TensorRT Profiler reports unnamed layers


I use TensorRT4.0RC and see that profiler reports several unnamed layers taking huge amount of time to execute:

leaky_re_lu_XX/LeakyRelu/Maximum                          , 0.0512ms
(Unnamed Layer* XX)                                       , 29.9571ms
conv2d_transpose_XX/conv2d_transpose                      , 0.1423ms
conv2d_transpose_XX/BiasAdd                               , 0.0870ms
leaky_re_lu_XX/LeakyRelu/mul                              , 0.0870ms
leaky_re_lu_XX/LeakyRelu/Maximum                          , 0.1116ms
conv2d_XX/convolution input reformatter 0                 , 0.1628ms
conv2d_XX/convolution                                     , 0.5325ms
conv2d_XX/convolution output reformatter 0                , 0.0717ms
leaky_re_lu_XX/LeakyRelu/mul                              , 0.0860ms
leaky_re_lu_XX/LeakyRelu/Maximum                          , 0.1137ms
(Unnamed Layer* XX)                                       , 127.1060ms
conv2d_transpose_XX/conv2d_transpose                      , 0.3297ms

Is there any way to figure out what is Unnamed layer? Why does it take so long (x1000 than other operations)to execute?

We created a new “Deep Learning Training and Inference” section in Devtalk to improve the experience for deep learning and accelerated computing, and HPC users:

We are moving active deep learning threads to the new section.

URLs for topics will not change with the re-categorization. So your bookmarks and links will continue to work as earlier.


Please file a bug here: https://developer.nvidia.com/nvidia-developer-program
Please include the steps/files used to reproduce the problem along with the output of infer_device.