc:158 Extend] Extended allocation by 1048576 bytes. | |
| 2021-08-09 10:27:27.735683788 [I:onnxruntime:log, bfc_arena.cc:161 Extend] Total allocated bytes: 50111488 | |
| 2021-08-09 10:27:27.735697290 [I:onnxruntime:log, bfc_arena.cc:164 Extend] Allocated memory at 0x7f4248d00000 to 0x7f4248e00000 | |
| 2021-08-09 10:27:27.739567196 [I:onnxruntime:log, bfc_arena.cc:273 AllocateRawInternal] Extending BFCArena for TensorrtPinned. bin_num:9 rounded_bytes:200704 | |
| 2021-08-09 10:27:27.740094820 [I:onnxruntime:log, bfc_arena.cc:158 Extend] Extended allocation by 1048576 bytes. | |
| 2021-08-09 10:27:27.740155024 [I:onnxruntime:log, bfc_arena.cc:161 Extend] Total allocated bytes: 1048576 | |
| 2021-08-09 10:27:27.740170332 [I:onnxruntime:log, bfc_arena.cc:164 Extend] Allocated memory at 0x7f445b201600 to 0x7f445b301600 | |
| I0809 10:27:27.744740 1 logging.cc:52] Applying generic optimizations to the graph for inference. | |
| I0809 10:27:27.744810 1 logging.cc:52] Original: 18 layers | |
| I0809 10:27:27.744881 1 logging.cc:52] After dead-layer removal: 18 layers | |
| I0809 10:27:27.744961 1 logging.cc:52] BinaryFusion: Fusing (Unnamed Layer* 0) [Constant] with (Unnamed Layer* 1) [Shuffle] | |
| I0809 10:27:27.745129 1 logging.cc:52] BinaryFusion: Fusing (Unnamed Layer* 5) [Constant] with (Unnamed Layer* 6) [Shuffle] | |
| I0809 10:27:27.745246 1 logging.cc:52] BinaryFusion: Fusing (Unnamed Layer* 11) [Constant] with (Unnamed Layer* 12) [Shuffle] | |
| I0809 10:27:27.745444 1 logging.cc:52] After Myelin optimization: 15 layers | |
| I0809 10:27:27.745726 1 logging.cc:52] After scale fusion: 15 layers | |
| I0809 10:27:27.746049 1 logging.cc:52] Swap the layer type of GlobalAveragePool_121 from REDUCE to POOLING | |
| I0809 10:27:27.748834 1 logging.cc:52] BinaryFusion: Fusing (Unnamed Layer* 15) [ElementWise] with ReduceL2_135 | |
| I0809 10:27:27.748977 1 logging.cc:52] BinaryFusion: Fusing (Unnamed Layer* 15) [ElementWise] + ReduceL2_135 with ReduceL2_135_8 | |
| I0809 10:27:27.757014 1 logging.cc:52] BinaryFusionBase: Fusing (Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle] with Pow_120 | |
| I0809 10:27:27.757794 1 logging.cc:52] BinaryFusionBase: Fusing (Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle] with Pow_123 | |
| I0809 10:27:27.768448 1 logging.cc:52] After vertical fusions: 11 layers | |
| I0809 10:27:27.768574 1 logging.cc:52] After dupe layer removal: 11 layers | |
| I0809 10:27:27.768619 1 logging.cc:52] After final dead-layer removal: 11 layers | |
| I0809 10:27:27.768648 1 logging.cc:52] After tensor merging: 11 layers | |
| I0809 10:27:27.768696 1 logging.cc:52] After concat removal: 11 layers | |
| I0809 10:27:27.768769 1 logging.cc:52] Graph construction and optimization completed in 0.0247426 seconds. | |
| I0809 10:27:27.776015 1 logging.cc:52] Constructing optimization profile number 0 [1/1]. | |
| I0809 10:27:27.778013 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:27.782783 1 logging.cc:52] Tactic: 1002 time 0.006176 | |
| I0809 10:27:27.784992 1 logging.cc:52] Tactic: 0 time 0.006176 | |
| I0809 10:27:27.785063 1 logging.cc:52] Fastest Tactic: 1002 Time: 0.006176 | |
| I0809 10:27:27.785298 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:27.790088 1 logging.cc:52] Tactic: 1002 time 0.008 | |
| I0809 10:27:27.792146 1 logging.cc:52] Tactic: 0 time 0.007808 | |
| I0809 10:27:27.792215 1 logging.cc:52] Fastest Tactic: 0 Time: 0.007808 | |
| I0809 10:27:27.792479 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:27.796935 1 logging.cc:52] Tactic: 1002 time 0.009024 | |
| I0809 10:27:27.799061 1 logging.cc:52] Tactic: 0 time 0.010176 | |
| I0809 10:27:27.799133 1 logging.cc:52] Fastest Tactic: 1002 Time: 0.009024 | |
| I0809 10:27:27.799244 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,100352) -> Float(1,7,49,100352) *************** | |
| I0809 10:27:32.139321 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWise) | |
| I0809 10:27:32.141510 1 logging.cc:52] Tactic: 128 time 0.0064 | |
| I0809 10:27:32.143471 1 logging.cc:52] Tactic: 256 time 0.006688 | |
| I0809 10:27:32.145345 1 logging.cc:52] Tactic: 512 time 0.00752 | |
| I0809 10:27:32.147053 1 logging.cc:52] Tactic: -32 time 0.028992 | |
| I0809 10:27:32.148745 1 logging.cc:52] Tactic: -64 time 0.017952 | |
| I0809 10:27:32.150343 1 logging.cc:52] Tactic: -128 time 0.011648 | |
| I0809 10:27:32.150373 1 logging.cc:52] Fastest Tactic: 128 Time: 0.0064 | |
| I0809 10:27:32.150389 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWiseV2) | |
