* support data_norm_op run in CUDA
* add two parameters sync_stats & summary_decay_rate
* add UT

* fix warnings oof gcc 8 compilation, test=develop

* fix boost::bad_get, test=develop

* refine PADDLE_ENFORCE, test=develop

* add launch_ps module so that we can launch a parameter server training job
1) a user can specify worker_num and server_num
2) parameter server can be killed after all workers exit
3) unit test is added
test=develop

* make search_compute support avx only

* clean search_compute.h

* rename sse_axpy to avx_axpy

test=develop

* update CMakeLists.txt

test=develop

test=develop

* refector geo sgd & communicator

* treat broadcast as non-initial, test=develop

* rename the class name

* rename the class name, test=develop

* unify DataLoader APIs, test=develop

* integrate iterable CPU Dataset, test=develop
add GPU dataset supporting, test=develop

* add unittests for dataset, test=develop

* add more docs to dataloader apis, test=develop, test=document_preview

* refine doc, test=develop

* refine doc again, test=develop

* increase coverage, test=develop

test=develop

* make test_pe_seresnext serial
test=develop

* Increase test_pe_seresnext time limit on MAC
test=develop

* refactor dygraph,test=develop

* fix failed unittest,test=develop

* polish code,test=develop

* check windows ci error,test=develop
try to fix windows ci error by np.allclose,test=develop

* polish vlog and profiler, test=develop

* try to fix preceding ops order,test=develop

* test transformer in windows ci, test=develop

* use python c-api to speed up tracer.trace,test=develop

* test=develop, fix docker with paddle nccl problem

* test=develop, add ut for debug string and gradient_accumulator

* test=develop, add tests for layer/gradient_accumulator/prepared_op

* test=develop, fix complie error for test_prepared_op

* test=develop, add more ut for dygraph

* test=develop, create API.spec for dygraph api change

* test=develop, refoctor name to make it easier to understand

* test=develop, refoctor name to make it easier to understand

* test=develop, fix multi-gpu failed problem , add Tracer tests, change PADDLEENFORCE to PADDLEENFORCE_EQ

* test=develop, fix ut failed on parallel se-resnext

* test=develop, change one more PADDLE_ENFORCE

Disable test_boxps in win32.
Adjust filename to avoid latent multi-thread problem.

* remove unused PADDLE_ASSERT(_IS_NOT_ERROR)

* replace PADDLE_ASSERT with PADDLE_ASSERT_MSG

test=develop

* fixed the code for coverage

* fixed the document,test=document_preview test=develop

* update executor feed

* supports multiple NCCL communicators preserved in NCCLCommContext
test=develop

* add ut for c_comm_init_all operator and fix cuda resource release problem
test=develop

* increase test_parallel_executor_seresnext time limit
test=develop

* split test_parallel_executor_seresnext
test=develop

* temporally disable reduce_and_allreduce test because of the random failure.
test=develop

* split gpu and cpu
test=develop

* add matrch_matrix_tensor op test=develop

* fix ignore unittest if with_mkl=off test=develop

* clean code and rm is_test param test=develop

* modify API.spec test=develop

* rm useless code in search_compute.h test=develop

* modify api.spec test=develop

* modify default_grad.spec test=develop

* Add API test code test=develop

* clean code in search_computer.h

* modify PADDLE_ENFORCE and clean search_compute.h test=develop

* fix code style test=develop

Unset unittests http_proxy env to avoid timeout.

Polish fleet API to support cuda collective mode and nccl2 mode

* fix overflow by int32 mul test=develop

* fix reference nullptr

* fix codestyle test=develop

* modify to point in ContextProjectFunctor test=develop

* modify to point in ContextProjectFunctor test=develop

* modify . to -> test=develop

* add var_conv_2d op test=develop

* edit api.spec test=develop

* ignore unittest if with_mkl=off test=develop

* fix python3 division test=develop

* fix ignore unittest bug test=develop

* remove useless code test=develop

* modify api.spec test=develop

* modify default_grad.spec test=develop

make listen and server as exclusive run 

make dist unit test exclusive run

test=develop

* first version memory optimize pass, test=develop

* remove move_tensor_sharing_pass, test=develop

* refine code comments, add unittests, test=develop

* turn off memory_optimize by default, test=develop

* follow huihuang's comments, test=develop

* follow chengduoZH's comments, test=develop

* fix grammar error, add const qualifier, fix pass_test exception message, test=develop

* follow chengduoZH's comments 2nd, test=develop

* split test_dist_se_resnext.py into 4 testcases

* extend matmul op to support multiple head multiplication

With the support of multiple head, the multiplication of two big matrixes is
split into multiplication of several (head_number) small matrixes. e.g. if
Mat A is [3, 24] and Mat B is [24, 4], when multiple A and B with head_number
as 4, Mat A will be split as 4 matrix of [3, 6] and Mat B will be 4 matrix of
[6, 4]. The result of final matrix will be 4 matrix of [3, 4], i.e. [3, 16].