Uploaded image for project: 'Apache MXNet (Retired)'
  1. Apache MXNet (Retired)
  2. MXNET-938

RNN.hybrid_forward() spends a significant time on _rnn_param_concat

    XMLWordPrintableJSON

Details

    • Improvement
    • Status: To Do
    • Major
    • Resolution: Unresolved
    • Gluon
    • None

    Description

      Issue :
      RNN.hybrid_forward() spends a significant time on _rnn_param_concat

      Root Cause:
      Creating a simple LSTM (256 -> 256) with gluon (gluon_lstm.py) and running inference on CPU with the profile shows the following output.  { Refer gluon_lstm.py for the code}

      _rnn_param_concat() takes more time than the RNN operator.

       

                         
      ------------------- ------- -------- -----------
      ------------
      ------------
      ------------
      operator                  
      =================                  
      Name Total Count Time (ms) Min Time (ms) Max Time (ms) Avg Time (ms)


      ------ --------- ---------- ---------- ----------
      RNN   72 72.312   0.065   33.15   1.0043
      _zeros   788 1.76   0.001   0.011   0.0022
      DeleteVariable   1110 0.637   0   0.002   0.0006
      SwapAxis   462 2.223   0.003   0.015   0.0048
      _rnn_param_concat   76 103.813   1.122   2.08   1.366

      Attachments

        1. gluon_lstm.py
          1 kB
          Gautham Kollu

        Issue Links

          Activity

            People

              Unassigned Unassigned
              kollg Gautham Kollu
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated:

                Time Tracking

                  Estimated:
                  Original Estimate - Not Specified
                  Not Specified
                  Remaining:
                  Remaining Estimate - 0h
                  0h
                  Logged:
                  Time Spent - 1h 50m
                  1h 50m