| I0809 10:27:32.153267 1 logging.cc:52] Tactic: 0 time 0.00576 | |
| I0809 10:27:32.156080 1 logging.cc:52] Tactic: 1 time 0.004416 | |
| I0809 10:27:32.158870 1 logging.cc:52] Tactic: 2 time 0.00464 | |
| I0809 10:27:32.161526 1 logging.cc:52] Tactic: 3 time 0.005504 | |
| I0809 10:27:32.164765 1 logging.cc:52] Tactic: 4 time 0.005536 | |
| I0809 10:27:32.170355 1 logging.cc:52] Tactic: 5 time 0.004416 | |
| I0809 10:27:32.176042 1 logging.cc:52] Tactic: 6 time 0.006144 | |
| I0809 10:27:32.181321 1 logging.cc:52] Tactic: 7 time 0.005888 | |
| I0809 10:27:32.185430 1 logging.cc:52] Tactic: 8 time 0.005408 | |
| I0809 10:27:32.189582 1 logging.cc:52] Tactic: 9 time 0.006016 | |
| I0809 10:27:32.193704 1 logging.cc:52] Tactic: 28 time 0.005824 | |
| I0809 10:27:32.193777 1 logging.cc:52] Fastest Tactic: 1 Time: 0.004416 | |
| I0809 10:27:32.193852 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 1 | |
| I0809 10:27:32.196038 1 logging.cc:52] | |
| I0809 10:27:32.222117 1 logging.cc:52] *************** Autotuning format combination: Float(2048,14336,1,100352) -> Float(2048,14336,1,100352) *************** | |
| I0809 10:27:32.222581 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWise) | |
| I0809 10:27:32.226775 1 logging.cc:52] Tactic: 128 time 0.008 | |
| I0809 10:27:32.228689 1 logging.cc:52] Tactic: 256 time 0.007808 | |
| I0809 10:27:32.230588 1 logging.cc:52] Tactic: 512 time 0.006464 | |
| I0809 10:27:32.232366 1 logging.cc:52] Tactic: -32 time 0.028992 | |
| I0809 10:27:32.234343 1 logging.cc:52] Tactic: -64 time 0.01776 | |
| I0809 10:27:32.236060 1 logging.cc:52] Tactic: -128 time 0.011808 | |
| I0809 10:27:32.236129 1 logging.cc:52] Fastest Tactic: 512 Time: 0.006464 | |
| I0809 10:27:32.236151 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWiseV2) | |
| I0809 10:27:32.236162 1 logging.cc:52] PointWiseV2 has no valid tactics for this config, skipping | |
| I0809 10:27:32.236172 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWise Tactic: 512 | |
| I0809 10:27:32.236182 1 logging.cc:52] | |
| I0809 10:27:32.236371 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49:32,3136) -> Float(1,7,49:32,3136) *************** | |
| I0809 10:27:34.440762 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWise) | |
| I0809 10:27:34.442789 1 logging.cc:52] Tactic: 128 time 0.006368 | |
| I0809 10:27:34.444669 1 logging.cc:52] Tactic: 256 time 0.006368 | |
| I0809 10:27:34.447099 1 logging.cc:52] Tactic: 512 time 0.007872 | |
| I0809 10:27:34.449221 1 logging.cc:52] Tactic: -32 time 0.02896 | |
| I0809 10:27:34.451487 1 logging.cc:52] Tactic: -64 time 0.017856 | |
| I0809 10:27:34.453305 1 logging.cc:52] Tactic: -128 time 0.011584 | |
| I0809 10:27:34.453364 1 logging.cc:52] Fastest Tactic: 128 Time: 0.006368 | |
| I0809 10:27:34.453393 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) (PointWiseV2) | |
| I0809 10:27:34.457952 1 logging.cc:52] Tactic: 24 time 0.005664 | |
| I0809 10:27:34.462510 1 logging.cc:52] Tactic: 25 time 0.006112 | |
| I0809 10:27:34.466684 1 logging.cc:52] Tactic: 26 time 0.006048 | |
| I0809 10:27:34.471045 1 logging.cc:52] Tactic: 27 time 0.006528 | |
| I0809 10:27:34.474919 1 logging.cc:52] Tactic: 31 time 0.005088 | |
| I0809 10:27:34.474975 1 logging.cc:52] Fastest Tactic: 31 Time: 0.005088 | |
| I0809 10:27:34.475022 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 31 | |
| I0809 10:27:34.477137 1 logging.cc:52] | |
| I0809 10:27:34.490714 1 logging.cc:52] *************** Autotuning format combination: -> Int32(1) *************** | |
| I0809 10:27:34.491035 1 logging.cc:52] --------------- Timing Runner: [HostToDeviceCopy] (ShapeHostToDevice) | |
| I0809 10:27:34.491089 1 logging.cc:52] Tactic: 0 is the only option, timing skipped | |
| I0809 10:27:34.491121 1 logging.cc:52] Fastest Tactic: 0 Time: 0 | |
| I0809 10:27:34.492993 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.496711 1 logging.cc:52] Tactic: 1002 time 0.007904 | |
| I0809 10:27:34.498181 1 logging.cc:52] Tactic: 0 time 0.007968 | |
| I0809 10:27:34.498236 1 logging.cc:52] Fastest Tactic: 1002 Time: 0.007904 | |
| I0809 10:27:34.498445 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.502225 1 logging.cc:52] Tactic: 1002 time 0.01664 | |
| I0809 10:27:34.503746 1 logging.cc:52] Tactic: 0 time 0.007776 | |
| I0809 10:27:34.503803 1 logging.cc:52] Fastest Tactic: 0 Time: 0.007776 | |
| I0809 10:27:34.503899 1 logging.cc:52] *************** Autotuning format combination: Float(1,7,49,100352) -> Float(1,1,1,2048) *************** | |
| I0809 10:27:34.504486 1 logging.cc:52] --------------- Timing Runner: GlobalAveragePool_121 (Pooling) | |
