Support RNN profiling in StreamExecutor for CUDA GPUs.
authorJames Qin <jamesqin@google.com>
Fri, 6 Apr 2018 18:56:08 +0000 (11:56 -0700)
committerTensorFlower Gardener <gardener@tensorflow.org>
Fri, 6 Apr 2018 19:00:45 +0000 (12:00 -0700)
commit4f7943f7358fc69af62dc280c6f6ba549ebe2167
tree19bdd3dddebeb7d26f9685328d0598bb58347bc0
parentf15c117c4f4d51a6660bf14b6d6cf73c52692cfb
Support RNN profiling in StreamExecutor for CUDA GPUs.

This change hasn't applied autotune on TF Cudnn kernels, only provides lower level support.

PiperOrigin-RevId: 191919566
tensorflow/core/kernels/cudnn_rnn_ops.cc
tensorflow/stream_executor/cuda/cuda_blas.cc
tensorflow/stream_executor/cuda/cuda_dnn.cc
tensorflow/stream_executor/cuda/cuda_dnn.h
tensorflow/stream_executor/cuda/cuda_timer.h
tensorflow/stream_executor/dnn.cc
tensorflow/stream_executor/dnn.h
tensorflow/stream_executor/stream.cc
tensorflow/stream_executor/stream.h
tensorflow/stream_executor/stream_executor_pimpl.cc
tensorflow/stream_executor/stream_executor_pimpl.h