Some optimization of CNN model computation.

* Add unite test with less receivers, more senders

* Fixed the check

* Adding an initial implementation for the unbuffered channel

* Including atomic header

* update comment

* Adding the closed attribute

* Add comments

* Updated locking mechanism

* Add simple unbuffered test

* Enhance unit test and fix bug

* Add details

Fix document errors.

* "clean code"

* "clean code"

remove libwarpctc.so in core.so and libpaddle_fluid.so

* initial commit

* add get_parameters method

* add get_parameters method

* small fix

* address comments

* address comments

* address comments

* fix

Make Fit a line a normal unittest

Make image_classification as a normal python unittest

Fix compilie error

Fix default build_type env

Rename rpc ops

CI build: take -DCUDA_ARCH_NAME from environment variable

Make parallel tests bind to different GPU.

Refine feed_op

Add documentation generate script for fluid

Merge test_understand_sentiment together

Fix CI Error

* "Need to re-design LoD "

* "add lod design"

* "fix lod gpu ptr pointer"

* "removed commented code"

* "fix CI"

* "remove set lod in pybind"

* "fix style check"

* "fix CI"

* "fix long type template error"

* "pybind reorder to use Place"

* "fix ci"

* "fix ci"

* fix ci

* "sperate as a new file"

* "fix CI"

* "fix ci"

* small fix

* "add test"

* "fix adam op"

* "fix lstmp op"

* "fix adam op"

* "follow comments"

* "fix ci"

Make word2vec uses parallel.do when CI

update ctc_beam_search_decoder design doc

change the default option of `WITH_TESTING` to OFF