| I0809 10:27:34.508084 1 logging.cc:52] Tactic: -1 time 0.008032 | |
| I0809 10:27:34.508144 1 logging.cc:52] Fastest Tactic: -1 Time: 0.008032 | |
| I0809 10:27:34.508212 1 logging.cc:52] --------------- Timing Runner: GlobalAveragePool_121 (TiledPooling) | |
| I0809 10:27:34.513779 1 logging.cc:52] Tactic: 8192257 time 0.00768 | |
| I0809 10:27:34.519648 1 logging.cc:52] Tactic: 8257793 time 0.008192 | |
| I0809 10:27:34.525711 1 logging.cc:52] Tactic: 8323329 time 0.007808 | |
| I0809 10:27:34.531543 1 logging.cc:52] Tactic: 8388865 time 0.007936 | |
| I0809 10:27:34.537607 1 logging.cc:52] Tactic: 8454401 time 0.007904 | |
| I0809 10:27:34.543608 1 logging.cc:52] Tactic: 8519937 time 0.00816 | |
| I0809 10:27:34.549691 1 logging.cc:52] Tactic: 8585473 time 0.008192 | |
| I0809 10:27:34.555618 1 logging.cc:52] Tactic: 8651009 time 0.007936 | |
| I0809 10:27:34.555720 1 logging.cc:52] Fastest Tactic: 8192257 Time: 0.00768 | |
| I0809 10:27:34.555742 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: TiledPooling Tactic: 8192257 | |
| I0809 10:27:34.555752 1 logging.cc:52] | |
| I0809 10:27:34.556392 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.560354 1 logging.cc:52] Tactic: 1002 time 0.006304 | |
| I0809 10:27:34.562912 1 logging.cc:52] Tactic: 0 time 0.004576 | |
| I0809 10:27:34.562982 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004576 | |
| I0809 10:27:34.563287 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.567563 1 logging.cc:52] Tactic: 1002 time 0.008192 | |
| I0809 10:27:34.569659 1 logging.cc:52] Tactic: 0 time 0.004288 | |
| I0809 10:27:34.569726 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004288 | |
| I0809 10:27:34.570026 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.574249 1 logging.cc:52] Tactic: 1002 time 0.00608 | |
| I0809 10:27:34.576366 1 logging.cc:52] Tactic: 0 time 0.004512 | |
| I0809 10:27:34.576440 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004512 | |
| I0809 10:27:34.576687 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.580820 1 logging.cc:52] Tactic: 1002 time 0.00816 | |
| I0809 10:27:34.582845 1 logging.cc:52] Tactic: 0 time 0.005568 | |
| I0809 10:27:34.582918 1 logging.cc:52] Fastest Tactic: 0 Time: 0.005568 | |
| I0809 10:27:34.583178 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.587427 1 logging.cc:52] Tactic: 1002 time 0.006144 | |
| I0809 10:27:34.589452 1 logging.cc:52] Tactic: 0 time 0.005408 | |
| I0809 10:27:34.589522 1 logging.cc:52] Fastest Tactic: 0 Time: 0.005408 | |
| I0809 10:27:34.589818 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.594195 1 logging.cc:52] Tactic: 1002 time 0.008128 | |
| I0809 10:27:34.596788 1 logging.cc:52] Tactic: 0 time 0.004512 | |
| I0809 10:27:34.596887 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004512 | |
| I0809 10:27:34.597244 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.606316 1 logging.cc:52] Tactic: 1002 time 0.008192 | |
| I0809 10:27:34.608554 1 logging.cc:52] Tactic: 0 time 0.004384 | |
| I0809 10:27:34.608644 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004384 | |
| I0809 10:27:34.609102 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:34.613442 1 logging.cc:52] Tactic: 1002 time 0.008192 | |
| I0809 10:27:34.615762 1 logging.cc:52] Tactic: 0 time 0.005312 | |
| I0809 10:27:34.615848 1 logging.cc:52] Fastest Tactic: 0 Time: 0.005312 | |
| I0809 10:27:34.616081 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1,2048) -> Float(1,1,1,2048) *************** | |
| I0809 10:27:34.616450 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.620692 1 logging.cc:52] Tactic: 1002 time 0.006208 | |
| I0809 10:27:34.622797 1 logging.cc:52] Tactic: 0 time 0.004288 | |
| I0809 10:27:34.622876 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004288 | |
| I0809 10:27:34.622900 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.622912 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.622923 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.622934 1 logging.cc:52] | |
| I0809 10:27:34.623072 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1,2048) -> Float(2048,2048,1,2048) *************** | |
| I0809 10:27:34.623236 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.627400 1 logging.cc:52] Tactic: 1002 time 0.006144 | |
| I0809 10:27:34.629467 1 logging.cc:52] Tactic: 0 time 0.004288 | |
| I0809 10:27:34.629573 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004288 | |
