Type Time (%) Time (ms) Calls       Avg       Min       Max                                               Name
GPU activities:    42.12    4.8836    37  131.99us     896ns  4.8456ms                                 [CUDA memcpy HtoD]
GPU activities:    18.46    2.1401    12  178.34us  85.791us  278.69us  void spatialDepthwiseConvolutionUpdateOutput<f...
GPU activities:    11.34    1.3154    18  73.076us  33.728us  121.57us  void kernelPointwiseApply2<TensorTakeOp<float,...
GPU activities:     8.73    1.0122    30  33.738us  1.5360us  122.94us  void kernelPointwiseApply3<TensorAddOp<long>, ...
GPU activities:     3.66   0.42435    24  17.681us  3.1680us  35.552us  void kernelPointwiseApply2<CopyOp<float, float...
GPU activities:      2.5    0.2895    12  24.125us  3.0720us  66.431us  void CatArrayBatchedCopy<float, unsigned int, ...
GPU activities:     2.41    0.2792    12  23.266us  5.0560us  42.432us  void kernelPointwiseApply3<TensorAddOp<float>,...
GPU activities:     2.35   0.27277    12  22.730us  7.2640us  38.208us  void kernelPointwiseApply3<TensorSubOp<float>,...
GPU activities:     1.88   0.21849    12  18.207us  7.1350us  32.575us  void kernelPointwiseApply2<CopyOp<float, float...
GPU activities:     1.86   0.21625     6  36.042us  13.920us  59.296us  void kernelPointwiseApply2<TensorDivConstantOp...
GPU activities:     1.11   0.12842     4  32.104us  29.760us  34.976us  void kernelPointwiseApply2<CopyOp<float, float...
GPU activities:     0.66  0.076063    72  1.0560us     992ns  1.5680us                                 [CUDA memcpy DtoH]
GPU activities:     0.63  0.073407    18  4.0780us  3.9040us  4.2560us  void kernelReduceAll<long, unsigned int, long,...
GPU activities:     0.59  0.068511    18  3.8060us  3.7120us  3.9040us  void kernelReduceAll<long, unsigned int, long,...
GPU activities:     0.47  0.054656     8  6.8320us  6.3360us  7.3600us  void kernelPointwiseApply2<CopyOp<float, float...
GPU activities:      0.4  0.046048    36  1.2790us     960ns  1.5360us  void kernelPointwiseApply2<TensorMulConstantOp...
GPU activities:     0.29  0.033408    30  1.1130us     800ns  1.8560us  void thrust::cuda_cub::core::_kernel_agent<thr...
GPU activities:     0.29  0.033216    36     922ns     768ns  1.1520us  void kernelPointwiseApply1<TensorFillOp<long>,...
GPU activities:     0.26  0.030112    18  1.6720us  1.6320us  1.8240us  void kernelPointwiseApply2<TensorRemainderOp<l...
         Total:   100.01   11.5957   415                                                                                 
Total (no mem):    57.23   6.63604   306                                                                                 
     API calls:    99.49   5675.93    20  283.80ms  6.6160us  5.66584s                                         cudaMalloc
     API calls:     0.14     8.154   185  44.075us     215ns  5.0181ms                               cuDeviceGetAttribute
     API calls:     0.13    7.3969   109  67.861us  5.4100us  5.9355ms                                    cudaMemcpyAsync
     API calls:     0.07    3.9797     2  1.9899ms  1.9777ms  2.0021ms                            cudaGetDeviceProperties
     API calls:     0.06    3.4974   306  11.429us  6.2050us  82.425us                                         cudaLaunch
     API calls:     0.03    1.5953    97  16.446us  1.5750us  309.44us                              cudaStreamSynchronize
     API calls:     0.02    1.3601  3483     390ns     264ns  14.949us                                      cudaGetDevice
     API calls:     0.02   0.95127     2  475.64us  15.446us  935.83us                                      cudaHostAlloc
     API calls:     0.01   0.47753  1148     415ns     287ns  12.962us                                      cudaSetDevice
     API calls:     0.01   0.40021     2  200.10us  199.96us  200.24us                                   cuDeviceTotalMem
     API calls:     0.01   0.39003     2  195.02us  191.94us  198.09us                                    cuDeviceGetName
     API calls:     0.01    0.3047  1458     208ns     108ns  7.8140us                                  cudaSetupArgument
     API calls:        0   0.11511    30  3.8370us  2.8070us  13.284us                              cudaFuncGetAttributes
     API calls:        0   0.09468   306     309ns     138ns  7.6200us                                  cudaConfigureCall
     API calls:        0   0.08245   348     236ns     126ns  7.2750us                                   cudaGetLastError
     API calls:        0  0.027772    13  2.1360us     926ns  3.4970us                                     cudaEventQuery
     API calls:        0    0.0201    60     335ns     104ns  5.6470us                                cudaPeekAtLastError
     API calls:        0  0.019118    11  1.7380us     667ns  6.8070us                                   cudaEventDestroy
     API calls:        0  0.018143    30     604ns     440ns  2.1690us                             cudaDeviceGetAttribute
     API calls:        0   0.01548    12  1.2900us     859ns  2.1570us                           cudaEventCreateWithFlags
     API calls:        0  0.014105    12  1.1750us     868ns  1.7880us                                    cudaEventRecord
     API calls:        0  0.005924    13     455ns     163ns  1.7660us                                 cudaGetDeviceCount
     API calls:        0  0.002697     4     674ns     208ns  1.7280us                                   cuDeviceGetCount
     API calls:        0   0.00154     3     513ns     246ns     933ns                                        cuDeviceGet
     API calls:        0  0.000816     1     816ns     816ns     816ns                                 cuDriverGetVersion
     API calls:        0   0.00073     1     730ns     730ns     730ns                                             cuInit
         Total:      100   5704.86  7658