Add reorder flag for DynamicRNN's memory function.

Fluid distributed supports CUDA place

remove `functor1` of ElementwiseGradCompute

Add static_input for DynamicRNN

Fix grpc bugs

Adapt warpctc grad op for gradient checking

* add copyright hook

* add copyright hook

* refine copyright hook

* "test copyright hook"

* fix check style

* fix ci

Add activation for glu

Add python wrapper for GLU.

1. Fix kernel
2. Add more test case

Fix random seed of dynamic rnn gradient check

Feature/transpiler split tensor to multiple pservers

ErrorClip test on MNIST

add a brief introduction of MKL-DNN work in root README.md

Enhance print_op.

* fix while_grad_op first step loss lod problem

* optimize code

* "unified operators"

* "add CUDNN register"

* "add use cudnn attribute"

* "add attribute"

* "test conv tranpose op"

* "remove duplicated attr"

* "fix op test"

* "add attribute to set cudnn"

* "add more log"

* "need layout op register support"

* "add more log"

* "change GetExpectedKernelType "

* "fix Get attr in conv_op"

* "fix CI"

* "fix tests"

* "removed kernel priority fallback"

* "fix CI"

* "fix stack pointer bug"

* "refine buggy interface"

* "add const cast to save life"

* "fix get_output_with_grad"

* "fix op test with dataformat"

* ""fix pooling

* "fix pooling test"

* "fix CI"

* "fix with_gpu error"

* "add transform needed functional check"

* "fix unpack list error"

* "comment out parallel.do temporary"

* "fix CI"

* "fix compile doc error"

* "make threshold larger"

Fix an error of the doc for how to write a new op.

2. Remove num_seq arguments.
3. Refine CUDA kernel of ScaleLoDTensorFunctor.
4. Change max_relative_error of gradient unitest to 0.007

add wrapper for element-wise math operator.

* update code comments.

* update the comments.

* follow comments.