| I0809 10:27:34.629599 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.629634 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.629646 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.629656 1 logging.cc:52] | |
| I0809 10:27:34.629791 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1,2048) -> Float(1,1,1:32,64) *************** | |
| I0809 10:27:34.629989 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.633941 1 logging.cc:52] Tactic: 1002 time 0.00816 | |
| I0809 10:27:34.635799 1 logging.cc:52] Tactic: 0 time 0.005312 | |
| I0809 10:27:34.635926 1 logging.cc:52] Fastest Tactic: 0 Time: 0.005312 | |
| I0809 10:27:34.635958 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.635969 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.635980 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.635990 1 logging.cc:52] | |
| I0809 10:27:34.636146 1 logging.cc:52] *************** Autotuning format combination: Float(2048,2048,1,2048) -> Float(1,1,1,2048) *************** | |
| I0809 10:27:34.636425 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.640332 1 logging.cc:52] Tactic: 1002 time 0.006336 | |
| I0809 10:27:34.642333 1 logging.cc:52] Tactic: 0 time 0.00432 | |
| I0809 10:27:34.642400 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00432 | |
| I0809 10:27:34.642423 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.642434 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.642445 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.642454 1 logging.cc:52] | |
| I0809 10:27:34.642581 1 logging.cc:52] *************** Autotuning format combination: Float(2048,2048,1,2048) -> Float(2048,2048,1,2048) *************** | |
| I0809 10:27:34.642762 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.646550 1 logging.cc:52] Tactic: 1002 time 0.005728 | |
| I0809 10:27:34.648419 1 logging.cc:52] Tactic: 0 time 0.00432 | |
| I0809 10:27:34.648502 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00432 | |
| I0809 10:27:34.648540 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.648559 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.648573 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.648584 1 logging.cc:52] | |
| I0809 10:27:34.648718 1 logging.cc:52] *************** Autotuning format combination: Float(2048,2048,1,2048) -> Float(1,1,1:32,64) *************** | |
| I0809 10:27:34.648914 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.652785 1 logging.cc:52] Tactic: 1002 time 0.00816 | |
| I0809 10:27:34.654829 1 logging.cc:52] Tactic: 0 time 0.00432 | |
| I0809 10:27:34.654931 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00432 | |
| I0809 10:27:34.654971 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.654991 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.655019 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.655030 1 logging.cc:52] | |
| I0809 10:27:34.655175 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1:32,64) -> Float(1,1,1,2048) *************** | |
| I0809 10:27:34.655437 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.659153 1 logging.cc:52] Tactic: 1002 time 0.00816 | |
| I0809 10:27:34.660932 1 logging.cc:52] Tactic: 0 time 0.00432 | |
| I0809 10:27:34.661009 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00432 | |
| I0809 10:27:34.661033 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.661044 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.661055 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.661064 1 logging.cc:52] | |
| I0809 10:27:34.661190 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1:32,64) -> Float(2048,2048,1,2048) *************** | |
| I0809 10:27:34.661449 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.665472 1 logging.cc:52] Tactic: 1002 time 0.00816 | |
| I0809 10:27:34.667429 1 logging.cc:52] Tactic: 0 time 0.005344 | |
| I0809 10:27:34.667503 1 logging.cc:52] Fastest Tactic: 0 Time: 0.005344 | |
| I0809 10:27:34.667526 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.667553 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.667565 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.667587 1 logging.cc:52] | |
| I0809 10:27:34.667754 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1:32,64) -> Float(1,1,1:32,64) *************** | |
| I0809 10:27:34.667945 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Reformat) | |
| I0809 10:27:34.672273 1 logging.cc:52] Tactic: 1002 time 0.007904 | |
| I0809 10:27:34.674474 1 logging.cc:52] Tactic: 0 time 0.00432 | |
| I0809 10:27:34.674548 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00432 | |
