* test=develop, fix docker with paddle nccl problem

* test=develop, fix/gcc_4.8_ubt_link_error

* test=develop, fix code format

* add mkldnn shapeblob cache clear strategy

test=develop

* refine with comments

test=develop

* make cache clear strategy more safey

test=develop

* add lock for GetShapeBlobSize

test=develop

* enhance execution error info
test=develop

test=develop

* rename mkldnn set/get_cur_thread_id() to set/get_cur_mkldnn_session_id()

test=develop

* update session id definition and adjust logic for default behavior

test=develop

* reset logic in mkldnn reuse as most of cases work in default.

test=develop

1. Since allreduce op has 4 reduce types, We split these four reduce types into four ops
2. We also refined the collective op code, e.g. we separated the collective op kernel into CPUKernel and CUDAKernel, and remove the device specified DeviceContext parameter in template as we already knew the target DeviceContext
3. We remove the newly added Collective op role to reduce the complexity of program and graph analysis

* Fix bug in quantize kernel which cause crash in vgg16/19 model

test=develop

* refine the code to reduce verbose code; test=develop

* remove useless code; test=develop

1. some key generation method is not aligned with PR#17965
2. enlarge ptr lifetime to avoid memory release if SetBlob fails
   otherwise it will get core dump.

test=develop

* add dependecy of collective_helper

* test=develop
fix dependecy of collective_helper

test=develop

* fix prepare context redundant code problem, optimize executor by caching create_varaiables
test=develop

* supports collective training in executor

* make fetch_list runable with variables, add more unittest for use_program_cache
test=develop

* fix comment
test=develop

* use unique name for nccl_id

* supports output to stream in program_to_code

* insert sync_comm_stream before regularization; add skip_op_callstack capability in program_to_code

* set op role in collective training

* add collective op role

* remove orig file

* add build optimizer by strategy

* add collective strategy

* refine collective strategy

* add multi-process role maker

* refine strategy building factory so that we can easily plugin more strategy

* scale loss grad in collective sgd transpiler

* add support for distributed fc

* code format

* revert some features for dist fc

* add support for distributed fc training

* fix prepare context redundant code problem, optimize executor by caching create_varaiables
test=develop

* supports collective training in executor

* make fetch_list runable with variables, add more unittest for use_program_cache
test=develop

* use unique name for nccl_id

* supports output to stream in program_to_code

* insert sync_comm_stream before regularization; add skip_op_callstack capability in program_to_code

* set op role in collective training

* add collective op role

* fix comment
test=develop

* remove orig file

* add build optimizer by strategy

* add collective strategy

* refine collective strategy

* add multi-process role maker

* refine strategy building factory so that we can easily plugin more strategy

* scale loss grad in collective sgd transpiler

* add support for distributed fc

* code format

* revert some features for dist fc

* add support for distributed fc training

* test=develop
add collective op unittest standard

* test=develop
remove the test_collective directory

* test=develop
remove the test_collective directory

* remove slicegather test

* code format for reducescatter

* update attr of shard_index_op

* Modify macro nccl_helper

* remove test without distribute

* macro collective_helper

* marcro update

* test=develop
update support python3.5

* test=develop change gpu memory use to 0.1 when test

* test=develop
update ut equal func

* test=develop
set flags to 1.5

* test=develop fix pickle dumple  py35

* test=develop
fix divide in slice and add sync_comm_stream
update atol and rtol to 1e-05
rm shard_index op and test
modify read input from file to read from memory
remove origin_program in framework and add i/o in c_sync_calc_stream

* test=develop update unittest sync operator I/O

* - Reusing of reuder used in elementwise_add_mkldnn

- Added MKL-DNN sum prim reusing

test=develop

- Compilation fixes

test=develop

- Yet another compilation fix

test=develop

- Yet another compilation fix

test=develo

- Yet another linking fix

test=develop

- Final compilation fix

test=develop

- lint fixes

test=develop

- Lint fixes

test=develop

* - Fixes after review

test=develop

* remove nccl dep when the number of GPU is 1
test=develop

* - removed is_reusing_

* - Added TID to keys for reusing apart from softmax PD

* - compilation fix

* - Yet another compilation fix

* - Batch Norm and Conv adapted

* - Fix to softmax MT

* - Fixes to MT code of MKL-DNN

* - Lint fixes

test=develop

Add Pipeline Concurrency Train Mode:
- Cpp: pipeline_trainer & section_worker
- Python: PipelineOptimizer
- Add a new data_feed type: PrivateInstantDataFeed
- Add a test demo of pipeline trainer and the test model is gnn
- Do not support win32 now

* remove attribute in Allocator::Allocate, test=develop

* fix travis ci error, test=develop

* fix cuda/cudnn version detection error, test=develop

* fix again, test=develop

test=develop

test=develop

* revise conv layer cudnn algo choose test=develop

* update for code style test=develop

* update for code style test=develop

test=develop

* add __str__ method for tensor and lodtensor to support print test=develop

*  Enable assign operator for a ngraph, test=develop

* Cross_entropy operators needs to be updated

* Revert "Revert "Fix allocator bug""

This reverts commit 174d0d0b.

* Revert "fix travis ci"

This reverts commit 5656fa9f.

test=develop

* add inlined_vector.h, test=develop

* add inlined_vector_test,test=develop

* Relu6 is the bottleneck op for Mobilenet-v2. As the mkldnn supports the conv/relu6 fusion, we implement it fusion via cpass way. Due to the int8 enabling for this fusion will be supported in MKLDNN v0.20, so this PR is focused on the fp32 optimization.

Below table shows the benchmark(FPS) which measured on skx-8180(28 cores)
Batch size | with fusion | without fusion
-- | -- | --
1 | 214.7 | 53.4
50 | 1219.727 | 137.280

test=develop

* Fix the format issue

test=develop

* Add the missing nolint comments.

test=develop

* Fix the typos.

test=develop

* Register the conv_brelu_mkldnn_fuse_pass for the MKLDNN engine.

test=develop

* Adjust the indentation.

test=develop

* Add the test_conv_brelu_mkldnn_fuse_pass case.

test=develop

* Slightly update the code per Baidu comments.
Let the parameter definition embedded into the code.
That's will make the code easy to understand.

test=develop

test=develop

* Add conv2d_grad_grad_op
* Extracte the cuDNN conv algo searching code in conv_cudnn_helper.h.
    - Now use it in conv2d_grad_grad.
    - Will simply the searching code in conv2d and conv2d_grad in next PR.
* Enhance and fix bug in unit testing of gradient_checker.
* Support to fetch empty variables，return None in Python.

* Refine elementwise kernel.

Add a simple cuda kernel if grad x and y both exist
Use 2D block cuda kernel to do broadcast.

test=develop
Signed-off-by: Nzhaoyuchen <zhaoyuchen01@baidu.com>

* refine code.

test=develop
Signed-off-by: Nzhaoyuchen <zhaoyuchen01@baidu.com>

* refine code.

test=develop
Signed-off-by: Nzhaoyuchen <zhaoyuchen01@baidu.com>

test=develop

* added lrn op test=develop

* Added CreateConstant method test=develop

* avoid duplicates test=develop

* remove unused FLAGS_warpctc_dir

test=develop

* remove FLAGS_warpctc_dir

test=develop

test=develop

1. Use CudnnWorkspaceHandle in exhaustive search of conv_cudnn.
2. For Ops using CudnnWorkspaceHandle in exhaustive search, release their GPU memory after exhaustive search.

test=develop

* make_conv_cudnn_ws_size_configurable, test=develop

* change std::max to std::min
test=develop