| I0809 10:27:34.674572 1 logging.cc:52] --------------- Timing Runner: Cast_122 (Cast) | |
| I0809 10:27:34.674584 1 logging.cc:52] Cast has no valid tactics for this config, skipping | |
| I0809 10:27:34.674595 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: Reformat Tactic: 0 | |
| I0809 10:27:34.674605 1 logging.cc:52] | |
| I0809 10:27:34.686691 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1,2048) -> Float(1,1,1,2048) *************** | |
| I0809 10:27:38.945405 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWise) | |
| I0809 10:27:38.947985 1 logging.cc:52] Tactic: 128 time 0.005088 | |
| I0809 10:27:38.949973 1 logging.cc:52] Tactic: 256 time 0.00576 | |
| I0809 10:27:38.951638 1 logging.cc:52] Tactic: 512 time 0.00576 | |
| I0809 10:27:38.953420 1 logging.cc:52] Tactic: -32 time 0.028992 | |
| I0809 10:27:38.955422 1 logging.cc:52] Tactic: -64 time 0.01664 | |
| I0809 10:27:38.957078 1 logging.cc:52] Tactic: -128 time 0.010496 | |
| I0809 10:27:38.957116 1 logging.cc:52] Fastest Tactic: 128 Time: 0.005088 | |
| I0809 10:27:38.957134 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWiseV2) | |
| I0809 10:27:38.959915 1 logging.cc:52] Tactic: 0 time 0.004352 | |
| I0809 10:27:38.962548 1 logging.cc:52] Tactic: 1 time 0.004352 | |
| I0809 10:27:38.965269 1 logging.cc:52] Tactic: 2 time 0.004352 | |
| I0809 10:27:38.967869 1 logging.cc:52] Tactic: 3 time 0.005568 | |
| I0809 10:27:38.970104 1 logging.cc:52] Tactic: 4 time 0.004416 | |
| I0809 10:27:38.972568 1 logging.cc:52] Tactic: 5 time 0.004352 | |
| I0809 10:27:38.975257 1 logging.cc:52] Tactic: 6 time 0.006144 | |
| I0809 10:27:38.978330 1 logging.cc:52] Tactic: 7 time 0.005728 | |
| I0809 10:27:38.980932 1 logging.cc:52] Tactic: 8 time 0.004576 | |
| I0809 10:27:38.983735 1 logging.cc:52] Tactic: 9 time 0.004448 | |
| I0809 10:27:38.986320 1 logging.cc:52] Tactic: 28 time 0.004352 | |
| I0809 10:27:38.986354 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004352 | |
| I0809 10:27:38.986373 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 0 | |
| I0809 10:27:38.987276 1 logging.cc:52] | |
| I0809 10:27:38.998928 1 logging.cc:52] *************** Autotuning format combination: Float(2048,2048,1,2048) -> Float(2048,2048,1,2048) *************** | |
| I0809 10:27:38.999165 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWise) | |
| I0809 10:27:39.003202 1 logging.cc:52] Tactic: 128 time 0.005568 | |
| I0809 10:27:39.004760 1 logging.cc:52] Tactic: 256 time 0.005792 | |
| I0809 10:27:39.006218 1 logging.cc:52] Tactic: 512 time 0.006144 | |
| I0809 10:27:39.007758 1 logging.cc:52] Tactic: -32 time 0.028928 | |
| I0809 10:27:39.009334 1 logging.cc:52] Tactic: -64 time 0.01664 | |
| I0809 10:27:39.011217 1 logging.cc:52] Tactic: -128 time 0.010816 | |
| I0809 10:27:39.011287 1 logging.cc:52] Fastest Tactic: 128 Time: 0.005568 | |
| I0809 10:27:39.011308 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWiseV2) | |
| I0809 10:27:39.011320 1 logging.cc:52] PointWiseV2 has no valid tactics for this config, skipping | |
| I0809 10:27:39.011330 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWise Tactic: 128 | |
| I0809 10:27:39.011339 1 logging.cc:52] | |
| I0809 10:27:39.011609 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1:32,64) -> Float(1,1,1:32,64) *************** | |
| I0809 10:27:41.431430 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWise) | |
| I0809 10:27:41.433810 1 logging.cc:52] Tactic: 128 time 0.005696 | |
| I0809 10:27:41.435484 1 logging.cc:52] Tactic: 256 time 0.00592 | |
| I0809 10:27:41.437262 1 logging.cc:52] Tactic: 512 time 0.00592 | |
| I0809 10:27:41.438918 1 logging.cc:52] Tactic: -32 time 0.028864 | |
| I0809 10:27:41.440789 1 logging.cc:52] Tactic: -64 time 0.017952 | |
| I0809 10:27:41.442785 1 logging.cc:52] Tactic: -128 time 0.010496 | |
| I0809 10:27:41.442834 1 logging.cc:52] Fastest Tactic: 128 Time: 0.005696 | |
| I0809 10:27:41.442850 1 logging.cc:52] --------------- Timing Runner: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) (PointWiseV2) | |
| I0809 10:27:41.445948 1 logging.cc:52] Tactic: 24 time 0.004352 | |
| I0809 10:27:41.448997 1 logging.cc:52] Tactic: 25 time 0.005504 | |
| I0809 10:27:41.452426 1 logging.cc:52] Tactic: 26 time 0.006112 | |
| I0809 10:27:41.455547 1 logging.cc:52] Tactic: 27 time 0.0064 | |
| I0809 10:27:41.458799 1 logging.cc:52] Tactic: 31 time 0.00448 | |
| I0809 10:27:41.458845 1 logging.cc:52] Fastest Tactic: 24 Time: 0.004352 | |
| I0809 10:27:41.458866 1 logging.cc:52] >>>>>>>>>>>>>>> Chose Runner Type: PointWiseV2 Tactic: 24 | |
| I0809 10:27:41.460057 1 logging.cc:52] | |
| I0809 10:27:41.467533 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:41.471511 1 logging.cc:52] Tactic: 1002 time 0.006272 | |
| I0809 10:27:41.473043 1 logging.cc:52] Tactic: 0 time 0.00544 | |
| I0809 10:27:41.473111 1 logging.cc:52] Fastest Tactic: 0 Time: 0.00544 | |
| I0809 10:27:41.473447 1 logging.cc:52] --------------- Timing Runner: <reformat> (Reformat) | |
| I0809 10:27:41.481280 1 logging.cc:52] Tactic: 1002 time 0.008128 | |
| I0809 10:27:41.483202 1 logging.cc:52] Tactic: 0 time 0.004288 | |
| I0809 10:27:41.483273 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004288 | |
| I0809 10:27:41.483531 1 logging.cc:52] *************** Autotuning format combination: Float(1,1,1,2048) -> Float(1,2048) *************** | |
| I0809 10:27:41.483897 1 logging.cc:52] --------------- Timing Runner: Reshape_133 (Shuffle) | |
| I0809 10:27:41.487876 1 logging.cc:52] Tactic: 0 time 0.004288 | |
| I0809 10:27:41.489739 1 logging.cc:52] Tactic: 1 time 0.009536 | |
| I0809 10:27:41.489841 1 logging.cc:52] Fastest Tactic: 0 Time: 0.004288 | |
| I0809 10:27:41.490037 1 logging.cc:52] *************** Autotuning format combination: -> Float(1,256) *************** | |
| I0809 10:27:41.490088 1 logging.cc:52] *************** Autotuning format combination: Float(1,2048), Float(1,256) -> Float(1,256) *************** | |
| I0809 10:27:41.492204 1 logging.cc:52] --------------- Timing Runner: Gemm_134 (MatrixMultiply) | |
| I0809 10:27:41.492294 1 logging.cc:52] Tactic: 0 is the only option, timing skipped | |
| I0809 10:27:41.492328 1 logging.cc:52] Fastest Tactic: 0 Time: 0 | |
| I0809 10:27:41.494688 1 logging.cc:52] *************** Autotuning format combination: -> Float(1,256) *************** | |
| I0809 10:27:41.494790 1 logging.cc:52] *************** Autotuning format combination: Float(1,256), Float(1,256) -> Float(1,256) *************** | |
| I0809 10:27:41.495274 1 logging.cc:52] --------------- Timing Runner: (Unnamed Layer* 13) [ElementWise] (ElementWise) | |
| I0809 10:27:41.499350 1 logging.cc:52] Tactic: 1 time 0.004352 | |
| I0809 10:27:41.501300 1 logging.cc:52] Tactic: 2 time 0.0064 | |
| I0809 10:27:41.501405 1 logging.cc:52] Fastest Tactic: 1 Time: 0.004352 | |
| I0809 10:27:41.501604 1 logging.cc:52] *************** Autotuning format combination: Float(1,256) -> Float(1,1) *************** | |
| I0809 10:27:41.501831 1 logging.cc:52] --------------- Timing Runner: (Unnamed Layer* 15) [ElementWise] + ReduceL2_135 + ReduceL2_135_8 (Reduce) | |
| I0809 10:27:41.505779 1 logging.cc:52] Tactic: 0 time 0.006144 | |
| I0809 10:27:41.507446 1 logging.cc:52] Tactic: 1 time 0.006144 | |
| I0809 10:27:41.509295 1 logging.cc:52] Tactic: 3 time 0.009792 | |
| I0809 10:27:41.512281 1 logging.cc:52] Tactic: 6 time 0.138496 | |
| I0809 10:27:41.512385 1 logging.cc:52] Fastest Tactic: 0 Time: 0.006144 | |
| I0809 10:27:41.513002 1 logging.cc:52] Adding reformat layer: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) reformatted input 0 (497) from Half(1,7,49,100352) to Float(1,7,49,100352) | |
| I0809 10:27:41.519032 1 logging.cc:52] Formats and tactics selection completed in 13.743 seconds. | |
| I0809 10:27:41.519115 1 logging.cc:52] After reformat layers: 12 layers | |
| I0809 10:27:41.519204 1 logging.cc:52] Block size 1073741824 | |
| I0809 10:27:41.519227 1 logging.cc:52] Block size 401408 | |
| I0809 10:27:41.519243 1 logging.cc:52] Block size 401408 | |
| I0809 10:27:41.519261 1 logging.cc:52] Block size 1 | |
| I0809 10:27:41.519313 1 logging.cc:52] Total Activation Memory: 1074544641 | |
| I0809 10:27:41.519442 1 logging.cc:49] Detected 1 inputs and 4 output network tensors. | |
| I0809 10:27:41.535992 1 logging.cc:52] Layer: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) input reformatter 0 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536083 1 logging.cc:52] Layer: PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) Weights: 0 HostPersistent: 276 DevicePersistent: 0 | |
| I0809 10:27:41.536108 1 logging.cc:52] Layer: [HostToDeviceCopy] Weights: 0 HostPersistent: 16 DevicePersistent: 0 | |
| I0809 10:27:41.536129 1 logging.cc:52] Layer: GlobalAveragePool_121 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536151 1 logging.cc:52] Layer: PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123) Weights: 0 HostPersistent: 276 DevicePersistent: 0 | |
| I0809 10:27:41.536170 1 logging.cc:52] Layer: Reshape_133 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536195 1 logging.cc:52] Layer: 693 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536217 1 logging.cc:52] Layer: Gemm_134 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536257 1 logging.cc:52] Layer: (Unnamed Layer* 11) [Constant] + (Unnamed Layer* 12) [Shuffle] Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536310 1 logging.cc:52] Layer: (Unnamed Layer* 13) [ElementWise] Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536333 1 logging.cc:52] Layer: (Unnamed Layer* 15) [ElementWise] + ReduceL2_135 + ReduceL2_135_8 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.536375 1 logging.cc:52] Total Host Persistent Memory: 568 | |
| I0809 10:27:41.536393 1 logging.cc:52] Total Device Persistent Memory: 0 | |
| I0809 10:27:41.536409 1 logging.cc:52] Total Weight Memory: 0 | |
| I0809 10:27:41.549649 1 logging.cc:52] Engine generation completed in 13.7808 seconds. | |
| I0809 10:27:41.549711 1 logging.cc:52] Builder timing cache: created 22 entries, 6 hit(s) | |
| I0809 10:27:41.551120 1 logging.cc:52] Engine Layer Information: | |
| I0809 10:27:41.551186 1 logging.cc:52] Layer(Reformat): PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) input reformatter 0, Tactic: 1002, 497[Half(2048,7,7)] -> PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) reformatted input 0[Float(2048,7,7)] | |
| I0809 10:27:41.551205 1 logging.cc:52] Layer(PointWiseV2): PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120), Tactic: 1, PWN((Unnamed Layer* 0) [Constant] + (Unnamed Layer* 1) [Shuffle], Pow_120) reformatted input 0[Float(2048,7,7)] -> 498[Float(2048,7,7)] | |
| I0809 10:27:41.551220 1 logging.cc:52] Layer(ShapeHostToDevice): [HostToDeviceCopy], Tactic: 0, -> 526[Int32()] | |
| I0809 10:27:41.551236 1 logging.cc:52] Layer(PoolingTiled): GlobalAveragePool_121, Tactic: 8192257, 498[Float(2048,7,7)] -> 499[Float(2048,1,1)] | |
| I0809 10:27:41.551251 1 logging.cc:52] Layer(PointWiseV2): PWN((Unnamed Layer* 5) [Constant] + (Unnamed Layer* 6) [Shuffle], Pow_123), Tactic: 0, 505[Float(2048,1,1)] -> OUTPUT__1[Float(2048,1,1)] | |
| I0809 10:27:41.551265 1 logging.cc:52] Layer(Shuffle): Reshape_133, Tactic: 0, OUTPUT__1[Float(2048,1,1)] -> 516[Float(2048)] | |
| I0809 10:27:41.551295 1 logging.cc:52] Layer(Constant): 693, Tactic: 0, -> (Unnamed Layer* 9) [Constant]_output[Float(256)] | |
| I0809 10:27:41.551312 1 logging.cc:52] Layer(MatrixMultiply): Gemm_134, Tactic: 0, 516[Float(2048)], (Unnamed Layer* 9) [Constant]_output[Float(256)] -> (Unnamed Layer* 10) [Matrix Multiply]_output[Float(256)] | |
| I0809 10:27:41.551325 1 logging.cc:52] Layer(Constant): (Unnamed Layer* 11) [Constant] + (Unnamed Layer* 12) [Shuffle], Tactic: 0, -> (Unnamed Layer* 12) [Shuffle]_output[Float(256)] | |
| I0809 10:27:41.551388 1 logging.cc:52] Layer(ElementWise): (Unnamed Layer* 13) [ElementWise], Tactic: 1, (Unnamed Layer* 10) [Matrix Multiply]_output[Float(256)], (Unnamed Layer* 12) [Shuffle]_output[Float(256)] -> 520[Float(256)] | |
| I0809 10:27:41.551423 1 logging.cc:52] Layer(Reduce): (Unnamed Layer* 15) [ElementWise] + ReduceL2_135 + ReduceL2_135_8, Tactic: 0, 520[Float(256)] -> 521[Float(1)] | |
| I0809 10:27:41.555075 1 logging.cc:52] Allocated persistent device memory of size 0 | |
| I0809 10:27:41.555170 1 logging.cc:52] Allocated activation device memory of size 803328 | |
| I0809 10:27:41.555213 1 logging.cc:52] Assigning persistent memory blocks for various profiles | |
| I0809 10:27:41.560832 1 logging.cc:52] Applying generic optimizations to the graph for inference. | |
| I0809 10:27:41.560898 1 logging.cc:52] Original: 2 layers | |
| I0809 10:27:41.560924 1 logging.cc:52] After dead-layer removal: 2 layers | |
| I0809 10:27:41.560982 1 logging.cc:52] After Myelin optimization: 2 layers | |
| I0809 10:27:41.561047 1 logging.cc:52] After scale fusion: 2 layers | |
| I0809 10:27:41.563665 1 logging.cc:52] After vertical fusions: 2 layers | |
| I0809 10:27:41.563739 1 logging.cc:52] After dupe layer removal: 2 layers | |
| I0809 10:27:41.563761 1 logging.cc:52] After final dead-layer removal: 2 layers | |
| I0809 10:27:41.563807 1 logging.cc:52] After tensor merging: 2 layers | |
| I0809 10:27:41.563837 1 logging.cc:52] After concat removal: 2 layers | |
| I0809 10:27:41.563874 1 logging.cc:52] Graph construction and optimization completed in 0.00347335 seconds. | |
| I0809 10:27:41.571539 1 logging.cc:52] Constructing optimization profile number 0 [1/1]. | |
| I0809 10:27:41.571767 1 logging.cc:52] *************** Autotuning format combination: Float(1,1) -> Float(1,256) *************** | |
| I0809 10:27:41.572039 1 logging.cc:52] --------------- Timing Runner: Expand_138 (Slice) | |
| I0809 10:27:41.572105 1 logging.cc:52] Tactic: 0 is the only option, timing skipped | |
| I0809 10:27:41.572142 1 logging.cc:52] Fastest Tactic: 0 Time: 0 | |
| I0809 10:27:41.573879 1 logging.cc:52] *************** Autotuning format combination: Float(1,256), Float(1,256) -> Float(1,256) *************** | |
| I0809 10:27:41.574086 1 logging.cc:52] --------------- Timing Runner: Div_139 (ElementWise) | |
| I0809 10:27:41.578386 1 logging.cc:52] Tactic: 1 time 0.005504 | |
| I0809 10:27:41.581559 1 logging.cc:52] Tactic: 2 time 0.00832 | |
| I0809 10:27:41.581632 1 logging.cc:52] Fastest Tactic: 1 Time: 0.005504 | |
| I0809 10:27:41.585001 1 logging.cc:52] Formats and tactics selection completed in 0.0134567 seconds. | |
| I0809 10:27:41.585058 1 logging.cc:52] After reformat layers: 2 layers | |
| I0809 10:27:41.585086 1 logging.cc:52] Block size 1073741824 | |
| I0809 10:27:41.585097 1 logging.cc:52] Block size 1024 | |
| I0809 10:27:41.585107 1 logging.cc:52] Total Activation Memory: 1073742848 | |
| I0809 10:27:41.585131 1 logging.cc:49] Detected 2 inputs and 1 output network tensors. | |
| I0809 10:27:41.585204 1 logging.cc:52] Layer: Expand_138 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.585228 1 logging.cc:52] Layer: Div_139 Weights: 0 HostPersistent: 0 DevicePersistent: 0 | |
| I0809 10:27:41.585241 1 logging.cc:52] Total Host Persistent Memory: 0 | |
| I0809 10:27:41.585252 1 logging.cc:52] Total Device Persistent Memory: 0 | |
| I0809 10:27:41.585262 1 logging.cc:52] Total Weight Memory: 0 | |
| I0809 10:27:41.603076 1 logging.cc:52] Engine generation completed in 0.0391664 seconds. | |
| I0809 10:27:41.603138 1 logging.cc:52] Builder timing cache: created 0 entries, 0 hit(s) | |
| I0809 10:27:41.605152 1 logging.cc:52] Engine Layer Information: | |
| I0809 10:27:41.605218 1 logging.cc:52] Layer(Slice): Expand_138, Tactic: 0, 525[Float(1)] -> 527[Float(256)] | |
| I0809 10:27:41.605236 1 logging.cc:52] Layer(ElementWise): Div_139, Tactic: 1, 520[Float(256)], 527[Float(256)] -> OUTPUT__0[Float(256)] | |
| I0809 10:27:41.610231 1 logging.cc:52] Allocated persistent device memory of size 0 | |
| I0809 10:27:41.610333 1 logging.cc:52] Allocated activation device memory of size 1024 | |
| I0809 10:27:41.610367 1 logging.cc:52] Assigning persistent memory blocks for various profiles | |
| 2021-08-09 10:27:41.610851661 [I:onnxruntime:, sequential_executor.cc:474 Execute] [Memory] ExecutionFrame dynamically allocates 64 bytes for Cuda | |
| 2021-08-09 10:27:41.610927199 [I:onnxruntime:, sequential_executor.cc:474 Execute] [Memory] ExecutionFrame dynamically allocates 200704 bytes for TensorrtPinned | |
| 2021-08-09 10:27:41.610951844 [I:onnxruntime:, sequential_executor.cc:474 Execute] [Memory] ExecutionFrame dynamically allocates 411776 bytes for Tensorrt | |
| 2021-08-09 10:27:41.610972502 [I:onnxruntime:, sequential_executor.cc:474 Execute] [Memory] ExecutionFrame dynamically allocates 602176 bytes for Cpu | |
| I0809 10:27:41.613130 1 infer_response.cc:165] add response output: output: OUTPUT__0, type: FP32, shape: [1,256] | |
| I0809 10:27:41.613276 1 grpc_server.cc:2230] GRPC: using buffer for 'OUTPUT__0', size: 1024, addr: 0x7f41c5a4aa70 | |
| I0809 10:27:41.613317 1 infer_response.cc:165] add response output: output: OUTPUT__1, type: FP32, shape: [1,2048,1,1] | |
| I0809 10:27:41.613345 1 grpc_server.cc:2230] GRPC: using buffer for 'OUTPUT__1', size: 8192, addr: 0x7f41c4dfdeb0 | |
| I0809 10:27:41.613364 1 infer_response.cc:165] add response output: output: OUTPUT__2, type: FP16, shape: [1,2048,7,7] | |
| I0809 10:27:41.613442 1 grpc_server.cc:2230] GRPC: using buffer for 'OUTPUT__2', size: 200704, addr: 0x7f41c5675af0 | |
| I0809 10:27:41.613529 1 grpc_server.cc:3240] ModelInferHandler::InferResponseComplete, 4 step ISSUED | |
| I0809 10:27:41.613623 1 grpc_server.cc:2265] GRPC free: size 1024, addr 0x7f41c5a4aa70 | |
| I0809 10:27:41.613647 1 grpc_server.cc:2265] GRPC free: size 8192, addr 0x7f41c4dfdeb0 | |
| I0809 10:27:41.613656 1 grpc_server.cc:2265] GRPC free: size 200704, addr 0x7f41c5675af0 | |
| I0809 10:27:41.614771 1 grpc_server.cc:2817] ModelInferHandler::InferRequestComplete | |
| I0809 10:27:41.614862 1 grpc_server.cc:3089] Process for ModelInferHandler, rpc_ok=1, 4 step COMPLETEI0809 10:27:41.614906 1 pinned_memory_manager.cc:158] pinned memory deallocation: addr 0x7f44b6000090 | |
| I0809 10:27:41.614938 1 grpc_server.cc:2139] Done for ModelInferHandler, 4 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment