fix merge conflict

739c3309 · sneaxiy · 1b59220d · 81f22bb2 · 739c3309 · 739c3309
1000 changed file
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -23,7 +23,7 @@ repos:
    -   id: clang-format-with-version-check
        name: clang-format
        description: Format files with ClangFormat.
-        entry: bash ./.clang_format.hook -i
+        entry: bash ./tools/codestyle/clang_format.hook -i
        language: system
        files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|proto)$
 -   repo: local
@@ -52,7 +52,7 @@ repos:
    hooks:
    -   id: copyright_checker
        name: copyright_checker
-        entry: python ./.copyright.hook
+        entry: python ./tools/codestyle/copyright.hook
        language: system
        files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|proto|py)$
        exclude: (?!.*third_party)^.*$ | (?!.*book)^.*$
--- a/AUTHORS.md
+++ b/AUTHORS.md
@@ -22,6 +22,7 @@
 | jczaja | Jacek Czaja |
 | JiayiFeng | Jia-Yi Feng |
 | kbinias | Krzysztof Binias |
+| kexinzhao | Ke-Xin Zhao |
 | kuke | Yi-Bing Liu |
 | lcy-seso | Ying Cao |
 | lipeng-unisound | Peng Li |

--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -61,7 +61,10 @@ option(EIGEN_USE_THREADS "Compile with multi-threaded Eigen"            OFF)
 option(WITH_ARM_FP16    "Use half precision support on armv8.2-a cpu"   OFF)
 option(WITH_FAST_BUNDLE_TEST    "Bundle tests that can be run in a single process together to reduce launch overhead"   OFF)
 option(WITH_CONTRIB     "Compile the third-party contributation"        OFF)
+option(REPLACE_ENFORCE_GLOG "Replace PADDLE_ENFORCE with glog/CHECK for better debug." OFF)
+option(WITH_ANAKIN      "Compile with Anakin library"                   OFF)
 option(WITH_GRPC     "Use grpc as the default rpc framework"            ${WITH_DISTRIBUTE})
+option(WITH_BRPC_RDMA     "Use brpc rdma as the rpc protocal"           OFF)

 # CMAKE_BUILD_TYPE
 if(NOT CMAKE_BUILD_TYPE)
@@ -130,6 +133,10 @@ if (NOT DEFINED WITH_MKLDNN)
        set(WITH_MKLDNN OFF)
    endif()
 endif()
+
+if (REPLACE_ENFORCE_GLOG)
+  add_definitions("-DREPLACE_ENFORCE_GLOG")
+endif()
 ########################################################################################

 include(external/mklml)     # download mklml package
@@ -152,12 +159,24 @@ include(external/cares)
 if(WITH_DISTRIBUTE)
    if(WITH_GRPC)
        include(external/grpc)
+        message(STATUS "Use grpc framework.")
    else()
+        message(STATUS "Use brpc framework.")
        include(external/leveldb)
        include(external/brpc)
    endif()
 endif()

+if(WITH_BRPC_RDMA)
+    message(STATUS "Use brpc with rdma.")
+    if(WITH_GRPC)
+        message(FATAL_ERROR "Can't use grpc with brpc rdma.")
+    endif()
+    if(NOT WITH_DISTRIBUTE)
+        message(FATAL_ERROR "Can't use brpc rdma in no distribute env.")
+    endif()
+endif()
+
 include(external/snappy)    # download snappy
 include(external/snappystream)
 include(external/threadpool)
@@ -177,7 +196,7 @@ include(inference_lib)      # add paddle fluid inference libraries


 include_directories("${PADDLE_SOURCE_DIR}")
-include_directories("${PADDLE_SOURCE_DIR}/paddle/cuda/include")
+include_directories("${PADDLE_SOURCE_DIR}/paddle/legacy/cuda/include")
 include_directories("${CMAKE_CURRENT_BINARY_DIR}/proto")
 include_directories("${CMAKE_CURRENT_BINARY_DIR}/go/pserver/client/c")

@@ -193,7 +212,10 @@ set(EXTERNAL_LIBS
 if(WITH_GPU)
    include(cuda)
    include(tensorrt)
-endif(WITH_GPU)
+    include(external/anakin)
+else()
+  set(WITH_ANAKIN OFF CACHE STRING "Anakin is valid only when GPU is set." FORCE)
+endif()

 if(WITH_AMD_GPU)
    find_package(HIP)
@@ -218,7 +240,7 @@ add_subdirectory(proto)
 if(NOT MOBILE_INFERENCE AND NOT WITH_FLUID_ONLY)
    # "add_subdirectory(go)" should be placed after the following loine,
    # because it depends on paddle/optimizer.
-    add_subdirectory(paddle/optimizer)
+    add_subdirectory(paddle/legacy/optimizer)
 endif()

 # "add_subdirectory(paddle)" and "add_subdirectory(python)" should be

--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -159,4 +159,4 @@ This will enable VLOG messages generated by `buddy_allocator.{h,cc}` and in the
 - verbose level 1: [framework](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/framework)
 - verbose level 3: [operators](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/operators)
 - verbose level 5: [memory](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/memory), [platform](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/platform)
- verbose level 7: [math](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/math)
+- verbose level 7: [math](https://github.com/PaddlePaddle/Paddle/tree/develop/paddle/legacy/math)
--- a/Dockerfile
+++ b/Dockerfile
@@ -76,7 +76,8 @@ RUN easy_install -U pip && \
    pip install sphinx-rtd-theme==0.1.9 recommonmark

 RUN pip install pre-commit 'ipython==5.3.0' && \
-    pip install 'ipykernel==4.6.0' 'jupyter==1.0.0'
+    pip install 'ipykernel==4.6.0' 'jupyter==1.0.0' && \
+    pip install opencv-python

 #For docstring checker
 RUN pip install pylint pytest astroid isort

--- a/README.md
+++ b/README.md
@@ -4,7 +4,6 @@
 [![Build Status](https://travis-ci.org/PaddlePaddle/Paddle.svg?branch=develop)](https://travis-ci.org/PaddlePaddle/Paddle)
 [![Documentation Status](https://img.shields.io/badge/docs-latest-brightgreen.svg?style=flat)](http://www.paddlepaddle.org/docs/develop/documentation/en/getstarted/index_en.html)
 [![Documentation Status](https://img.shields.io/badge/中文文档-最新-brightgreen.svg)](http://www.paddlepaddle.org/docs/develop/documentation/zh/getstarted/index_cn.html)
-[![Coverage Status](https://coveralls.io/repos/github/PaddlePaddle/Paddle/badge.svg?branch=develop)](https://coveralls.io/github/PaddlePaddle/Paddle?branch=develop)
 [![Release](https://img.shields.io/github/release/PaddlePaddle/Paddle.svg)](https://github.com/PaddlePaddle/Paddle/releases)
 [![License](https://img.shields.io/badge/license-Apache%202-blue.svg)](LICENSE)


--- a/benchmark/fluid/Dockerfile
+++ b/benchmark/fluid/Dockerfile
 FROM nvidia/cuda:9.0-cudnn7-devel-ubuntu16.04
+
+# Use UBUNTU_MIRROR can speed up apt-get speed.
+# ARG UBUNTU_MIRROR
+# RUN /bin/bash -c 'if [[ -n ${UBUNTU_MIRROR} ]]; then sed -i 's#http://archive.ubuntu.com/ubuntu#${UBUNTU_MIRROR}#g' /etc/apt/sources.list; fi'
+
 RUN apt-get update && apt-get install -y python python-pip iputils-ping libgtk2.0-dev wget vim net-tools iftop python-opencv
 RUN ln -s /usr/lib/x86_64-linux-gnu/libcudnn.so.7 /usr/lib/libcudnn.so && ln -s /usr/lib/x86_64-linux-gnu/libnccl.so.2 /usr/lib/libnccl.so
-RUN pip install -U pip
-RUN pip install -U kubernetes paddlepaddle

 # IMPORTANT:
 # Add "ENV http_proxy=http://ip:port" if your download is slow, and don't forget to unset it at runtime.
+# exmaple: unset http_proxy && unset https_proxy && python fluid_benchmark.py ...
+
+RUN pip install -U pip
+RUN pip install -U kubernetes paddlepaddle

 RUN sh -c 'echo "import paddle.v2 as paddle\npaddle.dataset.cifar.train10()\npaddle.dataset.flowers.fetch()" | python'
 RUN sh -c 'echo "import paddle.v2 as paddle\npaddle.dataset.mnist.train()\npaddle.dataset.mnist.test()\npaddle.dataset.imdb.fetch()" | python'
@@ -14,9 +21,11 @@ RUN pip uninstall -y paddlepaddle && mkdir /workspace

 ADD https://raw.githubusercontent.com/PaddlePaddle/cloud/develop/docker/paddle_k8s /usr/bin
 ADD https://raw.githubusercontent.com/PaddlePaddle/cloud/develop/docker/k8s_tools.py /root
+RUN chmod +x /usr/bin/paddle_k8s

 ADD *.whl /
-RUN pip install /*.whl && rm -f /*.whl && chmod +x /usr/bin/paddle_k8s
+RUN pip install /*.whl && rm -f /*.whl 

 ENV LD_LIBRARY_PATH=/usr/local/lib
-ADD fluid_benchmark.py recordio_converter.py models/ /workspace/
+ADD fluid_benchmark.py recordio_converter.py args.py recordio_converter.py run.sh run_fluid_benchmark.sh /workspace/
+ADD models/ /workspace/models/
--- a/benchmark/fluid/args.py
+++ b/benchmark/fluid/args.py
@@ -122,5 +122,9 @@ def parse_args():
        type=str,
        default="",
        help='Directory that contains all the training recordio files.')
+    parser.add_argument(
+        '--use_inference_transpiler',
+        action='store_true',
+        help='If set, uses inference transpiler to optimize the program.')
    args = parser.parse_args()
    return args
--- a/benchmark/fluid/fluid_benchmark.py
+++ b/benchmark/fluid/fluid_benchmark.py
@@ -97,7 +97,7 @@ def dist_transpile(trainer_id, args):
        return train_program, fluid.default_startup_program()
    else:
        raise ValueError(
-            'TRAINING_ROLE environment variable must be either TRAINER or PSERVER'
+            'PADDLE_TRAINING_ROLE environment variable must be either TRAINER or PSERVER'
        )


@@ -131,6 +131,11 @@ def train(avg_loss, infer_prog, optimizer, train_reader, test_reader, batch_acc,
    exe = fluid.Executor(place)
    exe.run(startup_prog)

+    # Use inference_transpiler to speedup
+    if args.use_inference_transpiler:
+        t = fluid.InferenceTranspiler()
+        t.transpile(infer_prog, place)
+
    if not args.use_reader_op:
        feed_var_list = [
            var for var in train_prog.global_block().vars.itervalues()
@@ -180,7 +185,7 @@ def train(avg_loss, infer_prog, optimizer, train_reader, test_reader, batch_acc,
        print_train_time(start_time, time.time(), num_samples)
        print("Pass: %d, Loss: %f" % (pass_id, np.mean(train_losses))),
        # evaluation
-        if not args.no_test and batch_acc:
+        if not args.no_test and batch_acc and not args.use_reader_op:
            pass_test_acc = test(exe, infer_prog, test_reader, feeder,
                                 batch_acc)
            print(", Test Accuracy: %f" % pass_test_acc)
@@ -264,8 +269,6 @@ def train_parallel(avg_loss, infer_prog, optimizer, train_reader, test_reader,
                    break
            else:
                loss, = exe.run([avg_loss.name], feed=feeder.feed(data))
-            if args.update_method == "pserver":
-                exe.bcast_params()
            if args.use_reader_op:
                num_samples += args.batch_size * args.gpus
            else:
@@ -277,11 +280,12 @@ def train_parallel(avg_loss, infer_prog, optimizer, train_reader, test_reader,
            batch_id += 1

        print_train_time(start_time, time.time(), num_samples)
-        if not args.no_test and batch_acc:
+        if not args.no_test and batch_acc and not args.use_reader_op:
+            # we have not implement record io for test
+            # skip test when use args.use_reader_op
            test_acc = test(startup_exe, infer_prog, test_reader, feeder,
                            batch_acc)
            print("Pass: %d, Test Accuracy: %f\n" % (pass_id, test_acc))
-        exit(0)


 def print_arguments(args):
@@ -300,9 +304,18 @@ def print_train_time(start_time, end_time, num_samples):
          (num_samples, train_elapsed, examples_per_sec))


+def print_paddle_envs():
+    print('----------- Configuration envs -----------')
+    for k in os.environ:
+        if "PADDLE_" in k:
+            print "ENV %s:%s" % (k, os.environ[k])
+    print('------------------------------------------------')
+
+
 def main():
    args = parse_args()
    print_arguments(args)
+    print_paddle_envs()

    # the unique trainer id, starting from 0, needed by trainer
    # only

--- a/benchmark/fluid/kube_gen_job.py
+++ b/benchmark/fluid/kube_gen_job.py
@@ -17,6 +17,7 @@ import copy
 import argparse
 import random
 import os
+import copy
 from kube_templates import pserver, trainer, envs


@@ -108,10 +109,9 @@ def gen_job():
    tn_container["ports"][0]["containerPort"] = spreadport

    envs.append({"name": "PADDLE_JOB_NAME", "value": args.jobname})
-    envs.append({"name": "TRAINERS", "value": str(args.trainers)})
-    envs.append({"name": "PSERVERS", "value": str(args.pservers)})
+    envs.append({"name": "PADDLE_TRAINERS", "value": str(args.trainers)})
+    envs.append({"name": "PADDLE_PSERVERS", "value": str(args.pservers)})
    envs.append({"name": "ENTRY", "value": args.entry})
-    envs.append({"name": "PADDLE_INIT_PORT", "value": str(args.port)})
    envs.append({"name": "PADDLE_PSERVER_PORT", "value": str(args.port)})
    # NOTE: these directories below are cluster specific, please modify
    # this settings before you run on your own cluster.
@@ -166,17 +166,23 @@ def gen_job():
    tn["spec"]["template"]["spec"]["volumes"] = volumes
    tn_container["volumeMounts"] = volumeMounts

-    ps_container["env"] = envs
-    ps_container["env"].append({"name": "TRAINING_ROLE", "value": "PSERVER"})
+    ps_container["env"] = copy.deepcopy(envs)
+    ps_container["env"].append({
+        "name": "PADDLE_TRAINING_ROLE",
+        "value": "PSERVER"
+    })
    tn_container["env"] = envs
    if args.disttype == "pserver":
        tn_container["env"].append({
-            "name": "TRAINING_ROLE",
+            "name": "PADDLE_TRAINING_ROLE",
            "value": "TRAINER"
        })
    elif args.disttype == "nccl2" or args.disttype == "local":
        # NCCL2 have no training role, set to plain WORKER
-        tn_container["env"].append({"name": "TRAINING_ROLE", "value": "WORKER"})
+        tn_container["env"].append({
+            "name": "PADDLE_TRAINING_ROLE",
+            "value": "WORKER"
+        })

    os.mkdir(args.jobname)
    if args.disttype == "pserver":

--- a/benchmark/fluid/models/machine_translation.py
+++ b/benchmark/fluid/models/machine_translation.py
@@ -173,21 +173,6 @@ def seq_to_seq_net(embedding_dim, encoder_size, decoder_size, source_dict_dim,
        return avg_cost, feeding_list


-def to_lodtensor(data, place):
-    seq_lens = [len(seq) for seq in data]
-    cur_len = 0
-    lod = [cur_len]
-    for l in seq_lens:
-        cur_len += l
-        lod.append(cur_len)
-    flattened_data = np.concatenate(data, axis=0).astype("int64")
-    flattened_data = flattened_data.reshape([len(flattened_data), 1])
-    lod_t = core.LoDTensor()
-    lod_t.set(flattened_data, place)
-    lod_t.set_lod([lod])
-    return lod_t, lod[-1]
-
-
 def lodtensor_to_ndarray(lod_tensor):
    dims = lod_tensor.get_dims()
    ndarray = np.zeros(shape=dims).astype('float32')

--- a/benchmark/fluid/models/resnet.py
+++ b/benchmark/fluid/models/resnet.py
@@ -199,7 +199,10 @@ def get_model(args):
    batched_train_reader = paddle.batch(
        paddle.reader.shuffle(
            train_reader, buf_size=5120),
-        batch_size=args.batch_size * args.gpus)
-    batched_test_reader = paddle.batch(train_reader, batch_size=args.batch_size)
+        batch_size=args.batch_size * args.gpus,
+        drop_last=True)
+    batched_test_reader = paddle.batch(
+        train_reader, batch_size=args.batch_size, drop_last=True)

-    return avg_cost, inference_program, optimizer, batched_train_reader, batched_test_reader, batch_acc
+    return avg_cost, inference_program, optimizer, batched_train_reader,\
+                   batched_test_reader, batch_acc
--- a/benchmark/fluid/models/stacked_dynamic_lstm.py
+++ b/benchmark/fluid/models/stacked_dynamic_lstm.py
@@ -125,18 +125,3 @@ def get_model(args):
        batch_size=args.batch_size)

    return loss, inference_program, adam, train_reader, test_reader, batch_acc
-
-
-def to_lodtensor(data, place):
-    seq_lens = [len(seq) for seq in data]
-    cur_len = 0
-    lod = [cur_len]
-    for l in seq_lens:
-        cur_len += l
-        lod.append(cur_len)
-    flattened_data = numpy.concatenate(data, axis=0).astype("int64")
-    flattened_data = flattened_data.reshape([len(flattened_data), 1])
-    res = fluid.LoDTensor()
-    res.set(flattened_data, place)
-    res.set_lod([lod])
-    return res
--- a/cmake/configure.cmake
+++ b/cmake/configure.cmake
@@ -118,6 +118,10 @@ endif()
 set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${SIMD_FLAG}")
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${SIMD_FLAG}")

+if(WITH_DISTRIBUTE)
+  add_definitions(-DPADDLE_WITH_DISTRIBUTE)
+endif()
+
 if(WITH_GOLANG)
  # we need to symlink Paddle directory into GOPATH. If we
  # don't do it and we have code that depends on Paddle, go
@@ -170,3 +174,7 @@ endif(WITH_GOLANG)
 if(WITH_GRPC)
    add_definitions(-DPADDLE_WITH_GRPC)
 endif(WITH_GRPC)
+
+if(WITH_BRPC_RDMA)
+    add_definitions(-DPADDLE_WITH_BRPC_RDMA)
+endif(WITH_BRPC_RDMA)
--- a/cmake/external/anakin.cmake
+++ b/cmake/external/anakin.cmake
+if (NOT WITH_ANAKIN)
+  return()
+endif()
+
+set(ANAKIN_INSTALL_DIR "${THIRD_PARTY_PATH}/install/anakin" CACHE PATH
+  "Anakin install path." FORCE)
+set(ANAKIN_INCLUDE "${ANAKIN_INSTALL_DIR}" CACHE STRING "root of Anakin header files")
+set(ANAKIN_LIBRARY "${ANAKIN_INSTALL_DIR}" CACHE STRING "path of Anakin library")
+
+set(ANAKIN_COMPILE_EXTRA_FLAGS -Wno-error=unused-variable -Wno-error=format-extra-args -Wno-error=comment -Wno-error=format -Wno-error=switch -Wno-error=return-type -Wno-error=non-virtual-dtor -Wno-reorder -Wno-error=cpp)
+
+set(ANAKIN_LIBRARY_URL "https://github.com/pangge/Anakin/releases/download/3.0/anakin_release_simple.tar.gz")
+
+# A helper function used in Anakin, currently, to use it, one need to recursively include
+# nearly all the header files.
+function(fetch_include_recursively root_dir)
+    if (IS_DIRECTORY ${root_dir})
+        include_directories(${root_dir})
+    endif()
+
+    file(GLOB ALL_SUB RELATIVE ${root_dir} ${root_dir}/*)
+    foreach(sub ${ALL_SUB})
+        if (IS_DIRECTORY ${root_dir}/${sub})
+            fetch_include_recursively(${root_dir}/${sub})
+        endif()
+    endforeach()
+endfunction()
+
+if (NOT EXISTS "${ANAKIN_INSTALL_DIR}")
+    # download library
+    message(STATUS "Download Anakin library from ${ANAKIN_LIBRARY_URL}")
+    execute_process(COMMAND bash -c "mkdir -p ${ANAKIN_INSTALL_DIR}")
+    execute_process(COMMAND bash -c "rm -rf ${ANAKIN_INSTALL_DIR}/*")
+    execute_process(COMMAND bash -c "cd ${ANAKIN_INSTALL_DIR}; wget -q ${ANAKIN_LIBRARY_URL}")
+    execute_process(COMMAND bash -c "mkdir -p ${ANAKIN_INSTALL_DIR}")
+    execute_process(COMMAND bash -c "cd ${ANAKIN_INSTALL_DIR}; tar xzf anakin_release_simple.tar.gz")
+endif()
+
+if (WITH_ANAKIN)
+    message(STATUS "Anakin for inference is enabled")
+    message(STATUS "Anakin is set INCLUDE:${ANAKIN_INCLUDE} LIBRARY:${ANAKIN_LIBRARY}")
+    fetch_include_recursively(${ANAKIN_INCLUDE})
+    link_directories(${ANAKIN_LIBRARY})
+endif()
--- a/cmake/external/brpc.cmake
+++ b/cmake/external/brpc.cmake
@@ -14,6 +14,15 @@

 INCLUDE(ExternalProject)

+find_library(SSL_LIBRARY NAMES ssl)
+ADD_LIBRARY(ssl SHARED IMPORTED GLOBAL)
+SET_PROPERTY(TARGET ssl PROPERTY IMPORTED_LOCATION ${SSL_LIBRARY})
+
+find_library(CRYPTO_LIBRARY NAMES crypto)
+ADD_LIBRARY(crypto SHARED IMPORTED GLOBAL)
+SET_PROPERTY(TARGET crypto PROPERTY IMPORTED_LOCATION ${CRYPTO_LIBRARY})
+
+
 SET(BRPC_SOURCES_DIR ${THIRD_PARTY_PATH}/brpc)
 SET(BRPC_INSTALL_DIR ${THIRD_PARTY_PATH}/install/brpc)
 SET(BRPC_INCLUDE_DIR "${BRPC_INSTALL_DIR}/include" CACHE PATH "brpc include directory." FORCE)
@@ -22,14 +31,14 @@ SET(BRPC_LIBRARIES "${BRPC_INSTALL_DIR}/lib/libbrpc.a" CACHE FILEPATH "brpc libr
 INCLUDE_DIRECTORIES(${BRPC_INCLUDE_DIR})

 # Reference https://stackoverflow.com/questions/45414507/pass-a-list-of-prefix-paths-to-externalproject-add-in-cmake-args
-set(prefix_path "${THIRD_PARTY_PATH}/install/gflags|${THIRD_PARTY_PATH}/install/leveldb|${THIRD_PARTY_PATH}/install/snappy|${THIRD_PARTY_PATH}/install/gtest|${THIRD_PARTY_PATH}/install/protobuf")
+set(prefix_path "${THIRD_PARTY_PATH}/install/gflags|${THIRD_PARTY_PATH}/install/leveldb|${THIRD_PARTY_PATH}/install/snappy|${THIRD_PARTY_PATH}/install/gtest|${THIRD_PARTY_PATH}/install/protobuf|${THIRD_PARTY_PATH}/install/zlib")

 # If minimal .a is need, you can set  WITH_DEBUG_SYMBOLS=OFF
 ExternalProject_Add(
    extern_brpc
    ${EXTERNAL_PROJECT_LOG_ARGS}
-    GIT_REPOSITORY  "https://github.com/brpc/brpc"
-    GIT_TAG         "6d153dd7ff00f960ae6895c9c5fff0ce9f07aff2"
+    GIT_REPOSITORY  "https://github.com/gongweibao/brpc"
+    GIT_TAG         "7dc04defad1fd4173aae170c3fcbde131b65155a"
    PREFIX          ${BRPC_SOURCES_DIR}
    UPDATE_COMMAND  ""
    CMAKE_ARGS      -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
@@ -42,6 +51,8 @@ ExternalProject_Add(
                    -DCMAKE_BUILD_TYPE=${THIRD_PARTY_BUILD_TYPE}
                    -DCMAKE_PREFIX_PATH=${prefix_path}
                    -DBRPC_WITH_GLOG=ON
+                    -DIOBUF_WITH_HUGE_BLOCK=ON
+                    -DBRPC_WITH_RDMA=${WITH_BRPC_RDMA}
                    ${EXTERNAL_OPTIONAL_ARGS}
    LIST_SEPARATOR |
    CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${BRPC_INSTALL_DIR}
@@ -49,7 +60,7 @@ ExternalProject_Add(
                     -DCMAKE_POSITION_INDEPENDENT_CODE:BOOL=ON
                     -DCMAKE_BUILD_TYPE:STRING=${THIRD_PARTY_BUILD_TYPE}
 )
-ADD_DEPENDENCIES(extern_brpc protobuf leveldb gflags glog gtest snappy)
+ADD_DEPENDENCIES(extern_brpc protobuf ssl crypto leveldb gflags glog gtest snappy)
 ADD_LIBRARY(brpc STATIC IMPORTED GLOBAL)
 SET_PROPERTY(TARGET brpc PROPERTY IMPORTED_LOCATION ${BRPC_LIBRARIES})
 ADD_DEPENDENCIES(brpc extern_brpc)

--- a/cmake/external/grpc.cmake
+++ b/cmake/external/grpc.cmake
@@ -40,12 +40,12 @@ ExternalProject_Add(
    # NOTE(wuyi):
    # this package is generated by following steps:
    # 1. git clone -b v1.8.x https://github.com/grpc/grpc.git
-    # 2. submodule update --init
+    # 2. git submodule update --init
    # 3. keep only zlib, cares, protobuf, boringssl under "third_party",
    #    checkout and clean other dirs under third_party
    # 4. remove .git, and package the directory.
-    URL "http://paddlepaddledeps.bj.bcebos.com/grpc-v1.8.x.tar.gz"
-    URL_MD5  "c9c58ee7d0e8929a63155af6a2ecdbd0"
+    URL "http://paddlepaddledeps.bj.bcebos.com/grpc-v1.10.x.tar.gz"
+    URL_MD5  "1f268a2aff6759839dccd256adcc91cf"
    PREFIX          ${GRPC_SOURCES_DIR}
    UPDATE_COMMAND  ""
    CONFIGURE_COMMAND ""

--- a/cmake/external/mkldnn.cmake
+++ b/cmake/external/mkldnn.cmake
@@ -45,7 +45,8 @@ IF(${CBLAS_PROVIDER} STREQUAL "MKLML")
 ELSE()
    MESSAGE(FATAL_ERROR "Should enable MKLML when build MKLDNN")
 ENDIF()
-SET(MKLDNN_FLAG "-Wno-error=strict-overflow -Wno-error=unused-result -Wno-unused-result")
+SET(MKLDNN_FLAG "-Wno-error=strict-overflow -Wno-error=unused-result")
+SET(MKLDNN_FLAG "${MKLDNN_FLAG} -Wno-unused-result -Wno-unused-value")
 SET(MKLDNN_CFLAG "${CMAKE_C_FLAGS} ${MKLDNN_FLAG}")
 SET(MKLDNN_CXXFLAG "${CMAKE_CXX_FLAGS} ${MKLDNN_FLAG}")
 ExternalProject_Add(
@@ -53,7 +54,7 @@ ExternalProject_Add(
    ${EXTERNAL_PROJECT_LOG_ARGS}
    DEPENDS             ${MKLDNN_DEPENDS}
    GIT_REPOSITORY      "https://github.com/01org/mkl-dnn.git"
-    GIT_TAG             "db3424ad44901513c03a1ea31ccaacdf633fbe9f"
+    GIT_TAG             "a29d8487a63afca3d5b8c5bbdbb473cf8ccc6e51"
    PREFIX              ${MKLDNN_SOURCES_DIR}
    UPDATE_COMMAND      ""
    CMAKE_ARGS          -DCMAKE_INSTALL_PREFIX=${MKLDNN_INSTALL_DIR}

--- a/cmake/external/openblas.cmake
+++ b/cmake/external/openblas.cmake
@@ -29,6 +29,8 @@ IF(NOT ${CBLAS_FOUND})
        "${CBLAS_INSTALL_DIR}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}openblas${CMAKE_STATIC_LIBRARY_SUFFIX}"
        CACHE FILEPATH "openblas library." FORCE)

+    ADD_DEFINITIONS(-DPADDLE_USE_OPENBLAS)
+
    SET(OPENBLAS_CC "${CMAKE_C_COMPILER} -Wno-unused-but-set-variable -Wno-unused-variable")
    SET(OPENBLAS_COMMIT "v0.2.20")

@@ -112,7 +114,12 @@ INCLUDE_DIRECTORIES(${CBLAS_INC_DIR})
 SET(dummyfile ${CMAKE_CURRENT_BINARY_DIR}/cblas_dummy.c)
 FILE(WRITE ${dummyfile} "const char *dummy_cblas = \"${dummyfile}\";")
 ADD_LIBRARY(cblas STATIC ${dummyfile})
-TARGET_LINK_LIBRARIES(cblas ${CBLAS_LIBRARIES})
+
+IF("${CBLAS_PROVIDER}" STREQUAL "MKLML")
+  TARGET_LINK_LIBRARIES(cblas dynload_mklml)
+ELSE()
+  TARGET_LINK_LIBRARIES(cblas ${CBLAS_LIBRARIES})
+ENDIF("${CBLAS_PROVIDER}" STREQUAL "MKLML")

 IF(NOT ${CBLAS_FOUND})
    ADD_DEPENDENCIES(cblas extern_openblas)

--- a/cmake/generic.cmake
+++ b/cmake/generic.cmake
@@ -96,6 +96,20 @@ if(NOT APPLE AND NOT ANDROID)
    set(CMAKE_CXX_LINK_EXECUTABLE "${CMAKE_CXX_LINK_EXECUTABLE} -pthread -ldl -lrt")
 endif(NOT APPLE AND NOT ANDROID)

+set_property(GLOBAL PROPERTY FLUID_MODULES "")
+# find all fluid modules is used for paddle fluid static library
+# for building inference libs
+function(find_fluid_modules TARGET_NAME)
+  get_filename_component(__target_path ${TARGET_NAME} ABSOLUTE)
+  string(REGEX REPLACE "^${PADDLE_SOURCE_DIR}/" "" __target_path ${__target_path})
+  string(FIND "${__target_path}" "fluid" pos)
+  if(pos GREATER 1)
+    get_property(fluid_modules GLOBAL PROPERTY FLUID_MODULES)
+    set(fluid_modules ${fluid_modules} ${TARGET_NAME})
+    set_property(GLOBAL PROPERTY FLUID_MODULES "${fluid_modules}")
+  endif()
+endfunction(find_fluid_modules)
+
 function(merge_static_libs TARGET_NAME)
  set(libs ${ARGN})
  list(REMOVE_DUPLICATES libs)
@@ -195,6 +209,15 @@ function(cc_library TARGET_NAME)
        list(REMOVE_ITEM cc_library_DEPS warpctc)
        add_dependencies(${TARGET_NAME} warpctc)
      endif()
+      # Only deps libmklml.so, not link
+      if("${cc_library_DEPS};" MATCHES "mklml;")
+        list(REMOVE_ITEM cc_library_DEPS mklml)
+        if(NOT "${TARGET_NAME}" MATCHES "dynload_mklml")
+          list(APPEND cc_library_DEPS dynload_mklml)
+        endif()
+        add_dependencies(${TARGET_NAME} mklml)
+        target_link_libraries(${TARGET_NAME} "-L${MKLML_LIB_DIR} -liomp5 -Wl,--as-needed")
+      endif()
      target_link_libraries(${TARGET_NAME} ${cc_library_DEPS})
      add_dependencies(${TARGET_NAME} ${cc_library_DEPS})
    endif()
@@ -241,6 +264,7 @@ function(cc_test TARGET_NAME)
             WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})
    if (${cc_test_SERIAL})
        set_property(TEST ${TARGET_NAME} PROPERTY SERIAL 1)
+    set_property(TEST ${TARGET_NAME} PROPERTY ENVIRONMENT FLAGS_init_allocated_mem=true)
    endif()
  endif()
 endfunction(cc_test)
@@ -305,6 +329,7 @@ function(nv_test TARGET_NAME)
    add_test(${TARGET_NAME} ${TARGET_NAME})
    if (nv_test_SERIAL)
        set_property(TEST ${TARGET_NAME} PROPERTY SERIAL 1)
+    set_property(TEST ${TARGET_NAME} PROPERTY ENVIRONMENT FLAGS_init_allocated_mem=true)
    endif()
  endif()
 endfunction(nv_test)
@@ -552,7 +577,7 @@ function(py_test TARGET_NAME)
    set(multiValueArgs SRCS DEPS ARGS ENVS)
    cmake_parse_arguments(py_test "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
    add_test(NAME ${TARGET_NAME}
-             COMMAND env PYTHONPATH=${PADDLE_BINARY_DIR}/python ${py_test_ENVS}
+             COMMAND env FLAGS_init_allocated_mem=true PYTHONPATH=${PADDLE_BINARY_DIR}/python ${py_test_ENVS}
             ${PYTHON_EXECUTABLE} -u ${py_test_SRCS} ${py_test_ARGS}
             WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})
  endif()

--- a/cmake/inference_lib.cmake
+++ b/cmake/inference_lib.cmake
@@ -12,19 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.

-set_property(GLOBAL PROPERTY FLUID_MODULES "")
-# find all fluid modules is used for paddle fluid static library
-function(find_fluid_modules TARGET_NAME)
-  get_filename_component(__target_path ${TARGET_NAME} ABSOLUTE)
-  string(REGEX REPLACE "^${PADDLE_SOURCE_DIR}/" "" __target_path ${__target_path})
-  string(FIND "${__target_path}" "fluid" pos)
-  if(pos GREATER 1)
-    get_property(fluid_modules GLOBAL PROPERTY FLUID_MODULES)
-    set(fluid_modules ${fluid_modules} ${TARGET_NAME})
-    set_property(GLOBAL PROPERTY FLUID_MODULES "${fluid_modules}")
-  endif()
-endfunction(find_fluid_modules)
-
 # make package for paddle fluid shared and static library
 function(copy TARGET)
    set(options "")
@@ -39,7 +26,7 @@ function(copy TARGET)
        message(FATAL_ERROR "${TARGET} source numbers are not equal to destination numbers")
    endif()
    math(EXPR len "${copy_lib_SRCS_len} - 1")
-    
+
    add_custom_target(${TARGET} DEPENDS ${copy_lib_DEPS})
    foreach(index RANGE ${len})
        list(GET copy_lib_SRCS ${index} src)
@@ -149,8 +136,29 @@ copy(memory_lib
  DSTS ${dst_dir}/${module} ${dst_dir}/${module}/detail
 )

+set(inference_deps paddle_fluid_shared paddle_fluid)
+
+if(WITH_CONTRIB)
+    message(STATUS "installing contrib")
+    set(contrib_dst_dir "${FLUID_INSTALL_DIR}/contrib/inference")
+    if (WITH_ANAKIN AND WITH_GPU)
+        copy(contrib_anakin_inference_lib DEPS paddle_inference_api inference_anakin_api
+            SRCS
+            ${PADDLE_BINARY_DIR}/paddle/contrib/inference/libinference_anakin_api* # compiled anakin api
+            ${PADDLE_BINARY_DIR}/third_party/install/anakin/*.tar.gz # anakin release
+            DSTS ${contrib_dst_dir}/anakin ${contrib_dst_dir}/anakin)
+        list(APPEND inference_deps contrib_anakin_inference_lib)
+   endif()
+
+  copy(contrib_inference_lib DEPS paddle_inference_api paddle_inference_api_shared
+        SRCS ${PADDLE_SOURCE_DIR}/paddle/contrib/inference/paddle_inference_api.h
+        ${PADDLE_BINARY_DIR}/paddle/contrib/inference/libpaddle_inference_api*
+        DSTS ${contrib_dst_dir} ${contrib_dst_dir})
+  list(APPEND inference_deps contrib_inference_lib)
+endif()
+
 set(module "inference")
-copy(inference_lib DEPS paddle_fluid_shared paddle_fluid
+copy(inference_lib DEPS ${inference_deps}
  SRCS ${src_dir}/${module}/*.h ${PADDLE_BINARY_DIR}/paddle/fluid/inference/libpaddle_fluid.*
  DSTS ${dst_dir}/${module} ${dst_dir}/${module}
 )

--- a/doc/about/about_us.rst
+++ b/doc/about/about_us.rst
+=========
+关于我们
+=========
+
+什么是PaddlePaddle
+--------------------
+
+- PaddlePaddle是百度自主研发并开源的深度学习框架，它能够让开发者和企业安全、快速地实现自己的AI想法
+
+- 项目团队汇聚了全球顶级的深度学习科学家，致力于为开发者和企业提供最好的深度学习研发体验
+
+- 框架具有易学、易用、安全、高效四大特性，是最适合中国开发者和企业的深度学习工具
+
+PaddlePaddle的技术特色
+-------------------------
+
+- 新一代深度学习框架： PaddlePaddle是基于“深度学习编程语言”的新一代深度学习框架，在保证性能的同时，极大的提升了框架对模型的表达能力，能够描述任意潜在可能出现的模型
+
+- 对大规模计算更加友好：经过百度内多种大规模计算业务的打磨，PaddlePaddle在分布式计算上表现优异，基于EDL技术能够节约大量计算资源，同时也能支持大规模稀疏模型的训练
+
+- 提供可视化的深度学习：通过Visual DL可以帮助开发者方便的观测训练整体趋势、数据样本质量和中间结果、参数分布和变化趋势、以及模型的结构，帮助开发者更便捷的完成编程过程
+
+提供基于PaddlePaddle的教育体系
+--------------------------------
+
+- 深度学习课程：百度与中国市场顶级的教育、培训机构共同开发了深度学习精品课程以及学习教材，帮助开发者从零掌握深度学习
+
+- 深度学习实训：对于目的是科研和学习的用户，PaddlePaddle提供了无需安装、线上运行的开发环境，并提供算法、算力、数据支持
+
+- 线下培训：提供丰富、高质量的线下教育活动，如青年教师培训、线下实战营、沙龙等多种形式的培训和交流
+
+
+提供基于PaddlePaddle的AI服务
+------------------------------
+
+- EadyDL：可以帮助零算法基础的企业快速完成一个深度学习任务，只需少量的数据即可得到优质的模型
+
+- AI市场：提供标准化的AI 能力、产品的交易机制，帮助企业快速找到所需，有效开展AI业务
+
+- 深度学习竞赛： PaddlePaddle汇聚顶尖深度学习开发者，企业可以发布自己的商业问题，通过竞赛方式快速找到最优的解决方案
+
+你对PaddlePaddle有任何的问题都可以通过以下方式联系到我们
+-----------------------------------------------------------
+
+- 学习/使用问题：可以在 `PaddlePaddle开源社区 <https://github.com/PaddlePaddle/Paddle/issues>`_，以及 `PaddlePaddle中文社区 <http://ai.baidu.com/forum/topic/list/168>`_ 向我们反馈
+
+- 对PaddlePaddle框架发展的建议：可发送邮件至Paddle-better@baidu.com
+
+我们期待与你一起打造世界顶级深度学习框架，共同推动AI技术的进步
+
+
+
+PaddlePaddle团队
--- a/doc/fluid/api/average.rst
+++ b/doc/fluid/api/average.rst
+..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
+    !DO NOT EDIT THIS FILE MANUALLY!
+
+=============
+fluid.average
+=============
+
+.. _api_fluid_average_WeightedAverage:
+
+WeightedAverage
+---------------
+
+..  autoclass:: paddle.fluid.average.WeightedAverage
+    :members:
+    :noindex:
+
--- a/doc/fluid/api/backward.rst
+++ b/doc/fluid/api/backward.rst
+..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
+    !DO NOT EDIT THIS FILE MANUALLY!
+
+==============
+fluid.backward
+==============
+
+.. _api_fluid_backward_append_backward:
+
+append_backward
+---------------
+
+..  autofunction:: paddle.fluid.backward.append_backward
+    :noindex:
+
+.. _api_fluid_backward_calc_gradient:
+
+calc_gradient
+-------------
+
+..  autofunction:: paddle.fluid.backward.calc_gradient
+    :noindex:
+
--- a/doc/fluid/api/clip.rst
+++ b/doc/fluid/api/clip.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-====
-clip
-====
+==========
+fluid.clip
+==========
+
+.. _api_fluid_clip_ErrorClipByValue:

 ErrorClipByValue
 ----------------
@@ -12,6 +14,8 @@ ErrorClipByValue
    :members:
    :noindex:

+.. _api_fluid_clip_GradientClipByValue:
+
 GradientClipByValue
 -------------------

@@ -19,6 +23,8 @@ GradientClipByValue
    :members:
    :noindex:

+.. _api_fluid_clip_GradientClipByNorm:
+
 GradientClipByNorm
 ------------------

@@ -26,6 +32,8 @@ GradientClipByNorm
    :members:
    :noindex:

+.. _api_fluid_clip_GradientClipByGlobalNorm:
+
 GradientClipByGlobalNorm
 ------------------------

@@ -33,15 +41,3 @@ GradientClipByGlobalNorm
    :members:
    :noindex:

-append_gradient_clip_ops
------------------------
-
-..  autofunction:: paddle.fluid.clip.append_gradient_clip_ops
-    :noindex:
-
-error_clip_callback
-------------------
-
-..  autofunction:: paddle.fluid.clip.error_clip_callback
-    :noindex:
-
--- a/doc/fluid/api/data.rst
+++ b/doc/fluid/api/data.rst
-==================================
-Data Reader Interface and DataSets
-==================================
-
-..  toctree::
-    :maxdepth: 1
-
-    data/data_reader.rst
-    data/image.rst
-    data/dataset.rst
--- a/doc/fluid/api/data_feeder.rst
+++ b/doc/fluid/api/data_feeder.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-===========
-data_feeder
-===========
+=================
+fluid.data_feeder
+=================
+
+.. _api_fluid_data_feeder_DataFeeder:

 DataFeeder
 ----------

--- a/doc/fluid/api/evaluator.rst
+++ b/doc/fluid/api/evaluator.rst
-..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
-    !DO NOT EDIT THIS FILE MANUALLY!
-
-=========
-evaluator
-=========
-
--- a/doc/fluid/api/executor.rst
+++ b/doc/fluid/api/executor.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-========
-executor
-========
+==============
+fluid.executor
+==============
+
+.. _api_fluid_executor_Executor:

 Executor
 --------
@@ -12,24 +14,32 @@ Executor
    :members:
    :noindex:

+.. _api_fluid_executor_global_scope:
+
 global_scope
 ------------

 ..  autofunction:: paddle.fluid.executor.global_scope
    :noindex:

+.. _api_fluid_executor_scope_guard:
+
 scope_guard
 -----------

 ..  autofunction:: paddle.fluid.executor.scope_guard
    :noindex:

-switch_scope
------------
+.. _api_fluid_executor__switch_scope:
+
+_switch_scope
+-------------

-..  autofunction:: paddle.fluid.executor.switch_scope
+..  autofunction:: paddle.fluid.executor._switch_scope
    :noindex:

+.. _api_fluid_executor_fetch_var:
+
 fetch_var
 ---------


--- a/doc/fluid/api/fluid.rst
+++ b/doc/fluid/api/fluid.rst
+..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
+    !DO NOT EDIT THIS FILE MANUALLY!
+
+=====
+fluid
+=====
+
+.. _api_fluid_Block:
+
+Block
+-----
+
+..  autoclass:: paddle.fluid.Block
+    :members:
+    :noindex:
+
+.. _api_fluid_Variable:
+
+Variable
+--------
+
+..  autoclass:: paddle.fluid.Variable
+    :members:
+    :noindex:
+
+.. _api_fluid_Program:
+
+Program
+-------
+
+..  autoclass:: paddle.fluid.Program
+    :members:
+    :noindex:
+
+.. _api_fluid_Operator:
+
+Operator
+--------
+
+..  autoclass:: paddle.fluid.Operator
+    :members:
+    :noindex:
+
+.. _api_fluid_default_startup_program:
+
+default_startup_program
+-----------------------
+
+..  autofunction:: paddle.fluid.default_startup_program
+    :noindex:
+
+.. _api_fluid_default_main_program:
+
+default_main_program
+--------------------
+
+..  autofunction:: paddle.fluid.default_main_program
+    :noindex:
+
+.. _api_fluid_program_guard:
+
+program_guard
+-------------
+
+..  autofunction:: paddle.fluid.program_guard
+    :noindex:
+
+.. _api_fluid_get_var:
+
+get_var
+-------
+
+..  autofunction:: paddle.fluid.get_var
+    :noindex:
+
+.. _api_fluid_Executor:
+
+Executor
+--------
+
+..  autoclass:: paddle.fluid.Executor
+    :members:
+    :noindex:
+
+.. _api_fluid_global_scope:
+
+global_scope
+------------
+
+..  autofunction:: paddle.fluid.global_scope
+    :noindex:
+
+.. _api_fluid_scope_guard:
+
+scope_guard
+-----------
+
+..  autofunction:: paddle.fluid.scope_guard
+    :noindex:
+
+.. _api_fluid__switch_scope:
+
+_switch_scope
+-------------
+
+..  autofunction:: paddle.fluid._switch_scope
+    :noindex:
+
+.. _api_fluid_fetch_var:
+
+fetch_var
+---------
+
+..  autofunction:: paddle.fluid.fetch_var
+    :noindex:
+
+.. _api_fluid_Go:
+
+Go
+--
+
+..  autoclass:: paddle.fluid.Go
+    :members:
+    :noindex:
+
+.. _api_fluid_make_channel:
+
+make_channel
+------------
+
+..  autofunction:: paddle.fluid.make_channel
+    :noindex:
+
+.. _api_fluid_channel_send:
+
+channel_send
+------------
+
+..  autofunction:: paddle.fluid.channel_send
+    :noindex:
+
+.. _api_fluid_channel_recv:
+
+channel_recv
+------------
+
+..  autofunction:: paddle.fluid.channel_recv
+    :noindex:
+
+.. _api_fluid_channel_close:
+
+channel_close
+-------------
+
+..  autofunction:: paddle.fluid.channel_close
+    :noindex:
+
+.. _api_fluid_Select:
+
+Select
+------
+
+..  autoclass:: paddle.fluid.Select
+    :members:
+    :noindex:
+
+.. _api_fluid_Trainer:
+
+Trainer
+-------
+
+..  autoclass:: paddle.fluid.Trainer
+    :members:
+    :noindex:
+
+.. _api_fluid_BeginEpochEvent:
+
+BeginEpochEvent
+---------------
+
+..  autoclass:: paddle.fluid.BeginEpochEvent
+    :members:
+    :noindex:
+
+.. _api_fluid_EndEpochEvent:
+
+EndEpochEvent
+-------------
+
+..  autoclass:: paddle.fluid.EndEpochEvent
+    :members:
+    :noindex:
+
+.. _api_fluid_BeginStepEvent:
+
+BeginStepEvent
+--------------
+
+..  autoclass:: paddle.fluid.BeginStepEvent
+    :members:
+    :noindex:
+
+.. _api_fluid_EndStepEvent:
+
+EndStepEvent
+------------
+
+..  autoclass:: paddle.fluid.EndStepEvent
+    :members:
+    :noindex:
+
+.. _api_fluid_CheckpointConfig:
+
+CheckpointConfig
+----------------
+
+..  autoclass:: paddle.fluid.CheckpointConfig
+    :members:
+    :noindex:
+
+.. _api_fluid_Inferencer:
+
+Inferencer
+----------
+
+..  autoclass:: paddle.fluid.Inferencer
+    :members:
+    :noindex:
+
+.. _api_fluid_DistributeTranspiler:
+
+DistributeTranspiler
+--------------------
+
+..  autoclass:: paddle.fluid.DistributeTranspiler
+    :members:
+    :noindex:
+
+.. _api_fluid_memory_optimize:
+
+memory_optimize
+---------------
+
+..  autofunction:: paddle.fluid.memory_optimize
+    :noindex:
+
+.. _api_fluid_release_memory:
+
+release_memory
+--------------
+
+..  autofunction:: paddle.fluid.release_memory
+    :noindex:
+
+.. _api_fluid_ParallelExecutor:
+
+ParallelExecutor
+----------------
+
+..  autoclass:: paddle.fluid.ParallelExecutor
+    :members:
+    :noindex:
+
+.. _api_fluid_ExecutionStrategy:
+
+ExecutionStrategy
+-----------------
+
+..  autoclass:: paddle.fluid.ExecutionStrategy
+    :members:
+    :noindex:
+
+.. _api_fluid_BuildStrategy:
+
+BuildStrategy
+-------------
+
+..  autoclass:: paddle.fluid.BuildStrategy
+    :members:
+    :noindex:
+
+.. _api_fluid_create_lod_tensor:
+
+create_lod_tensor
+-----------------
+
+..  autofunction:: paddle.fluid.create_lod_tensor
+    :noindex:
+
+.. _api_fluid_create_random_int_lodtensor:
+
+create_random_int_lodtensor
+---------------------------
+
+..  autofunction:: paddle.fluid.create_random_int_lodtensor
+    :noindex:
+
+.. _api_fluid_LoDTensor:
+
+LoDTensor
+---------
+
+..  autoclass:: paddle.fluid.LoDTensor
+    :members:
+    :noindex:
+
+.. _api_fluid_CPUPlace:
+
+CPUPlace
+--------
+
+..  autoclass:: paddle.fluid.CPUPlace
+    :members:
+    :noindex:
+
+.. _api_fluid_CUDAPlace:
+
+CUDAPlace
+---------
+
+..  autoclass:: paddle.fluid.CUDAPlace
+    :members:
+    :noindex:
+
+.. _api_fluid_CUDAPinnedPlace:
+
+CUDAPinnedPlace
+---------------
+
+..  autoclass:: paddle.fluid.CUDAPinnedPlace
+    :members:
+    :noindex:
+
+.. _api_fluid_Tensor:
+
+Tensor
+------
+
+..  autoclass:: paddle.fluid.Tensor
+    :members:
+    :noindex:
+
+.. _api_fluid_ParamAttr:
+
+ParamAttr
+---------
+
+..  autoclass:: paddle.fluid.ParamAttr
+    :members:
+    :noindex:
+
+.. _api_fluid_WeightNormParamAttr:
+
+WeightNormParamAttr
+-------------------
+
+..  autoclass:: paddle.fluid.WeightNormParamAttr
+    :members:
+    :noindex:
+
+.. _api_fluid_DataFeeder:
+
+DataFeeder
+----------
+
+..  autoclass:: paddle.fluid.DataFeeder
+    :members:
+    :noindex:
+
+.. _api_fluid_Scope:
+
+Scope
+-----
+
+..  autoclass:: paddle.fluid.Scope
+    :members:
+    :noindex:
+
--- a/doc/fluid/api/gen_doc.py
+++ b/doc/fluid/api/gen_doc.py
@@ -29,19 +29,27 @@ def parse_arg():


 class DocGenerator(object):
-    def __init__(self, module_name, stream=sys.stdout):
+    def __init__(self, module_name=None, stream=sys.stdout):
+        if module_name == "":
+            module_name = None
        self.stream = stream
-        self.module_name = module_name
-        if not hasattr(fluid, module_name):
-            raise ValueError("Cannot find fluid.{0}".format(module_name))
+        if module_name is None:
+            self.module_name = "fluid"
        else:
-            self.module = getattr(fluid, module_name)
+            self.module_name = "fluid." + module_name
+        if module_name is None:
+            self.module = fluid
+        else:
+            if not hasattr(fluid, module_name):
+                raise ValueError("Cannot find fluid.{0}".format(module_name))
+            else:
+                self.module = getattr(fluid, module_name)
        self.stream.write('''..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

 ''')

-        self._print_header_(module_name, dot='=', is_title=True)
+        self._print_header_(self.module_name, dot='=', is_title=True)

    def print_submodule(self, submodule_name):
        submodule = getattr(self.module, submodule_name)
@@ -60,25 +68,29 @@ class DocGenerator(object):
        self._print_header_(name, dot='=', is_title=False)

    def print_item(self, name):
-        item = getattr(self.module, name)
+        item = getattr(self.module, name, None)
+        if item is None:
+            return
        if isinstance(item, types.TypeType):
            self.print_class(name)
        elif isinstance(item, types.FunctionType):
            self.print_method(name)
        else:
-            raise RuntimeError("Unsupported item {0}".format(name))
+            pass

    def print_class(self, name):
+        self._print_ref_(name)
        self._print_header_(name, dot='-', is_title=False)
-        self.stream.write('''..  autoclass:: paddle.fluid.{0}.{1}
+        self.stream.write('''..  autoclass:: paddle.{0}.{1}
    :members:
    :noindex:

 '''.format(self.module_name, name))

    def print_method(self, name):
+        self._print_ref_(name)
        self._print_header_(name, dot='-', is_title=False)
-        self.stream.write('''..  autofunction:: paddle.fluid.{0}.{1}
+        self.stream.write('''..  autofunction:: paddle.{0}.{1}
    :noindex:

 '''.format(self.module_name, name))
@@ -94,6 +106,10 @@ class DocGenerator(object):
        self.stream.write('\n')
        self.stream.write('\n')

+    def _print_ref_(self, name):
+        self.stream.write(".. _api_{0}_{1}:\n\n".format("_".join(
+            self.module_name.split(".")), name))
+

 def main():
    args = parse_arg()

--- a/doc/fluid/api/gen_doc.sh
+++ b/doc/fluid/api/gen_doc.sh
 #!/bin/bash
-python gen_doc.py layers --submodules control_flow device io nn ops tensor > layers.rst
+python gen_doc.py layers --submodules control_flow device io nn ops tensor learning_rate_scheduler detection metric_op tensor > layers.rst

-for module in data_feeder clip metrics executor initializer io nets optimizer param_attr profiler regularizer
+for module in data_feeder clip metrics executor initializer io nets optimizer param_attr profiler regularizer transpiler recordio_writer backward average profiler
 do
  python gen_doc.py ${module} > ${module}.rst
 done
+
+python gen_doc.py "" > fluid.rst
--- a/doc/fluid/api/index_en.rst
+++ b/doc/fluid/api/index_en.rst
-======================
-Fluid
-======================
+=============
+API Reference
+=============

 ..  toctree::
    :maxdepth: 1

+    fluid.rst
    layers.rst
    data_feeder.rst
    executor.rst
@@ -18,3 +19,8 @@ Fluid
    regularizer.rst
    io.rst
    data.rst
+    transpiler.rst
+    recordio_writer.rst
+    backward.rst
+    average.rst
+    profiler.rst
--- a/doc/fluid/api/initializer.rst
+++ b/doc/fluid/api/initializer.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-===========
-initializer
-===========
+=================
+fluid.initializer
+=================
+
+.. _api_fluid_initializer_Constant:

 Constant
 --------
@@ -12,6 +14,8 @@ Constant
    :members:
    :noindex:

+.. _api_fluid_initializer_Uniform:
+
 Uniform
 -------

@@ -19,6 +23,8 @@ Uniform
    :members:
    :noindex:

+.. _api_fluid_initializer_Normal:
+
 Normal
 ------

@@ -26,6 +32,8 @@ Normal
    :members:
    :noindex:

+.. _api_fluid_initializer_Xavier:
+
 Xavier
 ------

@@ -33,18 +41,42 @@ Xavier
    :members:
    :noindex:

+.. _api_fluid_initializer_Bilinear:
+
+Bilinear
+--------
+
+..  autoclass:: paddle.fluid.initializer.Bilinear
+    :members:
+    :noindex:
+
+.. _api_fluid_initializer_MSRA:
+
+MSRA
+----
+
+..  autoclass:: paddle.fluid.initializer.MSRA
+    :members:
+    :noindex:
+
+.. _api_fluid_initializer_force_init_on_cpu:
+
 force_init_on_cpu
 -----------------

 ..  autofunction:: paddle.fluid.initializer.force_init_on_cpu
    :noindex:

+.. _api_fluid_initializer_init_on_cpu:
+
 init_on_cpu
 -----------

 ..  autofunction:: paddle.fluid.initializer.init_on_cpu
    :noindex:

+.. _api_fluid_initializer_ConstantInitializer:
+
 ConstantInitializer
 -------------------

@@ -52,6 +84,8 @@ ConstantInitializer
    :members:
    :noindex:

+.. _api_fluid_initializer_UniformInitializer:
+
 UniformInitializer
 ------------------

@@ -59,6 +93,8 @@ UniformInitializer
    :members:
    :noindex:

+.. _api_fluid_initializer_NormalInitializer:
+
 NormalInitializer
 -----------------

@@ -66,6 +102,8 @@ NormalInitializer
    :members:
    :noindex:

+.. _api_fluid_initializer_XavierInitializer:
+
 XavierInitializer
 -----------------

@@ -73,3 +111,21 @@ XavierInitializer
    :members:
    :noindex:

+.. _api_fluid_initializer_BilinearInitializer:
+
+BilinearInitializer
+-------------------
+
+..  autoclass:: paddle.fluid.initializer.BilinearInitializer
+    :members:
+    :noindex:
+
+.. _api_fluid_initializer_MSRAInitializer:
+
+MSRAInitializer
+---------------
+
+..  autoclass:: paddle.fluid.initializer.MSRAInitializer
+    :members:
+    :noindex:
+
--- a/doc/fluid/api/io.rst
+++ b/doc/fluid/api/io.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-==
-io
-==
+========
+fluid.io
+========
+
+.. _api_fluid_io_save_vars:

 save_vars
 ---------
@@ -11,69 +13,115 @@ save_vars
 ..  autofunction:: paddle.fluid.io.save_vars
    :noindex:

+.. _api_fluid_io_save_params:
+
 save_params
 -----------

 ..  autofunction:: paddle.fluid.io.save_params
    :noindex:

+.. _api_fluid_io_save_persistables:
+
 save_persistables
 -----------------

 ..  autofunction:: paddle.fluid.io.save_persistables
    :noindex:

+.. _api_fluid_io_load_vars:
+
 load_vars
 ---------

 ..  autofunction:: paddle.fluid.io.load_vars
    :noindex:

+.. _api_fluid_io_load_params:
+
 load_params
 -----------

 ..  autofunction:: paddle.fluid.io.load_params
    :noindex:

+.. _api_fluid_io_load_persistables:
+
 load_persistables
 -----------------

 ..  autofunction:: paddle.fluid.io.load_persistables
    :noindex:

+.. _api_fluid_io_save_inference_model:
+
 save_inference_model
 --------------------

 ..  autofunction:: paddle.fluid.io.save_inference_model
    :noindex:

+.. _api_fluid_io_load_inference_model:
+
 load_inference_model
 --------------------

 ..  autofunction:: paddle.fluid.io.load_inference_model
    :noindex:

+.. _api_fluid_io_get_inference_program:
+
 get_inference_program
 ---------------------

 ..  autofunction:: paddle.fluid.io.get_inference_program
    :noindex:

+.. _api_fluid_io_save_checkpoint:
+
 save_checkpoint
 ---------------

 ..  autofunction:: paddle.fluid.io.save_checkpoint
    :noindex:

+.. _api_fluid_io_load_checkpoint:
+
 load_checkpoint
 ---------------

 ..  autofunction:: paddle.fluid.io.load_checkpoint
    :noindex:

+.. _api_fluid_io_clean_checkpoint:
+
 clean_checkpoint
 ----------------

 ..  autofunction:: paddle.fluid.io.clean_checkpoint
    :noindex:

+.. _api_fluid_io_load_persist_vars_without_grad:
+
+load_persist_vars_without_grad
+------------------------------
+
+..  autofunction:: paddle.fluid.io.load_persist_vars_without_grad
+    :noindex:
+
+.. _api_fluid_io_save_persist_vars_without_grad:
+
+save_persist_vars_without_grad
+------------------------------
+
+..  autofunction:: paddle.fluid.io.save_persist_vars_without_grad
+    :noindex:
+
+.. _api_fluid_io_get_latest_checkpoint_serial:
+
+get_latest_checkpoint_serial
+----------------------------
+
+..  autofunction:: paddle.fluid.io.get_latest_checkpoint_serial
+    :noindex:
+
--- a/doc/fluid/api/layers.rst
+++ b/doc/fluid/api/layers.rst
--- a/doc/fluid/api/metrics.rst
+++ b/doc/fluid/api/metrics.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-=======
-metrics
-=======
+=============
+fluid.metrics
+=============
+
+.. _api_fluid_metrics_MetricBase:

 MetricBase
 ----------
@@ -12,6 +14,8 @@ MetricBase
    :members:
    :noindex:

+.. _api_fluid_metrics_CompositeMetric:
+
 CompositeMetric
 ---------------

@@ -19,6 +23,26 @@ CompositeMetric
    :members:
    :noindex:

+.. _api_fluid_metrics_Precision:
+
+Precision
+---------
+
+..  autoclass:: paddle.fluid.metrics.Precision
+    :members:
+    :noindex:
+
+.. _api_fluid_metrics_Recall:
+
+Recall
+------
+
+..  autoclass:: paddle.fluid.metrics.Recall
+    :members:
+    :noindex:
+
+.. _api_fluid_metrics_Accuracy:
+
 Accuracy
 --------

@@ -26,6 +50,8 @@ Accuracy
    :members:
    :noindex:

+.. _api_fluid_metrics_ChunkEvaluator:
+
 ChunkEvaluator
 --------------

@@ -33,6 +59,8 @@ ChunkEvaluator
    :members:
    :noindex:

+.. _api_fluid_metrics_EditDistance:
+
 EditDistance
 ------------

@@ -40,6 +68,8 @@ EditDistance
    :members:
    :noindex:

+.. _api_fluid_metrics_DetectionMAP:
+
 DetectionMAP
 ------------

@@ -47,6 +77,8 @@ DetectionMAP
    :members:
    :noindex:

+.. _api_fluid_metrics_Auc:
+
 Auc
 ---


--- a/doc/fluid/api/nets.rst
+++ b/doc/fluid/api/nets.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-====
-nets
-====
+==========
+fluid.nets
+==========
+
+.. _api_fluid_nets_simple_img_conv_pool:

 simple_img_conv_pool
 --------------------
@@ -11,18 +13,24 @@ simple_img_conv_pool
 ..  autofunction:: paddle.fluid.nets.simple_img_conv_pool
    :noindex:

+.. _api_fluid_nets_sequence_conv_pool:
+
 sequence_conv_pool
 ------------------

 ..  autofunction:: paddle.fluid.nets.sequence_conv_pool
    :noindex:

+.. _api_fluid_nets_glu:
+
 glu
 ---

 ..  autofunction:: paddle.fluid.nets.glu
    :noindex:

+.. _api_fluid_nets_scaled_dot_product_attention:
+
 scaled_dot_product_attention
 ----------------------------


--- a/doc/fluid/api/optimizer.rst
+++ b/doc/fluid/api/optimizer.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-=========
-optimizer
-=========
+===============
+fluid.optimizer
+===============
+
+.. _api_fluid_optimizer_SGD:

 SGD
 ---
@@ -12,6 +14,8 @@ SGD
    :members:
    :noindex:

+.. _api_fluid_optimizer_Momentum:
+
 Momentum
 --------

@@ -19,6 +23,8 @@ Momentum
    :members:
    :noindex:

+.. _api_fluid_optimizer_Adagrad:
+
 Adagrad
 -------

@@ -26,6 +32,8 @@ Adagrad
    :members:
    :noindex:

+.. _api_fluid_optimizer_Adam:
+
 Adam
 ----

@@ -33,6 +41,8 @@ Adam
    :members:
    :noindex:

+.. _api_fluid_optimizer_Adamax:
+
 Adamax
 ------

@@ -40,6 +50,8 @@ Adamax
    :members:
    :noindex:

+.. _api_fluid_optimizer_DecayedAdagrad:
+
 DecayedAdagrad
 --------------

@@ -47,6 +59,17 @@ DecayedAdagrad
    :members:
    :noindex:

+.. _api_fluid_optimizer_Ftrl:
+
+Ftrl
+----
+
+..  autoclass:: paddle.fluid.optimizer.Ftrl
+    :members:
+    :noindex:
+
+.. _api_fluid_optimizer_SGDOptimizer:
+
 SGDOptimizer
 ------------

@@ -54,6 +77,8 @@ SGDOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_MomentumOptimizer:
+
 MomentumOptimizer
 -----------------

@@ -61,6 +86,8 @@ MomentumOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_AdagradOptimizer:
+
 AdagradOptimizer
 ----------------

@@ -68,6 +95,8 @@ AdagradOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_AdamOptimizer:
+
 AdamOptimizer
 -------------

@@ -75,6 +104,8 @@ AdamOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_AdamaxOptimizer:
+
 AdamaxOptimizer
 ---------------

@@ -82,6 +113,8 @@ AdamaxOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_DecayedAdagradOptimizer:
+
 DecayedAdagradOptimizer
 -----------------------

@@ -89,6 +122,8 @@ DecayedAdagradOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_RMSPropOptimizer:
+
 RMSPropOptimizer
 ----------------

@@ -96,6 +131,17 @@ RMSPropOptimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_FtrlOptimizer:
+
+FtrlOptimizer
+-------------
+
+..  autoclass:: paddle.fluid.optimizer.FtrlOptimizer
+    :members:
+    :noindex:
+
+.. _api_fluid_optimizer_Adadelta:
+
 Adadelta
 --------

@@ -103,6 +149,8 @@ Adadelta
    :members:
    :noindex:

+.. _api_fluid_optimizer_ModelAverage:
+
 ModelAverage
 ------------

@@ -110,6 +158,8 @@ ModelAverage
    :members:
    :noindex:

+.. _api_fluid_optimizer_Optimizer:
+
 Optimizer
 ---------

@@ -117,3 +167,12 @@ Optimizer
    :members:
    :noindex:

+.. _api_fluid_optimizer_RMSPropOptimizer:
+
+RMSPropOptimizer
+----------------
+
+..  autoclass:: paddle.fluid.optimizer.RMSPropOptimizer
+    :members:
+    :noindex:
+
--- a/doc/fluid/api/param_attr.rst
+++ b/doc/fluid/api/param_attr.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-==========
-param_attr
-==========
+================
+fluid.param_attr
+================
+
+.. _api_fluid_param_attr_ParamAttr:

 ParamAttr
 ---------
@@ -12,6 +14,8 @@ ParamAttr
    :members:
    :noindex:

+.. _api_fluid_param_attr_WeightNormParamAttr:
+
 WeightNormParamAttr
 -------------------


--- a/doc/fluid/api/profiler.rst
+++ b/doc/fluid/api/profiler.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-========
-profiler
-========
+==============
+fluid.profiler
+==============
+
+.. _api_fluid_profiler_cuda_profiler:

 cuda_profiler
 -------------
@@ -11,24 +13,32 @@ cuda_profiler
 ..  autofunction:: paddle.fluid.profiler.cuda_profiler
    :noindex:

+.. _api_fluid_profiler_reset_profiler:
+
 reset_profiler
 --------------

 ..  autofunction:: paddle.fluid.profiler.reset_profiler
    :noindex:

+.. _api_fluid_profiler_profiler:
+
 profiler
 --------

 ..  autofunction:: paddle.fluid.profiler.profiler
    :noindex:

+.. _api_fluid_profiler_start_profiler:
+
 start_profiler
 --------------

 ..  autofunction:: paddle.fluid.profiler.start_profiler
    :noindex:

+.. _api_fluid_profiler_stop_profiler:
+
 stop_profiler
 -------------


--- a/doc/fluid/api/recordio_writer.rst
+++ b/doc/fluid/api/recordio_writer.rst
+..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
+    !DO NOT EDIT THIS FILE MANUALLY!
+
+=====================
+fluid.recordio_writer
+=====================
+
+.. _api_fluid_recordio_writer_convert_reader_to_recordio_file:
+
+convert_reader_to_recordio_file
+-------------------------------
+
+..  autofunction:: paddle.fluid.recordio_writer.convert_reader_to_recordio_file
+    :noindex:
+
+.. _api_fluid_recordio_writer_convert_reader_to_recordio_files:
+
+convert_reader_to_recordio_files
+--------------------------------
+
+..  autofunction:: paddle.fluid.recordio_writer.convert_reader_to_recordio_files
+    :noindex:
+
--- a/doc/fluid/api/regularizer.rst
+++ b/doc/fluid/api/regularizer.rst
 ..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
    !DO NOT EDIT THIS FILE MANUALLY!

-===========
-regularizer
-===========
+=================
+fluid.regularizer
+=================
+
+.. _api_fluid_regularizer_append_regularization_ops:

 append_regularization_ops
 -------------------------
@@ -11,12 +13,7 @@ append_regularization_ops
 ..  autofunction:: paddle.fluid.regularizer.append_regularization_ops
    :noindex:

-WeightDecayRegularizer
----------------------
-
-..  autoclass:: paddle.fluid.regularizer.WeightDecayRegularizer
-    :members:
-    :noindex:
+.. _api_fluid_regularizer_L1Decay:

 L1Decay
 -------
@@ -25,6 +22,8 @@ L1Decay
    :members:
    :noindex:

+.. _api_fluid_regularizer_L2Decay:
+
 L2Decay
 -------

@@ -32,6 +31,8 @@ L2Decay
    :members:
    :noindex:

+.. _api_fluid_regularizer_L1DecayRegularizer:
+
 L1DecayRegularizer
 ------------------

@@ -39,6 +40,8 @@ L1DecayRegularizer
    :members:
    :noindex:

+.. _api_fluid_regularizer_L2DecayRegularizer:
+
 L2DecayRegularizer
 ------------------


--- a/doc/fluid/api/transpiler.rst
+++ b/doc/fluid/api/transpiler.rst
+..  THIS FILE IS GENERATED BY `gen_doc.{py|sh}`
+    !DO NOT EDIT THIS FILE MANUALLY!
+
+================
+fluid.transpiler
+================
+
+.. _api_fluid_transpiler_DistributeTranspiler:
+
+DistributeTranspiler
+--------------------
+
+..  autoclass:: paddle.fluid.transpiler.DistributeTranspiler
+    :members:
+    :noindex:
+
+.. _api_fluid_transpiler_memory_optimize:
+
+memory_optimize
+---------------
+
+..  autofunction:: paddle.fluid.transpiler.memory_optimize
+    :noindex:
+
+.. _api_fluid_transpiler_release_memory:
+
+release_memory
+--------------
+
+..  autofunction:: paddle.fluid.transpiler.release_memory
+    :noindex:
+
+.. _api_fluid_transpiler_HashName:
+
+HashName
+--------
+
+..  autoclass:: paddle.fluid.transpiler.HashName
+    :members:
+    :noindex:
+
+.. _api_fluid_transpiler_RoundRobin:
+
+RoundRobin
+----------
+
+..  autoclass:: paddle.fluid.transpiler.RoundRobin
+    :members:
+    :noindex:
+
--- a/doc/fluid/design/concepts/lod_tensor.md
+++ b/doc/fluid/design/concepts/lod_tensor.md
@@ -173,6 +173,7 @@ are transformed into offsets of elements/words as follows:

 ## Slicing of LoD Tensors

+
 When we use the above 2-level LoD Tensor as the input to a nested-RNN, we need to retrieve certain sequences.  Here we define the sequence identified by branch <i,j,...> as the **<i,j,...>-slice**.

 For example, the <2>-slice of above example is
@@ -189,3 +190,22 @@ and the <2,0>-slice of above slice is
 10  12
  ||
 ```
+
+## Length Representation vs Offset Representation
+
+The offset representation is an implementation-oriented decision and it makes understanding the idea behind LoDTensor difficult.
+Hence, we encapsulate this implementation detail in C++ and expose the original length representation in our Python API. 
+Specifically, we call this length representation `recursive_sequence_lengths` and users can use the following code to set or get the `recursive_sequence_lengths` of a LoDTensor in Python:
+```Python
+# length representation of lod called recursive_sequence_lengths
+recursive_seq_lens = [[3, 1, 2], [2, 2, 1, 3, 1, 2]]
+# Create a LoDTensor that has the above recursive_sequence_lengths info.
+# This recursive_sequence_lengths will be converted to an offset representation of LoD in the C++ implementation under the hood.
+tensor = fluid.LoDTensor(lod)
+
+# Set/Change the recursive_sequence_lengths info of LoDTensor
+tensor.set_recursive_sequence_lengths([[3, 1, 2]])
+# Get the recursive_sequence_lengths info of a LoDTensor (the offset-based LoD representation stored in C++ will be converted 
+# back to length-based recursive_sequence_lengths), new_recursive_seq_lens = [[3, 1, 2]]
+new_recursive_seq_lens = tensor.recursive_sequence_lengths()
+```
--- a/doc/fluid/design/concepts/python_data_feeding.md
+++ b/doc/fluid/design/concepts/python_data_feeding.md
+# Python Data Feeding
+
+In the former implementation of Paddle Fluid, there are two ways to feed data:
+
+- Use `reader_op` in backend C++ side. This method only supports data feeding from recordio files and random data generators, but supports many kinds of `decorated_readers`. For examples, `double_buffer_reader` uses two threads to achieve better performance: one for time-consuming I/O operations, and the other for `Executor::Run()`. See [C++ Data Feeding](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/fluid/design/concepts/cpp_data_feeding.md) for details.
+
+- Feed data directly using `DataFeeder.feed()` in Python codes. It is more flexible than the first way. Many kinds of preprocessing steps can be performed before feeding using Python or any other languages, instead of adding many uncommon `operators` in C++ side. But this method is less efficient: the program cannot read the next mini-batch data before `Executor::Run()` ends. Moreover, `decorated_readers` such as `double_buffer_reader` cannot be used for better performance.
+
+In this document, we design a Python Data Feeding process combining the efficiency of the first way and the flexibility of the second way. A data queue `LoDTensorBlockingQueue` is designed to be shared by the Python and C++ side, while `LoDTensorArray` is pushed into the queue in Python side and `reader_op` in C++ side reads out the data from the queue.
+
+
+## Design of LoDTensorBlockingQueue
+`LoDTensorBlockingQueue` is a blocking queue with a fixed `capacity` and accepts `std::vector<framework::LoDTensor>` with shapes indicated by `dims`. Since `LoDTensorBlockingQueue` must be constructed using `capacity` and `dims`, it cannot be a `Variable` type. Therefore, a `LoDTensorBlockingQueueHolder` is designed to defer construction of `LoDTensorBlockingQueue`.
+
+```C++
+class LoDTensorBlockingQueueHolder;
+
+class LoDTensorBlockingQueue {
+  friend class LoDTensorBlockingQueueHolder;
+ private:
+  // `LoDTensorBlockingQueue` can only be constructed by 
+  // `LoDTensorBlockingQueueHolder::InitOnce()`
+  LoDTensorBlockingQueue(size_t capacity, const std::vector<framework::DDim>& dims);
+ 
+ public:
+  size_t Size() const { return queue_.Size(); } // Get the current size of the queue
+
+  size_t Cap() const { return queue_.Cap(); }// Get the capacity of the queue
+
+  void Close() { return queue_.Close(); }
+
+  bool IsClosed() const { return queue_.IsClosed(); }
+
+  // Block if Size() == Cap()
+  // Return false only when queue_.IsClosed() == true
+  bool Push(const std::vector<framework::LoDTensor> &lod_tensor_vec);
+  
+  // Block if Size() == 0.
+  // *Success == false when queue_.IsClosed() == true
+  std::vector<framework::LoDTensor> Pop(bool *success = nullptr);
+ 
+ private:
+  // Use reader::BlockingQueue as the inner data structure
+  BlockingQueue<std::vector<framework::LoDTensor>> queue_;
+  std::vector<framework::DDim> dims_;
+};
+
+class LoDTensorBlockingQueueHolder {
+ public:  
+  // Call the constructor of `LoDTensorBlockingQueue` to create queue_
+  // `InitOnce` can only called once, otherwise an exception would raise
+  void InitOnce(size_t capacity, const std::vector<framework::DDim>& dims) {
+    PADDLE_ENFORCE(queue_ == nullptr);
+    queue_.reset(new LoDTensorBlockingQueue(capacity, dims));
+  }
+
+  const std::shared_ptr<LoDTensorBlockingQueue>& GetQueue() const { return queue_; }
+
+ private:
+  std::shared_ptr<LoDTensorBlockingQueue> queue_;
+};
+```
+
+There are some major things that must be concerned:
+- `LoDTensorBlockingQueueHolder` should be a `Variable` in global scope, so that `reader_op` can find it when reading data.
+- A `Variable` of `LoDTensorBlockingQueueHolder` but not `VarDesc` must be created in Python code before `Executor::Run()` so that `Executor::Run()` can get the feeding data when it is called.
+- `Create_reader_op` should accept the name of the `LoDTensorBlockingQueueHolder` variable as an input.
+
+
+## Release of the GIL in pybind
+`Pybind11::gil_scoped_release` is used to release GIL (Global Interpreter Lock) when `LoDTensorBlockingQueue::Push()` or `Executor::Run()` method are invoked in Python side, making `LoDTensorBlockingQueue::Push()` and `Executor::Run()` run in parallel.
+
+
+## Design of PyReader
+`PyReader` is a reader which holds a `LoDTensorBlockingQueue` object.
+```C++
+class PyReader : public ReaderBase {
+ public:
+  explicit PyReader(const std::shared_ptr<LoDTensorBlockingQueue>& queue);
+  
+  void ReadNext(std::vector<framework::LoDTensor>* out) override {
+    bool success;
+    *out = queue_->Pop(&success);
+    if (!success) out->clear();
+  }
+  
+  void ReInit() override { return; }
+
+ private:
+  std::shared_ptr<LoDTensorBlockingQueue> queue_;
+};
+```
+
+
+## Design of CreatePyReaderOp
+`CreatePyReaderOp` is used to create the `PyReader` object. It requires an input `blocking_queue` which indicates the name of the `LoDTensorBlockingQueueHolder` variable.
+```C++
+class CreatePyReaderOp : public framework::OperatorBase {
+ public:
+  using framework::OperatorBase::OperatorBase;
+ private:
+  void RunImpl(const framework::Scope& scope,
+               const platform::Place& dev_place) const override {
+    auto* out = scope.FindVar(Output("Out"))
+                    ->template GetMutable<framework::ReaderHolder>();
+    if (out->Get() != nullptr) return;
+    
+    const std::string& queue_name = Input("blocking_queue");
+    auto* queue_holder_var = scope.FindVar(queue_name);
+    PADDLE_ENFORCE(queue_holder_var != nullptr);
+		auto* queue_holder = queue_holder_var
+                    ->template GetMutable<framework::LoDTensorBlockingQueueHolder>();
+    out->Reset(new PyReader(queue_holder->GetQueue()));
+  }
+};
+```
+
+## Design of Python codes
+The design of Python codes are as follows. First, we construct a variable of `LoDTensorBlockingQueueHolder` and init it with given parameters, returning the `LoDTensorBlockingQueue` object after initialization. After that, a layer of `CreatePyReaderOp` is constructed and accepts the name of the `LoDTensorBlockingQueueHolder` variable. The `LoDTensorBlockingQueue` object and result of the layer are both returned.
+```Python
+def py_reader(capacity, shapes):
+  queue_name = unique_name.generate("lod_tensor_blocking_queue")
+  var = global_scope().var(feeder_name) # create LoDTensorBlockingQueueHolder Variable
+  feed_queue = core.init_lod_tensor_blocking_queue(var, capacity, shapes) # init the queue
+  out = create_var()
+  create_py_reader_op_with_queue_name(
+      inputs={'blocking_queue': queue_name},
+      outputs={'Out':[out]})  
+  return out, feed_queue
+```
--- a/doc/fluid/design/multi_devices/kernel_selection.md
+++ b/doc/fluid/design/multi_devices/kernel_selection.md
@@ -74,10 +74,10 @@ void OperatorWithKernel::Run(
    auto kernel_type_for_var = this->GetKernelTypeForVar(...);
    if (kernel_type_for_var.place_ != expected_kernel_key.place_) {
      auto* trans_var = new_scope.Var(var_name);
-      auto* out = DataTransform(expected_kernel_key,
+      auto* out = TransformData(expected_kernel_key,
                                kernel_type_for_var,
                                *tensor_in);
-      CopyVariableWithTensor(...);
+      SetTensorToVariable(...);
    }
  }


--- a/doc/fluid/howto/cluster/fluid_cluster_train_cn.md
+++ b/doc/fluid/howto/cluster/fluid_cluster_train_cn.md
@@ -168,13 +168,13 @@ cd /paddle/python/paddle/fluid/tests/book

 第二步，启动Parameter Server：
 ```bash
-PADDLE_INIT_PORT=6174 PADDLE_INIT_PSERVERS=192.168.1.2 TRAINERS=2 POD_IP=192.168.1.2 PADDLE_INIT_TRAINER_ID=1 TRAINING_ROLE=PSERVER python test_fit_a_line.py
+PADDLE_PSERVER_PORT=6174 PADDLE_PSERVER_IPS=192.168.1.2 PADDLE_TRAINERS=2 PADDLE_CURRENT_IP=192.168.1.2 PADDLE_TRAINER_ID=1 PADDLE_TRAINING_ROLE=PSERVER python test_fit_a_line.py
 ```
 执行命令后请等待出现提示： ```Server listening on 192.168.1.2:6174 ```, 表示Paramter Server已经正常启动。

 第三步，启动Trainer：
 ```bash
-PADDLE_INIT_PORT=6174 PADDLE_INIT_PSERVERS=192.168.1.3 TRAINERS=2 POD_IP=192.168.1.3 PADDLE_INIT_TRAINER_ID=1 TRAINING_ROLE=TRAINER python test_fit_a_line.py
+PADDLE_PSERVER_PORT=6174 PADDLE_PSERVER_IPS=192.168.1.3 PADDLE_TRAINERS=2 PADDLE_CURRENT_IPP=192.168.1.3 PADDLE_TRAINER_ID=1 PADDLE_TRAINING_ROLE=TRAINER python test_fit_a_line.py
 ```
 由于我们定义的Trainer的数量是2个，因此需要在另外一个计算节点上再启动一个Trainer。


--- a/doc/fluid/howto/cluster/fluid_recordio.md
+++ b/doc/fluid/howto/cluster/fluid_recordio.md
@@ -114,8 +114,8 @@ def gen_train_list(file_pattern, trainers, trainer_id):
           ret_list.append(f)
   return ret_list

-trainers = int(os.getenv("TRAINERS"))
-trainer_id = int(os.getenv("PADDLE_INIT_TRAINER_ID"))
+trainers = int(os.getenv("PADDLE_TRAINERS"))
+trainer_id = int(os.getenv("PADDLE_TRAINER_ID"))
 data_file = fluid.layers.io.open_files(
    filenames=gen_train_list("./mnist-[0-9]*.recordio", 2, 0),
    thread_num=1,

--- a/doc/fluid/howto/inference/build_and_install_lib_cn.rst
+++ b/doc/fluid/howto/inference/build_and_install_lib_cn.rst
@@ -13,6 +13,7 @@ cpu_noavx_openblas       `fluid.tgz <https://guest:@paddleci.ngrok.io/repository
 cuda7.5_cudnn5_avx_mkl   `fluid.tgz <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda75cudnn5cp27cp27mu/.lastSuccessful/fluid.tgz>`_
 cuda8.0_cudnn5_avx_mkl   `fluid.tgz <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda80cudnn5cp27cp27mu/.lastSuccessful/fluid.tgz>`_
 cuda8.0_cudnn7_avx_mkl   `fluid.tgz <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda8cudnn7cp27cp27mu/.lastSuccessful/fluid.tgz>`_
+cuda9.0_cudnn7_avx_mkl   `fluid.tgz <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda90cudnn7avxMkl/.lastSuccessful/fluid.tgz>`_
 ======================   ========================================

 从源码编译

--- a/doc/fluid/howto/optimization/host_memory_profiling_cn.md
+++ b/doc/fluid/howto/optimization/host_memory_profiling_cn.md
-## 堆内存分析和优化
+# 堆内存分析和优化

 计算机程序都可能有内存泄漏的风险。**内存泄漏**一般是由于程序在堆(heap)上分配了内存而没有释放，随着程序的运行占用的内存越来越大，一方面会影响程序的稳定性，可能让运行速度越来越慢，或者造成oom，甚至会影响运行程序的机器的稳定性，造成宕机。

@@ -20,11 +20,11 @@ Paddle也提供了基于gperftool的[CPU性能分析教程](https://github.com/P

 对于堆内存的分析，主要用到thread-caching malloc和heap-profiling using tcmalloc。

-## 使用流程
-#### 环境
+## 环境
+
 本教程基于paddle提供的Docker开发环境paddlepaddle/paddle:latest-dev，基于Ubuntu 16.04.4 LTS环境。

-#### 使用流程
+## 使用流程

 - 安装google-perftools


--- a/doc/fluid/howto/optimization/timeline_cn.md
+++ b/doc/fluid/howto/optimization/timeline_cn.md
+# 如何使用timeline工具做性能分析
+
+1. 在训练的主循环外加上`with profiler.profiler(...)`。运行之后，代码会在`/tmp/profile`目录下生成一个profile的记录文件。
+
+	**提示：**
+	请不要在timeline记录信息时运行太多次迭代，因为timeline中的记录数量和迭代次数是成正比的。
+
+	```python
+	with profiler.profiler('All', 'total', '/tmp/profile') as prof:
+	    for pass_id in range(pass_num):
+	        for batch_id, data in enumerate(train_reader()):
+	            exe.run(fluid.default_main_program(),
+	                    feed=feeder.feed(data),
+	                    fetch_list=[])
+	            ...
+	```
+
+1. 运行`python paddle/tools/timeline.py`来处理`/tmp/profile`，这个程序默认会生成一个`/tmp/timeline`文件，你也可以用命令行参数来修改这个路径，请参考[timeline.py](https://github.com/PaddlePaddle/Paddle/blob/develop/tools/timeline.py)。
+
+1. 打开chrome浏览器，访问<chrome://tracing/>，用`load`按钮来加载生成的`timeline`文件。
+
+	![chrome tracing](./tracing.jpeg)
+
+1. 结果如下图所示，可以放到来查看timetime的细节信息。
+
+	![chrome timeline](./timeline.jpeg)
--- a/doc/fluid/howto/optimization/timeline.md
+++ b/doc/fluid/howto/optimization/timeline.md
--- a/doc/survey/dynamic_graph.md
+++ b/doc/survey/dynamic_graph.md
@@ -171,7 +171,7 @@ Pytorch chooses immediate evaluation. It avoids ever materializing a "forward gr

 ## What can fluid learn from them?

-TBD
+Please refer to `paddle/contrib/dynamic/`.

 # Appendix


--- a/doc/v2/api/config/evaluators.rst
+++ b/doc/v2/api/config/evaluators.rst
@@ -101,7 +101,7 @@ value_printer
    :noindex:

 Detection
-=====
+==========

 detection_map
 -------------

--- a/doc/v2/api/config/layer.rst
+++ b/doc/v2/api/config/layer.rst
@@ -11,7 +11,7 @@ Data layer

 data
 ----
-..  autoclass:: paddle.v2.layer.data
+..  autofunction:: paddle.v2.layer.data
    :noindex:

 Fully Connected Layers
@@ -21,12 +21,12 @@ Fully Connected Layers

 fc
 --
-..  autoclass:: paddle.v2.layer.fc
+..  autofunction:: paddle.v2.layer.fc
    :noindex:

 selective_fc
 ------------
-..  autoclass:: paddle.v2.layer.selective_fc
+..  autofunction:: paddle.v2.layer.selective_fc
    :noindex:

 Conv Layers
@@ -34,34 +34,34 @@ Conv Layers

 conv_operator
 -------------
-..  autoclass:: paddle.v2.layer.conv_operator
+..  autofunction:: paddle.v2.layer.conv_operator
    :noindex:

 conv_projection
 ---------------
-..  autoclass:: paddle.v2.layer.conv_projection
+..  autofunction:: paddle.v2.layer.conv_projection
    :noindex:

 conv_shift
 ----------
-..  autoclass:: paddle.v2.layer.conv_shift
+..  autofunction:: paddle.v2.layer.conv_shift
    :noindex:

 img_conv
 --------
-..  autoclass:: paddle.v2.layer.img_conv
+..  autofunction:: paddle.v2.layer.img_conv
    :noindex:

 ..  _api_v2.layer_context_projection:

 context_projection
 ------------------
-..  autoclass:: paddle.v2.layer.context_projection
+..  autofunction:: paddle.v2.layer.context_projection
    :noindex:

 row_conv
 --------
-..  autoclass:: paddle.v2.layer.row_conv
+..  autofunction:: paddle.v2.layer.row_conv
    :noindex:

 Image Pooling Layer
@@ -69,27 +69,27 @@ Image Pooling Layer

 img_pool
 --------
-..  autoclass:: paddle.v2.layer.img_pool
+..  autofunction:: paddle.v2.layer.img_pool
    :noindex:

 spp
 ---
-..  autoclass:: paddle.v2.layer.spp
+..  autofunction:: paddle.v2.layer.spp
    :noindex:

 maxout
 ------
-..  autoclass:: paddle.v2.layer.maxout
+..  autofunction:: paddle.v2.layer.maxout
    :noindex:

 roi_pool
 --------
-..  autoclass:: paddle.v2.layer.roi_pool
+..  autofunction:: paddle.v2.layer.roi_pool
    :noindex:

 pad
 ----
-..  autoclass:: paddle.v2.layer.pad
+..  autofunction:: paddle.v2.layer.pad
    :noindex:

 Norm Layer
@@ -97,27 +97,27 @@ Norm Layer

 img_cmrnorm
 -----------
-..  autoclass:: paddle.v2.layer.img_cmrnorm
+..  autofunction:: paddle.v2.layer.img_cmrnorm
    :noindex:

 batch_norm
 ----------
-..  autoclass:: paddle.v2.layer.batch_norm
+..  autofunction:: paddle.v2.layer.batch_norm
    :noindex:

 sum_to_one_norm
 ---------------
-..  autoclass:: paddle.v2.layer.sum_to_one_norm
+..  autofunction:: paddle.v2.layer.sum_to_one_norm
    :noindex:

 cross_channel_norm
 ------------------
-..  autoclass:: paddle.v2.layer.cross_channel_norm
+..  autofunction:: paddle.v2.layer.cross_channel_norm
    :noindex:

 row_l2_norm
 -----------
-..  autoclass:: paddle.v2.layer.row_l2_norm
+..  autofunction:: paddle.v2.layer.row_l2_norm
    :noindex:

 Recurrent Layers
@@ -125,22 +125,22 @@ Recurrent Layers

 recurrent
 ---------
-..  autoclass:: paddle.v2.layer.recurrent
+..  autofunction:: paddle.v2.layer.recurrent
    :noindex:

 lstmemory
 ---------
-..  autoclass:: paddle.v2.layer.lstmemory
+..  autofunction:: paddle.v2.layer.lstmemory
    :noindex:

 grumemory
 ---------
-..  autoclass:: paddle.v2.layer.grumemory
+..  autofunction:: paddle.v2.layer.grumemory
    :noindex:

 gated_unit
 -----------
-..  autoclass:: paddle.v2.layer.gated_unit
+..  autofunction:: paddle.v2.layer.gated_unit
    :noindex:

 Recurrent Layer Group
@@ -148,32 +148,32 @@ Recurrent Layer Group

 memory
 ------
-..  autoclass:: paddle.v2.layer.memory
+..  autofunction:: paddle.v2.layer.memory
    :noindex:

 recurrent_group
 ---------------
-..  autoclass:: paddle.v2.layer.recurrent_group
+..  autofunction:: paddle.v2.layer.recurrent_group
    :noindex:

 lstm_step
 ---------
-..  autoclass:: paddle.v2.layer.lstm_step
+..  autofunction:: paddle.v2.layer.lstm_step
    :noindex:

 gru_step
 --------
-..  autoclass:: paddle.v2.layer.gru_step
+..  autofunction:: paddle.v2.layer.gru_step
    :noindex:

 beam_search
 ------------
-..  autoclass:: paddle.v2.layer.beam_search
+..  autofunction:: paddle.v2.layer.beam_search
    :noindex:

 get_output
 ----------
-..  autoclass:: paddle.v2.layer.get_output
+..  autofunction:: paddle.v2.layer.get_output
    :noindex:

 Mixed Layer
@@ -183,54 +183,54 @@ Mixed Layer

 mixed
 -----
-..  autoclass:: paddle.v2.layer.mixed
+..  autofunction:: paddle.v2.layer.mixed
    :noindex:

 ..  _api_v2.layer_embedding:

 embedding
 ---------
-..  autoclass:: paddle.v2.layer.embedding
+..  autofunction:: paddle.v2.layer.embedding
    :noindex:

 scaling_projection
 ------------------
-..  autoclass:: paddle.v2.layer.scaling_projection
+..  autofunction:: paddle.v2.layer.scaling_projection
    :noindex:

 dotmul_projection
 -----------------
-..  autoclass:: paddle.v2.layer.dotmul_projection
+..  autofunction:: paddle.v2.layer.dotmul_projection
    :noindex:

 dotmul_operator
 ---------------
-..  autoclass:: paddle.v2.layer.dotmul_operator
+..  autofunction:: paddle.v2.layer.dotmul_operator
    :noindex:

 full_matrix_projection
 ----------------------
-..  autoclass:: paddle.v2.layer.full_matrix_projection
+..  autofunction:: paddle.v2.layer.full_matrix_projection
    :noindex:

 identity_projection
 -------------------
-..  autoclass:: paddle.v2.layer.identity_projection
+..  autofunction:: paddle.v2.layer.identity_projection
    :noindex:

 slice_projection
 -------------------
-..  autoclass:: paddle.v2.layer.slice_projection
+..  autofunction:: paddle.v2.layer.slice_projection
    :noindex:

 table_projection
 ----------------
-..  autoclass:: paddle.v2.layer.table_projection
+..  autofunction:: paddle.v2.layer.table_projection
    :noindex:

 trans_full_matrix_projection
 ----------------------------
-..  autoclass:: paddle.v2.layer.trans_full_matrix_projection
+..  autofunction:: paddle.v2.layer.trans_full_matrix_projection
    :noindex:

 Aggregate Layers
@@ -245,51 +245,46 @@ AggregateLevel

 pooling
 -------
-..  autoclass:: paddle.v2.layer.pooling
+..  autofunction:: paddle.v2.layer.pooling
    :noindex:

 ..  _api_v2.layer_last_seq:

 last_seq
 --------
-..  autoclass:: paddle.v2.layer.last_seq
+..  autofunction:: paddle.v2.layer.last_seq
    :noindex:

 ..  _api_v2.layer_first_seq:

 first_seq
 ---------
-..  autoclass:: paddle.v2.layer.first_seq
+..  autofunction:: paddle.v2.layer.first_seq
    :noindex:

 sub_seq
 ---------
-..  autoclass:: paddle.v2.layer.sub_seq
+..  autofunction:: paddle.v2.layer.sub_seq
    :noindex:

 concat
 ------
-..  autoclass:: paddle.v2.layer.concat
+..  autofunction:: paddle.v2.layer.concat
    :noindex:

 seq_concat
 ----------
-..  autoclass:: paddle.v2.layer.seq_concat
+..  autofunction:: paddle.v2.layer.seq_concat
    :noindex:

 seq_slice
 ---------
-..  autoclass:: paddle.v2.layer.seq_slice
-    :noindex:
-
-kmax_sequence_score
-------------------
-..  autoclass:: paddle.v2.layer.kmax_sequence_score
+..  autofunction:: paddle.v2.layer.seq_slice
    :noindex:

 sub_nested_seq
 --------------
-..  autoclass:: paddle.v2.layer.sub_nested_seq
+..  autofunction:: paddle.v2.layer.sub_nested_seq
    :noindex:

 Reshaping Layers
@@ -297,7 +292,7 @@ Reshaping Layers

 block_expand
 ------------
-..  autoclass:: paddle.v2.layer.block_expand
+..  autofunction:: paddle.v2.layer.block_expand
    :noindex:

 ..  _api_v2.layer_expand:
@@ -309,22 +304,22 @@ ExpandLevel

 expand
 ------
-..  autoclass:: paddle.v2.layer.expand
+..  autofunction:: paddle.v2.layer.expand
    :noindex:

 repeat
 ------
-..  autoclass:: paddle.v2.layer.repeat
+..  autofunction:: paddle.v2.layer.repeat
    :noindex:

 rotate
 ------
-..  autoclass:: paddle.v2.layer.rotate
+..  autofunction:: paddle.v2.layer.rotate
    :noindex:

 seq_reshape
 -----------
-..  autoclass:: paddle.v2.layer.seq_reshape
+..  autofunction:: paddle.v2.layer.seq_reshape
    :noindex:

 Math Layers
@@ -332,94 +327,94 @@ Math Layers

 addto
 -----
-..  autoclass:: paddle.v2.layer.addto
+..  autofunction:: paddle.v2.layer.addto
    :noindex:

 linear_comb
 -----------
-..  autoclass:: paddle.v2.layer.linear_comb
+..  autofunction:: paddle.v2.layer.linear_comb
    :noindex:

 interpolation
 -------------
-..  autoclass:: paddle.v2.layer.interpolation
+..  autofunction:: paddle.v2.layer.interpolation
    :noindex:

 bilinear_interp
 ---------------
-..  autoclass:: paddle.v2.layer.bilinear_interp
+..  autofunction:: paddle.v2.layer.bilinear_interp
    :noindex:

 dropout
 --------
-..  autoclass:: paddle.v2.layer.dropout
+..  autofunction:: paddle.v2.layer.dropout
    :noindex:

 dot_prod
 ---------
-.. autoclass:: paddle.v2.layer.dot_prod
+.. autofunction:: paddle.v2.layer.dot_prod
    :noindex:

 out_prod
 --------
-.. autoclass:: paddle.v2.layer.out_prod
+.. autofunction:: paddle.v2.layer.out_prod
    :noindex:

 power
 -----
-..  autoclass:: paddle.v2.layer.power
+..  autofunction:: paddle.v2.layer.power
    :noindex:

 scaling
 -------
-..  autoclass:: paddle.v2.layer.scaling
+..  autofunction:: paddle.v2.layer.scaling
    :noindex:

 clip
 ----
-..  autoclass:: paddle.v2.layer.clip
+..  autofunction:: paddle.v2.layer.clip
    :noindex:

 resize
 ------
-..  autoclass:: paddle.v2.layer.resize
+..  autofunction:: paddle.v2.layer.resize
    :noindex:

 slope_intercept
 ---------------
-..  autoclass:: paddle.v2.layer.slope_intercept
+..  autofunction:: paddle.v2.layer.slope_intercept
    :noindex:

 tensor
 ------
-..  autoclass:: paddle.v2.layer.tensor
+..  autofunction:: paddle.v2.layer.tensor
    :noindex:

 ..  _api_v2.layer_cos_sim:

 cos_sim
 -------
-..  autoclass:: paddle.v2.layer.cos_sim
+..  autofunction:: paddle.v2.layer.cos_sim
    :noindex:

 l2_distance
 -----------
-..  autoclass:: paddle.v2.layer.l2_distance
+..  autofunction:: paddle.v2.layer.l2_distance
    :noindex:

 trans
 -----
-..  autoclass:: paddle.v2.layer.trans
+..  autofunction:: paddle.v2.layer.trans
    :noindex:

 scale_shift
 -----------
-..  autoclass:: paddle.v2.layer.scale_shift
+..  autofunction:: paddle.v2.layer.scale_shift
    :noindex:

 factorization_machine
 ---------------------
-..  autoclass:: paddle.v2.layer.factorization_machine
+..  autofunction:: paddle.v2.layer.factorization_machine
    :noindex:

 Sampling Layers
@@ -427,17 +422,17 @@ Sampling Layers

 maxid
 -----
-..  autoclass:: paddle.v2.layer.max_id
+..  autofunction:: paddle.v2.layer.max_id
    :noindex:

 sampling_id
 -----------
-..  autoclass:: paddle.v2.layer.sampling_id
+..  autofunction:: paddle.v2.layer.sampling_id
    :noindex:

 multiplex
 ---------
-..  autoclass:: paddle.v2.layer.multiplex
+..  autofunction:: paddle.v2.layer.multiplex
    :noindex:

 ..  _api_v2.layer_costs:
@@ -447,97 +442,97 @@ Cost Layers

 cross_entropy_cost
 ------------------
-..  autoclass:: paddle.v2.layer.cross_entropy_cost
+..  autofunction:: paddle.v2.layer.cross_entropy_cost
    :noindex:

 cross_entropy_with_selfnorm_cost
 --------------------------------
-..  autoclass:: paddle.v2.layer.cross_entropy_with_selfnorm_cost
+..  autofunction:: paddle.v2.layer.cross_entropy_with_selfnorm_cost
    :noindex:

 multi_binary_label_cross_entropy_cost
 -------------------------------------
-..  autoclass:: paddle.v2.layer.multi_binary_label_cross_entropy_cost
+..  autofunction:: paddle.v2.layer.multi_binary_label_cross_entropy_cost
    :noindex:

 classification_cost
 -------------------
-.. autoclass:: paddle.v2.layer.classification_cost
+.. autofunction:: paddle.v2.layer.classification_cost
   :noindex:

 huber_regression_cost
 -------------------------
-..  autoclass:: paddle.v2.layer.huber_regression_cost
+..  autofunction:: paddle.v2.layer.huber_regression_cost
    :noindex:

 huber_classification_cost
 -------------------------
-..  autoclass:: paddle.v2.layer.huber_classification_cost
+..  autofunction:: paddle.v2.layer.huber_classification_cost
    :noindex:

 lambda_cost
 -----------
-..  autoclass:: paddle.v2.layer.lambda_cost
+..  autofunction:: paddle.v2.layer.lambda_cost
    :noindex:

 square_error_cost
 -----------------
-..  autoclass:: paddle.v2.layer.square_error_cost
+..  autofunction:: paddle.v2.layer.square_error_cost
    :noindex:

 rank_cost
 ---------
-..  autoclass:: paddle.v2.layer.rank_cost
+..  autofunction:: paddle.v2.layer.rank_cost
    :noindex:

 sum_cost
 ---------
-..  autoclass:: paddle.v2.layer.sum_cost
+..  autofunction:: paddle.v2.layer.sum_cost
    :noindex:

 crf
 ---
-..  autoclass:: paddle.v2.layer.crf
+..  autofunction:: paddle.v2.layer.crf
    :noindex:

 crf_decoding
 ------------
-..  autoclass:: paddle.v2.layer.crf_decoding
+..  autofunction:: paddle.v2.layer.crf_decoding
    :noindex:

 ctc
 ---
-..  autoclass:: paddle.v2.layer.ctc
+..  autofunction:: paddle.v2.layer.ctc
    :noindex:

 warp_ctc
 --------
-..  autoclass:: paddle.v2.layer.warp_ctc
+..  autofunction:: paddle.v2.layer.warp_ctc
    :noindex:

 nce
 ---
-..  autoclass:: paddle.v2.layer.nce
+..  autofunction:: paddle.v2.layer.nce
    :noindex:

 hsigmoid
 ---------
-..  autoclass:: paddle.v2.layer.hsigmoid
+..  autofunction:: paddle.v2.layer.hsigmoid
    :noindex:

 smooth_l1_cost
 --------------
-..  autoclass:: paddle.v2.layer.smooth_l1_cost
+..  autofunction:: paddle.v2.layer.smooth_l1_cost
    :noindex:

 multibox_loss
 --------------
-..  autoclass:: paddle.v2.layer.multibox_loss
+..  autofunction:: paddle.v2.layer.multibox_loss
    :noindex:

 detection_output
 ----------------
-..  autoclass:: paddle.v2.layer.detection_output
+..  autofunction:: paddle.v2.layer.detection_output
    :noindex:

 Check Layer
@@ -545,7 +540,7 @@ Check Layer

 eos
 ---
-..  autoclass:: paddle.v2.layer.eos
+..  autofunction:: paddle.v2.layer.eos
    :noindex:

 Activation
@@ -553,5 +548,5 @@ Activation

 prelu
 --------
-..  autoclass:: paddle.v2.layer.prelu
+..  autofunction:: paddle.v2.layer.prelu
    :noindex:
--- a/doc/v2/api/index_en.rst
+++ b/doc/v2/api/index_en.rst
@@ -8,4 +8,3 @@ API
    model_configs.rst
    data.rst
    run_logic.rst
-    fluid/index.rst
--- a/doc/v2/build_and_install/pip_install_cn.rst
+++ b/doc/v2/build_and_install/pip_install_cn.rst
@@ -60,6 +60,7 @@ paddlepaddle-gpu==0.11.0            使用CUDA 7.5和cuDNN 5编译的0.11.0版
    "cpu_noavx_openblas", "`paddlepaddle-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_CpuNoavxOpenblas/.lastSuccessful/paddlepaddle-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_CpuNoavxOpenblas/.lastSuccessful/paddlepaddle-latest-cp27-cp27m-linux_x86_64.whl>`_"
    "cuda8.0_cudnn5_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda80cudnn5cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda80cudnn5cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"
    "cuda8.0_cudnn7_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda8cudnn7cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda8cudnn7cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"
+    "cuda9.0_cudnn7_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda90cudnn7avxMkl/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda90cudnn7avxMkl/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"

 .. _pip_dependency:


--- a/doc/v2/build_and_install/pip_install_en.rst
+++ b/doc/v2/build_and_install/pip_install_en.rst
@@ -63,6 +63,7 @@ If the links below shows up the login form, just click "Log in as guest" to star
    "cpu_noavx_openblas", "`paddlepaddle-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_CpuNoavxOpenblas/.lastSuccessful/paddlepaddle-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_CpuNoavxOpenblas/.lastSuccessful/paddlepaddle-latest-cp27-cp27m-linux_x86_64.whl>`__"
    "cuda8.0_cudnn5_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda80cudnn5cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda80cudnn5cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"
    "cuda8.0_cudnn7_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda8cudnn7cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda8cudnn7cp27cp27mu/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"
+    "cuda9.0_cudnn7_avx_mkl", "`paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda90cudnn7avxMkl/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27mu-linux_x86_64.whl>`__", "`paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl <https://guest:@paddleci.ngrok.io/repository/download/Manylinux1_Cuda90cudnn7avxMkl/.lastSuccessful/paddlepaddle_gpu-latest-cp27-cp27m-linux_x86_64.whl>`__"

 .. _pip_dependency:


--- a/doc/v2/design/interface/00.why_plain_c.md
+++ b/doc/v2/design/interface/00.why_plain_c.md
@@ -65,7 +65,7 @@ paddle_error paddle_matrix_get_shape(paddle_matrix matrix,
 而在CPP里面实现这个C的接口，文件 `paddle_matrix.cpp`

 ```cpp
-#include "paddle/math/matrix.h"
+#include "paddle/legacy/math/matrix.h"
 extern "C"
 paddle_error paddle_matrix_shape(paddle_matrix matrix,
                                 uint64_t *width,

--- a/doc/v2/dev/contribute_to_paddle_cn.md
+++ b/doc/v2/dev/contribute_to_paddle_cn.md
@@ -104,7 +104,7 @@ no changes added to commit (use "git add" and/or "git commit -a")
 ➜  docker run -it -v $(pwd):/paddle paddle:latest-dev bash -c "cd /paddle/build && ctest"
 ```

-关于构建和测试的更多信息，请参见[这篇文档](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/getstarted/build_and_install/docker_install_cn.rst)。
+关于构建和测试的更多信息，请参见[使用Docker安装运行](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/v2/build_and_install/docker_install_cn.rst)。

 ## 提交（commit）


--- a/doc/v2/dev/new_layer_cn.rst
+++ b/doc/v2/dev/new_layer_cn.rst
@@ -58,7 +58,7 @@ PaddlePaddle的base layer类可以自动计算上面的导数。
 实现C++类
 ===================

-一个网络层的C++类需要实现初始化，前向和后向。全连接层的实现位于:code:`paddle/gserver/layers/FullyConnectedLayer.h`及:code:`paddle/gserver/layers/FullyConnectedLayer.cpp`。这里我们展示一份简化过的代码。
+一个网络层的C++类需要实现初始化，前向和后向。全连接层的实现位于:code:`paddle/legacy/gserver/layers/FullyConnectedLayer.h`及:code:`paddle/legacy/gserver/layers/FullyConnectedLayer.cpp`。这里我们展示一份简化过的代码。

 这个类需要继承 :code:`paddle::Layer` 这个基类，并且需要重写基类中的以下几个虚函数：

@@ -153,7 +153,7 @@ PaddlePaddle的base layer类可以自动计算上面的导数。

 - 每个层在其 :code:`forward` 函数的开头必须调用 :code:`Layer::forward(passType);` 。
 - 之后使用 :code:`reserveOutput(batchSize, size);` 为输出分配内存。由于我们支持训练数据有不同的批次大小，所以这一步是必要的。 :code:`reserveOutput`  会相应地改变输出的尺寸。为了保证效率，如果需要扩大矩阵，我们会重新分配内存；如果需要缩减矩阵，我们会继续使用现有的内存块。
- 之后使用矩阵运算函数来计算 :math:`\sum_i W_i x + b`。:code:`getInput(i).value` 返回第i个输入矩阵。每个输入都是一个 :math:`batchSize \times dim` 的矩阵，每行表示一个批次中的单个输入。对于我们支持的全部矩阵操作，请参考 :code:`paddle/math/Matrix.h`和:code:`paddle/math/BaseMatrix.h` 。
+- 之后使用矩阵运算函数来计算 :math:`\sum_i W_i x + b`。:code:`getInput(i).value` 返回第i个输入矩阵。每个输入都是一个 :math:`batchSize \times dim` 的矩阵，每行表示一个批次中的单个输入。对于我们支持的全部矩阵操作，请参考 :code:`paddle/legacy/math/Matrix.h`和:code:`paddle/legacy/math/BaseMatrix.h` 。
 - 最终，使用 :code:`forwardActivation();` 进行激活操作。这会自动进行网络配置中声明的激活操作。


@@ -262,7 +262,7 @@ PaddlePaddle的base layer类可以自动计算上面的导数。
    REGISTER_LAYER(fc, FullyConnectedLayer);
    }

-若 :code:`cpp` 被放在 :code:`paddle/gserver/layers` 目录下，其会自动被加入编译列表。
+若 :code:`cpp` 被放在 :code:`paddle/legacy/gserver/layers` 目录下，其会自动被加入编译列表。


 写梯度检查单元测试
@@ -270,7 +270,7 @@ PaddlePaddle的base layer类可以自动计算上面的导数。

 写梯度检查单元测试是一个验证新实现的层是否正确的相对简单的办法。梯度检查单元测试通过有限差分法来验证一个层的梯度。首先对输入做一个小的扰动 :math:`\Delta x` ，然后观察到输出的变化为 :math:`\Delta y` ，那么，梯度就可以通过这个方程计算得到 :math:`\frac{\Delta y}{\Delta x }` 。之后，再用这个梯度去和 :code:`backward` 函数得到的梯度去对比，以保证梯度计算的正确性。需要注意的是梯度检查仅仅验证了梯度的计算，并不保证 :code:`forward` 和 :code:`backward` 函数的实现是正确的。你需要一些更复杂的单元测试来保证你实现的网络层是正确的。

-所有网络层的梯度检查单测都位于 :code:`paddle/gserver/tests/test_LayerGrad.cpp` 。我们建议你在写新网络层时把测试代码放入新的文件中。下面列出了全连接层的梯度检查单元测试。它包含以下几步：
+所有网络层的梯度检查单测都位于 :code:`paddle/legacy/gserver/tests/test_LayerGrad.cpp` 。我们建议你在写新网络层时把测试代码放入新的文件中。下面列出了全连接层的梯度检查单元测试。它包含以下几步：

 + 生成网络层配置。网络层配置包含以下几项：
   - 偏置参数的大小。（例子中是4096）
@@ -322,7 +322,7 @@ PaddlePaddle的base layer类可以自动计算上面的导数。
      }
    }

-如果你要为了测试而增加新的文件，例如 :code:`paddle/gserver/tests/testFCGrad.cpp` ，你需要把该文件加入 :code:`paddle/gserver/tests/CMakeLists.txt` 中。下面给出了一个例子。当你执行命令 :code:`make tests` 时，所有的单测都会被执行一次。注意，有些层可能需要高精度来保证梯度检查单测正确执行。你需要在配置cmake时将 :code:`WITH_DOUBLE` 设置为 `ON` 。
+如果你要为了测试而增加新的文件，例如 :code:`paddle/legacy/gserver/tests/testFCGrad.cpp` ，你需要把该文件加入 :code:`paddle/legacy/gserver/tests/CMakeLists.txt` 中。下面给出了一个例子。当你执行命令 :code:`make tests` 时，所有的单测都会被执行一次。注意，有些层可能需要高精度来保证梯度检查单测正确执行。你需要在配置cmake时将 :code:`WITH_DOUBLE` 设置为 `ON` 。

 .. code-block:: bash


--- a/doc/v2/dev/new_layer_en.rst
+++ b/doc/v2/dev/new_layer_en.rst
@@ -58,7 +58,7 @@ Finally we can use chain rule to calculate :math:`\frac{\partial z}{\partial x}`
 Implement C++ Class
 ===================

-The C++ class of the layer implements the initialization, forward, and backward part of the layer. The fully connected layer is at :code:`paddle/gserver/layers/FullyConnectedLayer.h` and :code:`paddle/gserver/layers/FullyConnectedLayer.cpp`. We list simplified version of the code below.
+The C++ class of the layer implements the initialization, forward, and backward part of the layer. The fully connected layer is at :code:`paddle/legacy/gserver/layers/FullyConnectedLayer.h` and :code:`paddle/legacy/gserver/layers/FullyConnectedLayer.cpp`. We list simplified version of the code below.

 It needs to derive the base class :code:`paddle::Layer`, and it needs to override the following functions:

@@ -154,7 +154,7 @@ The implementation of the forward part has the following steps.

 - Every layer must call :code:`Layer::forward(passType);` at the beginning of its :code:`forward` function.
 - Then it allocates memory for the output using :code:`reserveOutput(batchSize, size);`. This step is necessary because we support the batches to have different batch sizes. :code:`reserveOutput` will change the size of the output accordingly. For the sake of efficiency, we will allocate new memory if we want to expand the matrix, but we will reuse the existing memory block if we want to shrink the matrix.
- Then it computes :math:`\sum_i W_i x + b` using Matrix operations. :code:`getInput(i).value` retrieve the matrix of the i-th input. Each input is a :math:`batchSize \times dim` matrix, where each row represents an single input in a batch. For a complete lists of supported matrix operations, please refer to :code:`paddle/math/Matrix.h` and :code:`paddle/math/BaseMatrix.h`.
+- Then it computes :math:`\sum_i W_i x + b` using Matrix operations. :code:`getInput(i).value` retrieve the matrix of the i-th input. Each input is a :math:`batchSize \times dim` matrix, where each row represents an single input in a batch. For a complete lists of supported matrix operations, please refer to :code:`paddle/legacy/math/Matrix.h` and :code:`paddle/legacy/math/BaseMatrix.h`.
 - Finally it applies the activation function using :code:`forwardActivation();`. It will automatically applies the corresponding activation function specifies in the network configuration.


@@ -263,7 +263,7 @@ Finally, you can use :code:`REGISTER_LAYER(fc, FullyConnectedLayer);` to registe
    REGISTER_LAYER(fc, FullyConnectedLayer);
    }

-If the :code:`cpp` file is put into :code:`paddle/gserver/layers`, it will be automatically added to the compilation list.
+If the :code:`cpp` file is put into :code:`paddle/legacy/gserver/layers`, it will be automatically added to the compilation list.


 Write Gradient Check Unit Test
@@ -271,7 +271,7 @@ Write Gradient Check Unit Test

 An easy way to verify the correctness of new layer's implementation is to write a gradient check unit test. Gradient check unit test utilizes finite difference method to verify the gradient of a layer. It modifies the input with a small perturbation :math:`\Delta x` and observes the changes of output :math:`\Delta y`, the gradient can be computed as :math:`\frac{\Delta y}{\Delta x }`. This gradient can be compared with the gradient computed by the :code:`backward` function of the layer to ensure the correctness of the gradient computation. Notice that the gradient check only tests the correctness of the gradient computation, it does not necessarily guarantee the correctness of the implementation of the :code:`forward` and :code:`backward` function. You need to write more sophisticated unit tests to make sure your layer is implemented correctly.

-All the gradient check unit tests are located in :code:`paddle/gserver/tests/test_LayerGrad.cpp`. You are recommended to put your test into a new test file if you are planning to write a new layer. The gradient test of the gradient check unit test of the fully connected layer is listed below. It has the following steps.
+All the gradient check unit tests are located in :code:`paddle/legacy/gserver/tests/test_LayerGrad.cpp`. You are recommended to put your test into a new test file if you are planning to write a new layer. The gradient test of the gradient check unit test of the fully connected layer is listed below. It has the following steps.

 + Create layer configuration. A layer configuration can include the following attributes:
   - size of the bias parameter. (4096 in our example)
@@ -323,7 +323,7 @@ All the gradient check unit tests are located in :code:`paddle/gserver/tests/tes
      }
    }

-If you are creating a new file for the test, such as :code:`paddle/gserver/tests/testFCGrad.cpp`, you need to add the file to :code:`paddle/gserver/tests/CMakeLists.txt`. An example is given below. All the unit tests will run when you execute the command :code:`make tests`. Notice that some layers might need high accuracy for the gradient check unit tests to work well. You need to configure :code:`WITH_DOUBLE` to `ON` when configuring cmake.
+If you are creating a new file for the test, such as :code:`paddle/legacy/gserver/tests/testFCGrad.cpp`, you need to add the file to :code:`paddle/legacy/gserver/tests/CMakeLists.txt`. An example is given below. All the unit tests will run when you execute the command :code:`make tests`. Notice that some layers might need high accuracy for the gradient check unit tests to work well. You need to configure :code:`WITH_DOUBLE` to `ON` when configuring cmake.

 .. code-block:: bash


--- a/doc/v2/faq/build_and_install/index_cn.rst
+++ b/doc/v2/faq/build_and_install/index_cn.rst
@@ -213,3 +213,12 @@ virtualenv本身也是Python的一个包，可以用pip进行安装：
 保存并关闭文件。

 这样，每次打开终端时就会自动启动名为‘paddle’的Python环境了。
+
+10. 通过pip安装的PaddlePaddle在  :code:`import paddle.fluid` 报找不到 :code:`libmkldnn.so` 或 :code:`libmklml_intel.so`
+------------------------------------------------------------------------------------------
+出现这种问题的原因是在导入 :code:`paddle.fluid` 时需要加载 :code:`libmkldnn.so` 和 :code:`libmklml_intel.so`，
+但是系统没有找到该文件。一般通过pip安装PaddlePaddle时会将 :code:`libmkldnn.so` 和 :code:`libmklml_intel.so`
+拷贝到 :code:`/usr/local/lib` 路径下，所以解决办法是将该路径加到 :code:`LD_LIBRARY_PATH` 环境变量下，
+即： :code:`export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH` 。
+
+**注意**：如果是在虚拟环境中安装PaddlePaddle， :code:`libmkldnn.so` 和 :code:`libmklml_intel.so` 可能不在 :code:`/usr/local/lib` 路径下。
\ No newline at end of file
--- a/doc/v2/faq/parameter/index_cn.rst
+++ b/doc/v2/faq/parameter/index_cn.rst
@@ -196,6 +196,6 @@ PaddlePaddle保存的模型参数文件内容由16字节头信息和网络参数
        obj="process",
        args={"src_dict_path": src_dict_path})

-完整源码可参考 `sequence_recurrent <https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/gserver/tests/sequence_recurrent.py>`_ 示例。
+完整源码可参考 `sequence_recurrent <https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/legacy/gserver/tests/sequence_recurrent.py>`_ 示例。


--- a/doc/v2/howto/optimization/gpu_profiling_cn.rst
+++ b/doc/v2/howto/optimization/gpu_profiling_cn.rst
@@ -50,12 +50,12 @@ GPU则还需要高并行性，才能发挥其全部能力。这正是它们速
 **nvprof** 是Nvidia性能分析工具， **nvvp** 则是带GUI的Nvidia可视化性能分析工具。
 在这个教程中，我们主要会介绍nvprof和nvvp。

-:code:`test_GpuProfiler` from :code:`paddle/math/tests` directory will be used to evaluate
+:code:`test_GpuProfiler` from :code:`paddle/legacy/math/tests` directory will be used to evaluate
 above profilers.

-:code:`paddle/math/test` 目录中的 :code:`test_GpuProfiler` 就是用于展示上述分析工具的用法。
+:code:`paddle/legacy/math/test` 目录中的 :code:`test_GpuProfiler` 就是用于展示上述分析工具的用法。

-.. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+.. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
   :language: c++
   :lines: 137-151
   :linenos:
@@ -83,7 +83,7 @@ program crashes when CPU version of PaddlePaddle invokes them.

 1. 加入 :code:`REGISTER_TIMER_INFO` 和 :code:`printAllStatus` 函数（如高亮部分）。

-    .. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
        :language: c++
        :lines: 137-151
        :emphasize-lines: 8-12,14
@@ -101,8 +101,8 @@ program crashes when CPU version of PaddlePaddle invokes them.
    .. code-block:: bash
        :emphasize-lines: 1,12-15

-        > ./paddle/math/tests/test_GpuProfiler
-        I1117 11:13:42.313065 2522362816 Util.cpp:155] commandline: ./paddle/math/tests/test_GpuProfiler
+        > ./paddle/legacy/math/tests/test_GpuProfiler
+        I1117 11:13:42.313065 2522362816 Util.cpp:155] commandline: ./paddle/legacy/math/tests/test_GpuProfiler
        I1117 11:13:42.845065 2522362816 Util.cpp:130] Calling runInitFunctions
        I1117 11:13:42.845208 2522362816 Util.cpp:143] Call runInitFunctions done.
        [==========] Running 1 test from 1 test case.
@@ -130,7 +130,7 @@ nvprof 工具

 1. 将 :code:`REGISTER_GPU_PROFILER` 函数加到代码中（参考强调部分）。

-    .. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
        :language: c++
        :lines: 137-151
        :emphasize-lines: 6-7
@@ -147,13 +147,13 @@ nvprof 工具

    .. code-block:: bash

-        nvprof  ./paddle/math/tests/test_GpuProfiler
+        nvprof  ./paddle/legacy/math/tests/test_GpuProfiler

 然后，您就能获得如下的分析结果：

 .. code-block:: bash

-    ==78544== Profiling application: ./paddle/math/tests/test_GpuProfiler
+    ==78544== Profiling application: ./paddle/legacy/math/tests/test_GpuProfiler
    ==78544== Profiling result:
    Time(%)     Time     Calls       Avg       Min       Max  Name
    27.60%  9.6305ms         5  1.9261ms  3.4560us  6.4035ms  [CUDA memcpy HtoD]

--- a/doc/v2/howto/optimization/gpu_profiling_en.rst
+++ b/doc/v2/howto/optimization/gpu_profiling_en.rst
@@ -51,10 +51,10 @@ For general GPU profiling, a bunch of tools are provided from both NVIDIA and th
 **nvprof** is Nvidia profiler and **nvvp** is (GUI based) Nvidia visual profiler.
 In this tutorial, we will focus on nvprof and nvvp.

-:code:`test_GpuProfiler` from :code:`paddle/math/tests` directory will be used to evaluate
+:code:`test_GpuProfiler` from :code:`paddle/legacy/math/tests` directory will be used to evaluate
 above profilers.

-.. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+.. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
   :language: c++
   :lines: 137-151
   :linenos:
@@ -80,7 +80,7 @@ As a simple example, consider the following:

 1. Add :code:`REGISTER_TIMER_INFO` and :code:`printAllStatus` functions (see the emphasize-lines).

-    .. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
        :language: c++
        :lines: 137-151
        :emphasize-lines: 8-12,14
@@ -98,8 +98,8 @@ As a simple example, consider the following:
    .. code-block:: bash
        :emphasize-lines: 1,12-15

-        > ./paddle/math/tests/test_GpuProfiler
-        I1117 11:13:42.313065 2522362816 Util.cpp:155] commandline: ./paddle/math/tests/test_GpuProfiler
+        > ./paddle/legacy/math/tests/test_GpuProfiler
+        I1117 11:13:42.313065 2522362816 Util.cpp:155] commandline: ./paddle/legacy/math/tests/test_GpuProfiler
        I1117 11:13:42.845065 2522362816 Util.cpp:130] Calling runInitFunctions
        I1117 11:13:42.845208 2522362816 Util.cpp:143] Call runInitFunctions done.
        [==========] Running 1 test from 1 test case.
@@ -127,7 +127,7 @@ To use this command line profiler **nvprof**, you can simply issue the following

 1. Add :code:`REGISTER_GPU_PROFILER` function (see the emphasize-lines).

-    .. literalinclude:: ../../../../paddle/math/tests/test_GpuProfiler.cpp
+    .. literalinclude:: ../../../../paddle/legacy/math/tests/test_GpuProfiler.cpp
        :language: c++
        :lines: 137-151
        :emphasize-lines: 6-7
@@ -144,13 +144,13 @@ To use this command line profiler **nvprof**, you can simply issue the following

    .. code-block:: bash

-        nvprof  ./paddle/math/tests/test_GpuProfiler
+        nvprof  ./paddle/legacy/math/tests/test_GpuProfiler

 Then, you can get the following profiling result:

 .. code-block:: bash

-    ==78544== Profiling application: ./paddle/math/tests/test_GpuProfiler
+    ==78544== Profiling application: ./paddle/legacy/math/tests/test_GpuProfiler
    ==78544== Profiling result:
    Time(%)     Time     Calls       Avg       Min       Max  Name
    27.60%  9.6305ms         5  1.9261ms  3.4560us  6.4035ms  [CUDA memcpy HtoD]

--- a/doc/v2/howto/rnn/hrnn_rnn_api_compare_cn.rst
+++ b/doc/v2/howto/rnn/hrnn_rnn_api_compare_cn.rst
--- a/doc/v2/howto/rnn/hrnn_rnn_api_compare_en.rst
+++ b/doc/v2/howto/rnn/hrnn_rnn_api_compare_en.rst
--- a/go/pserver/optimizer.go
+++ b/go/pserver/optimizer.go
@@ -16,7 +16,7 @@ package pserver

 // #cgo CFLAGS: -I ../../
 // #cgo LDFLAGS: ${SRCDIR}/client/c/libpaddle_go_optimizer.a -lstdc++ -lm
-// #include "paddle/optimizer/optimizer.h"
+// #include "paddle/legacy/optimizer/optimizer.h"
 // #include <stdlib.h>
 // #include <string.h>
 import "C"

--- a/paddle/CMakeLists.txt
+++ b/paddle/CMakeLists.txt
 if(NOT WITH_FLUID_ONLY)
-  add_subdirectory(cuda)
-  add_subdirectory(function)
+  add_subdirectory(legacy/cuda)
+  add_subdirectory(legacy/function)
  add_subdirectory(utils)
-  add_subdirectory(math)
-  add_subdirectory(gserver)
-  add_subdirectory(parameter)
+  add_subdirectory(legacy/math)
+  add_subdirectory(legacy/gserver)
+  add_subdirectory(legacy/parameter)

  if(MOBILE_INFERENCE)
    add_subdirectory(capi)
  else()
-    add_subdirectory(pserver)
+    add_subdirectory(legacy/pserver)
    add_subdirectory(trainer)
    add_subdirectory(scripts)


--- a/paddle/api/Arguments.cpp
+++ b/paddle/api/Arguments.cpp
@@ -15,7 +15,7 @@ limitations under the License. */
 #include "PaddleAPI.h"
 #include "PaddleAPIPrivate.h"

-#include "paddle/parameter/Argument.h"
+#include "paddle/legacy/parameter/Argument.h"

 size_t Arguments::getSlotNum() const { return m->outputs.size(); }


--- a/paddle/api/GradientMachine.cpp
+++ b/paddle/api/GradientMachine.cpp
@@ -16,7 +16,7 @@ limitations under the License. */
 #include "PaddleAPIPrivate.h"

 #include "Internal.h"
-#include "paddle/gserver/gradientmachines/NeuralNetwork.h"
+#include "paddle/legacy/gserver/gradientmachines/NeuralNetwork.h"

 std::vector<int> GradientMachine::defaultParamTypes = {
    PARAMETER_VALUE, PARAMETER_GRADIENT, PARAMETER_MOMENTUM};

--- a/paddle/api/Matrix.cpp
+++ b/paddle/api/Matrix.cpp
@@ -12,12 +12,12 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License. */

-#include "paddle/math/Matrix.h"
+#include "paddle/legacy/math/Matrix.h"
 #include <cstring>
 #include <iostream>
 #include "PaddleAPI.h"
-#include "paddle/math/CpuSparseMatrix.h"
-#include "paddle/math/SparseMatrix.h"
+#include "paddle/legacy/math/CpuSparseMatrix.h"
+#include "paddle/legacy/math/SparseMatrix.h"

 struct MatrixPrivate {
  std::shared_ptr<paddle::Matrix> mat;

--- a/paddle/api/PaddleAPI.h
+++ b/paddle/api/PaddleAPI.h
--- a/paddle/api/PaddleAPIPrivate.h
+++ b/paddle/api/PaddleAPIPrivate.h
--- a/paddle/api/Parameter.cpp
+++ b/paddle/api/Parameter.cpp
--- a/paddle/api/ParameterOptimizer.cpp
+++ b/paddle/api/ParameterOptimizer.cpp
--- a/paddle/api/SequenceGenerator.cpp
+++ b/paddle/api/SequenceGenerator.cpp
--- a/paddle/api/Trainer.cpp
+++ b/paddle/api/Trainer.cpp
--- a/paddle/api/Util.cpp
+++ b/paddle/api/Util.cpp
--- a/paddle/api/Vector.cpp
+++ b/paddle/api/Vector.cpp
--- a/paddle/capi/capi_private.h
+++ b/paddle/capi/capi_private.h
--- a/paddle/capi/gradient_machine.cpp
+++ b/paddle/capi/gradient_machine.cpp
--- a/paddle/capi/tests/test_GradientMachine.cpp
+++ b/paddle/capi/tests/test_GradientMachine.cpp
--- a/paddle/contrib/inference/CMakeLists.txt
+++ b/paddle/contrib/inference/CMakeLists.txt
--- a/paddle/contrib/inference/demo/CMakeLists.txt
+++ b/paddle/contrib/inference/demo/CMakeLists.txt
--- a/paddle/contrib/inference/demo/README.md
+++ b/paddle/contrib/inference/demo/README.md
--- a/paddle/contrib/inference/demo/simple_on_word2vec.cc
+++ b/paddle/contrib/inference/demo/simple_on_word2vec.cc
--- a/paddle/contrib/inference/demo/utils.h
+++ b/paddle/contrib/inference/demo/utils.h
--- a/paddle/contrib/inference/demo/vis_demo.cc
+++ b/paddle/contrib/inference/demo/vis_demo.cc
--- a/paddle/contrib/inference/high_level_api.md
+++ b/paddle/contrib/inference/high_level_api.md
--- a/paddle/contrib/inference/high_level_api_cn.md
+++ b/paddle/contrib/inference/high_level_api_cn.md
--- a/paddle/contrib/inference/paddle_inference_api.cc
+++ b/paddle/contrib/inference/paddle_inference_api.cc
--- a/paddle/contrib/inference/paddle_inference_api.h
+++ b/paddle/contrib/inference/paddle_inference_api.h
--- a/paddle/contrib/inference/paddle_inference_api_anakin_engine.cc
+++ b/paddle/contrib/inference/paddle_inference_api_anakin_engine.cc
--- a/paddle/contrib/inference/paddle_inference_api_anakin_engine.h
+++ b/paddle/contrib/inference/paddle_inference_api_anakin_engine.h
--- a/paddle/contrib/inference/paddle_inference_api_anakin_engine_tester.cc
+++ b/paddle/contrib/inference/paddle_inference_api_anakin_engine_tester.cc
--- a/paddle/contrib/inference/paddle_inference_api_impl.cc
+++ b/paddle/contrib/inference/paddle_inference_api_impl.cc
--- a/paddle/contrib/inference/paddle_inference_api_impl.h
+++ b/paddle/contrib/inference/paddle_inference_api_impl.h
--- a/paddle/contrib/inference/paddle_inference_api_tensorrt_subgraph_engine.cc
+++ b/paddle/contrib/inference/paddle_inference_api_tensorrt_subgraph_engine.cc
--- a/paddle/contrib/inference/test_paddle_inference_api_impl.cc
+++ b/paddle/contrib/inference/test_paddle_inference_api_impl.cc
--- a/paddle/contrib/inference/test_paddle_inference_api_tensorrt_subgraph_engine.cc
+++ b/paddle/contrib/inference/test_paddle_inference_api_tensorrt_subgraph_engine.cc
--- a/paddle/cuda/include/hl_base.h
+++ b/paddle/cuda/include/hl_base.h
--- a/paddle/cuda/src/hl_top_k.cu
+++ b/paddle/cuda/src/hl_top_k.cu
--- a/paddle/fluid/framework/CMakeLists.txt
+++ b/paddle/fluid/framework/CMakeLists.txt
--- a/paddle/fluid/framework/data_layout_transform.cc
+++ b/paddle/fluid/framework/data_layout_transform.cc
--- a/paddle/fluid/framework/data_layout_transform.h
+++ b/paddle/fluid/framework/data_layout_transform.h
--- a/paddle/fluid/framework/data_transform.cc
+++ b/paddle/fluid/framework/data_transform.cc
--- a/paddle/fluid/framework/data_transform.h
+++ b/paddle/fluid/framework/data_transform.h
--- a/paddle/fluid/framework/details/broadcast_op_handle.cc
+++ b/paddle/fluid/framework/details/broadcast_op_handle.cc
--- a/paddle/fluid/framework/details/multi_devices_graph_builder.cc
+++ b/paddle/fluid/framework/details/multi_devices_graph_builder.cc
--- a/paddle/fluid/framework/details/multi_devices_graph_builder.h
+++ b/paddle/fluid/framework/details/multi_devices_graph_builder.h
--- a/paddle/fluid/framework/details/op_handle_base.cc
+++ b/paddle/fluid/framework/details/op_handle_base.cc
--- a/paddle/fluid/framework/details/ssa_graph_builder.h
+++ b/paddle/fluid/framework/details/ssa_graph_builder.h
--- a/paddle/fluid/framework/details/ssa_graph_checker.h
+++ b/paddle/fluid/framework/details/ssa_graph_checker.h
--- a/paddle/fluid/framework/details/ssa_graph_printer.h
+++ b/paddle/fluid/framework/details/ssa_graph_printer.h
--- a/paddle/fluid/framework/details/threaded_ssa_graph_executor.cc
+++ b/paddle/fluid/framework/details/threaded_ssa_graph_executor.cc
--- a/paddle/fluid/framework/details/threaded_ssa_graph_executor.h
+++ b/paddle/fluid/framework/details/threaded_ssa_graph_executor.h
--- a/paddle/fluid/framework/executor.cc
+++ b/paddle/fluid/framework/executor.cc
--- a/paddle/fluid/framework/executor.h
+++ b/paddle/fluid/framework/executor.h
--- a/paddle/fluid/framework/framework.proto
+++ b/paddle/fluid/framework/framework.proto
--- a/paddle/fluid/framework/init.cc
+++ b/paddle/fluid/framework/init.cc
--- a/paddle/fluid/framework/lod_tensor.cc
+++ b/paddle/fluid/framework/lod_tensor.cc
--- a/paddle/fluid/framework/lod_tensor.h
+++ b/paddle/fluid/framework/lod_tensor.h
--- a/paddle/fluid/framework/lod_tensor_test.cc
+++ b/paddle/fluid/framework/lod_tensor_test.cc
--- a/paddle/fluid/framework/op_desc.cc
+++ b/paddle/fluid/framework/op_desc.cc
--- a/paddle/fluid/framework/op_desc.h
+++ b/paddle/fluid/framework/op_desc.h
--- a/paddle/fluid/framework/op_kernel_type.h
+++ b/paddle/fluid/framework/op_kernel_type.h
--- a/paddle/fluid/framework/operator.cc
+++ b/paddle/fluid/framework/operator.cc
--- a/paddle/fluid/framework/operator.h
+++ b/paddle/fluid/framework/operator.h
--- a/paddle/fluid/framework/parallel_executor.cc
+++ b/paddle/fluid/framework/parallel_executor.cc
--- a/paddle/fluid/framework/parallel_executor.h
+++ b/paddle/fluid/framework/parallel_executor.h
--- a/paddle/fluid/framework/reader.h
+++ b/paddle/fluid/framework/reader.h
--- a/paddle/fluid/framework/scope.cc
+++ b/paddle/fluid/framework/scope.cc
--- a/paddle/fluid/framework/scope.h
+++ b/paddle/fluid/framework/scope.h
--- a/paddle/fluid/framework/tensor_impl.h
+++ b/paddle/fluid/framework/tensor_impl.h
--- a/paddle/fluid/framework/tensor_util.cc
+++ b/paddle/fluid/framework/tensor_util.cc
--- a/paddle/fluid/framework/tensor_util.h
+++ b/paddle/fluid/framework/tensor_util.h
--- a/paddle/fluid/framework/type_defs.h
+++ b/paddle/fluid/framework/type_defs.h
--- a/paddle/fluid/framework/var_type.h
+++ b/paddle/fluid/framework/var_type.h
--- a/paddle/fluid/inference/CMakeLists.txt
+++ b/paddle/fluid/inference/CMakeLists.txt
--- a/paddle/fluid/inference/analysis/CMakeLists.txt
+++ b/paddle/fluid/inference/analysis/CMakeLists.txt
--- a/paddle/fluid/inference/analysis/README.md
+++ b/paddle/fluid/inference/analysis/README.md
--- a/paddle/fluid/inference/analysis/analyzer.cc
+++ b/paddle/fluid/inference/analysis/analyzer.cc
--- a/paddle/fluid/inference/analysis/analyzer.h
+++ b/paddle/fluid/inference/analysis/analyzer.h
--- a/paddle/fluid/inference/analysis/analyzer_tester.cc
+++ b/paddle/fluid/inference/analysis/analyzer_tester.cc
--- a/paddle/fluid/inference/analysis/argument.cc
+++ b/paddle/fluid/inference/analysis/argument.cc
--- a/paddle/fluid/inference/analysis/argument.h
+++ b/paddle/fluid/inference/analysis/argument.h
--- a/paddle/fluid/inference/analysis/data_flow_graph.cc
+++ b/paddle/fluid/inference/analysis/data_flow_graph.cc
--- a/paddle/fluid/inference/analysis/data_flow_graph.h
+++ b/paddle/fluid/inference/analysis/data_flow_graph.h
--- a/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass.cc
+++ b/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass.cc
--- a/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass.h
+++ b/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass.h
--- a/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass_tester.cc
+++ b/paddle/fluid/inference/analysis/data_flow_graph_to_fluid_pass_tester.cc
--- a/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass.cc
+++ b/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass.cc
--- a/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass.h
+++ b/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass.h
--- a/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass_tester.cc
+++ b/paddle/fluid/inference/analysis/dfg_graphviz_draw_pass_tester.cc
--- a/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass.cc
+++ b/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass.cc
--- a/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass.h
+++ b/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass.h
--- a/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass_tester.cc
+++ b/paddle/fluid/inference/analysis/fluid_to_data_flow_graph_pass_tester.cc
--- a/paddle/fluid/inference/analysis/helper.cc
+++ b/paddle/fluid/inference/analysis/helper.cc
--- a/paddle/fluid/inference/analysis/helper.h
+++ b/paddle/fluid/inference/analysis/helper.h
--- a/paddle/fluid/inference/analysis/node.cc
+++ b/paddle/fluid/inference/analysis/node.cc
--- a/paddle/fluid/inference/analysis/node.h
+++ b/paddle/fluid/inference/analysis/node.h
--- a/paddle/fluid/inference/analysis/node_attr_flags.h
+++ b/paddle/fluid/inference/analysis/node_attr_flags.h
--- a/paddle/fluid/inference/analysis/pass.h
+++ b/paddle/fluid/inference/analysis/pass.h
--- a/paddle/fluid/inference/analysis/pass_manager.cc
+++ b/paddle/fluid/inference/analysis/pass_manager.cc
--- a/paddle/fluid/inference/analysis/pass_manager.h
+++ b/paddle/fluid/inference/analysis/pass_manager.h
--- a/paddle/fluid/inference/analysis/pass_manager_tester.cc
+++ b/paddle/fluid/inference/analysis/pass_manager_tester.cc
--- a/paddle/fluid/inference/analysis/subgraph_splitter.cc
+++ b/paddle/fluid/inference/analysis/subgraph_splitter.cc
--- a/paddle/fluid/inference/analysis/subgraph_splitter_tester.cc
+++ b/paddle/fluid/inference/analysis/subgraph_splitter_tester.cc
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass.cc
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass.cc
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass.h
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass.h
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass_tester.cc
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_node_mark_pass_tester.cc
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_pass.cc
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_pass.cc
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_pass.h
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_pass.h
--- a/paddle/fluid/inference/analysis/tensorrt_subgraph_pass_tester.cc
+++ b/paddle/fluid/inference/analysis/tensorrt_subgraph_pass_tester.cc
--- a/paddle/fluid/inference/analysis/ut_helper.h
+++ b/paddle/fluid/inference/analysis/ut_helper.h
--- a/paddle/fluid/inference/io.cc
+++ b/paddle/fluid/inference/io.cc
--- a/paddle/fluid/inference/tensorrt/convert/op_converter.h
+++ b/paddle/fluid/inference/tensorrt/convert/op_converter.h
--- a/paddle/fluid/inference/tensorrt/engine.h
+++ b/paddle/fluid/inference/tensorrt/engine.h
--- a/paddle/fluid/inference/tests/book/test_inference_nlp.cc
+++ b/paddle/fluid/inference/tests/book/test_inference_nlp.cc
--- a/paddle/fluid/memory/detail/system_allocator.cc
+++ b/paddle/fluid/memory/detail/system_allocator.cc
--- a/paddle/fluid/memory/malloc.cc
+++ b/paddle/fluid/memory/malloc.cc
--- a/paddle/fluid/operators/CMakeLists.txt
+++ b/paddle/fluid/operators/CMakeLists.txt
--- a/paddle/fluid/operators/activation_mkldnn_op.cc
+++ b/paddle/fluid/operators/activation_mkldnn_op.cc
--- a/paddle/fluid/operators/activation_op.cc
+++ b/paddle/fluid/operators/activation_op.cc
--- a/paddle/fluid/operators/adam_op.cc
+++ b/paddle/fluid/operators/adam_op.cc
--- a/paddle/fluid/operators/adam_op.h
+++ b/paddle/fluid/operators/adam_op.h
--- a/paddle/fluid/operators/argsort_op.cc
+++ b/paddle/fluid/operators/argsort_op.cc
--- a/paddle/fluid/operators/argsort_op.cu
+++ b/paddle/fluid/operators/argsort_op.cu
--- a/paddle/fluid/operators/argsort_op.h
+++ b/paddle/fluid/operators/argsort_op.h
--- a/paddle/fluid/operators/assign_value_op.cc
+++ b/paddle/fluid/operators/assign_value_op.cc
--- a/paddle/fluid/operators/average_accumulates_op.cc
+++ b/paddle/fluid/operators/average_accumulates_op.cc
--- a/paddle/fluid/operators/average_accumulates_op.h
+++ b/paddle/fluid/operators/average_accumulates_op.h
--- a/paddle/fluid/operators/batch_norm_mkldnn_op.cc
+++ b/paddle/fluid/operators/batch_norm_mkldnn_op.cc
--- a/paddle/fluid/operators/batch_norm_op.cc
+++ b/paddle/fluid/operators/batch_norm_op.cc
--- a/paddle/fluid/operators/batch_norm_op.h
+++ b/paddle/fluid/operators/batch_norm_op.h
--- a/paddle/fluid/operators/beam_search_decode_op.cc
+++ b/paddle/fluid/operators/beam_search_decode_op.cc
--- a/paddle/fluid/operators/beam_search_decode_op.h
+++ b/paddle/fluid/operators/beam_search_decode_op.h
--- a/paddle/fluid/operators/beam_search_decode_op_test.cc
+++ b/paddle/fluid/operators/beam_search_decode_op_test.cc
--- a/paddle/fluid/operators/beam_search_op.cc
+++ b/paddle/fluid/operators/beam_search_op.cc
--- a/paddle/fluid/operators/beam_search_op.h
+++ b/paddle/fluid/operators/beam_search_op.h
--- a/paddle/fluid/operators/beam_search_op_test.cc
+++ b/paddle/fluid/operators/beam_search_op_test.cc
--- a/paddle/fluid/operators/bilinear_interp_op.cc
+++ b/paddle/fluid/operators/bilinear_interp_op.cc
--- a/paddle/fluid/operators/bilinear_interp_op.h
+++ b/paddle/fluid/operators/bilinear_interp_op.h
--- a/paddle/fluid/operators/checkpoint_notify_op.cc
+++ b/paddle/fluid/operators/checkpoint_notify_op.cc
--- a/paddle/fluid/operators/chunk_eval_op.cc
+++ b/paddle/fluid/operators/chunk_eval_op.cc
--- a/paddle/fluid/operators/clip_by_norm_op.cc
+++ b/paddle/fluid/operators/clip_by_norm_op.cc
--- a/paddle/fluid/operators/compare_op.cc
+++ b/paddle/fluid/operators/compare_op.cc
--- a/paddle/fluid/operators/concat_op.cc
+++ b/paddle/fluid/operators/concat_op.cc
--- a/paddle/fluid/operators/concat_op.cu.cc
+++ b/paddle/fluid/operators/concat_op.cu.cc
--- a/paddle/fluid/operators/concat_op.h
+++ b/paddle/fluid/operators/concat_op.h
--- a/paddle/fluid/operators/conditional_block_op.cc
+++ b/paddle/fluid/operators/conditional_block_op.cc
--- a/paddle/fluid/operators/conv_mkldnn_op.cc
+++ b/paddle/fluid/operators/conv_mkldnn_op.cc
--- a/paddle/fluid/operators/conv_op.cc
+++ b/paddle/fluid/operators/conv_op.cc
--- a/paddle/fluid/operators/conv_transpose_op.cc
+++ b/paddle/fluid/operators/conv_transpose_op.cc
--- a/paddle/fluid/operators/conv_transpose_op.cu.cc
+++ b/paddle/fluid/operators/conv_transpose_op.cu.cc
--- a/paddle/fluid/operators/conv_transpose_op.h
+++ b/paddle/fluid/operators/conv_transpose_op.h
--- a/paddle/fluid/operators/cos_sim_op.cc
+++ b/paddle/fluid/operators/cos_sim_op.cc
--- a/paddle/fluid/operators/crf_decoding_op.cc
+++ b/paddle/fluid/operators/crf_decoding_op.cc
--- a/paddle/fluid/operators/crop_op.h
+++ b/paddle/fluid/operators/crop_op.h
--- a/paddle/fluid/operators/cumsum_op.cc
+++ b/paddle/fluid/operators/cumsum_op.cc
--- a/paddle/fluid/operators/detail/CMakeLists.txt
+++ b/paddle/fluid/operators/detail/CMakeLists.txt
--- a/paddle/fluid/operators/detail/brpc_client.cc
+++ b/paddle/fluid/operators/detail/brpc_client.cc
--- a/paddle/fluid/operators/detail/brpc_client.h
+++ b/paddle/fluid/operators/detail/brpc_client.h
--- a/paddle/fluid/operators/detail/brpc_server.cc
+++ b/paddle/fluid/operators/detail/brpc_server.cc
--- a/paddle/fluid/operators/detail/brpc_server.h
+++ b/paddle/fluid/operators/detail/brpc_server.h
--- a/paddle/fluid/operators/detail/bytebuffer_stream.cc
+++ b/paddle/fluid/operators/detail/bytebuffer_stream.cc
--- a/paddle/fluid/operators/detail/bytebuffer_stream.h
+++ b/paddle/fluid/operators/detail/bytebuffer_stream.h
--- a/paddle/fluid/operators/detail/grpc_client.cc
+++ b/paddle/fluid/operators/detail/grpc_client.cc
--- a/paddle/fluid/operators/detail/grpc_client.h
+++ b/paddle/fluid/operators/detail/grpc_client.h
--- a/paddle/fluid/operators/detail/grpc_serde_test.cc
+++ b/paddle/fluid/operators/detail/grpc_serde_test.cc
--- a/paddle/fluid/operators/detail/grpc_server.cc
+++ b/paddle/fluid/operators/detail/grpc_server.cc
--- a/paddle/fluid/operators/detail/grpc_server.h
+++ b/paddle/fluid/operators/detail/grpc_server.h
--- a/paddle/fluid/operators/detail/grpc_service.h
+++ b/paddle/fluid/operators/detail/grpc_service.h
--- a/paddle/fluid/operators/detail/macros.h
+++ b/paddle/fluid/operators/detail/macros.h
--- a/paddle/fluid/operators/detail/proto_encoder_helper.h
+++ b/paddle/fluid/operators/detail/proto_encoder_helper.h
--- a/paddle/fluid/operators/detail/request_handler.h
+++ b/paddle/fluid/operators/detail/request_handler.h
--- a/paddle/fluid/operators/detail/request_handler_impl.cc
+++ b/paddle/fluid/operators/detail/request_handler_impl.cc
--- a/paddle/fluid/operators/detail/request_handler_impl.h
+++ b/paddle/fluid/operators/detail/request_handler_impl.h
--- a/paddle/fluid/operators/detail/rpc_client.cc
+++ b/paddle/fluid/operators/detail/rpc_client.cc
--- a/paddle/fluid/operators/detail/rpc_client.h
+++ b/paddle/fluid/operators/detail/rpc_client.h
--- a/paddle/fluid/operators/detail/rpc_server.cc
+++ b/paddle/fluid/operators/detail/rpc_server.cc
--- a/paddle/fluid/operators/detail/rpc_server.h
+++ b/paddle/fluid/operators/detail/rpc_server.h
--- a/paddle/fluid/operators/detail/rpc_server_test.cc
+++ b/paddle/fluid/operators/detail/rpc_server_test.cc
--- a/paddle/fluid/operators/detail/send_recv.proto
+++ b/paddle/fluid/operators/detail/send_recv.proto
--- a/paddle/fluid/operators/detail/sendrecvop_utils.cc
+++ b/paddle/fluid/operators/detail/sendrecvop_utils.cc
--- a/paddle/fluid/operators/detail/sendrecvop_utils.h
+++ b/paddle/fluid/operators/detail/sendrecvop_utils.h
--- a/paddle/fluid/operators/detail/variable_response.cc
+++ b/paddle/fluid/operators/detail/variable_response.cc
--- a/paddle/fluid/operators/detail/variable_response.h
+++ b/paddle/fluid/operators/detail/variable_response.h
--- a/paddle/fluid/operators/detection/CMakeLists.txt
+++ b/paddle/fluid/operators/detection/CMakeLists.txt
--- a/paddle/fluid/operators/detection/anchor_generator_op.cc
+++ b/paddle/fluid/operators/detection/anchor_generator_op.cc
--- a/paddle/fluid/operators/detection/anchor_generator_op.cu
+++ b/paddle/fluid/operators/detection/anchor_generator_op.cu
--- a/paddle/fluid/operators/detection/anchor_generator_op.h
+++ b/paddle/fluid/operators/detection/anchor_generator_op.h
--- a/paddle/fluid/operators/detection/bipartite_match_op.cc
+++ b/paddle/fluid/operators/detection/bipartite_match_op.cc
--- a/paddle/fluid/operators/detection/box_coder_op.cc
+++ b/paddle/fluid/operators/detection/box_coder_op.cc
--- a/paddle/fluid/operators/detection/iou_similarity_op.cc
+++ b/paddle/fluid/operators/detection/iou_similarity_op.cc
--- a/paddle/fluid/operators/detection/polygon_box_transform_op.cc
+++ b/paddle/fluid/operators/detection/polygon_box_transform_op.cc
--- a/paddle/fluid/operators/detection_map_op.cc
+++ b/paddle/fluid/operators/detection_map_op.cc
--- a/paddle/fluid/operators/distributed/CMakeLists.txt
+++ b/paddle/fluid/operators/distributed/CMakeLists.txt
--- a/paddle/fluid/operators/distributed/brpc_client.cc
+++ b/paddle/fluid/operators/distributed/brpc_client.cc
--- a/paddle/fluid/operators/distributed/brpc_client.h
+++ b/paddle/fluid/operators/distributed/brpc_client.h
--- a/paddle/fluid/operators/distributed/brpc_server.cc
+++ b/paddle/fluid/operators/distributed/brpc_server.cc
--- a/paddle/fluid/operators/distributed/brpc_server.h
+++ b/paddle/fluid/operators/distributed/brpc_server.h
--- a/paddle/fluid/operators/distributed/bytebuffer_stream.cc
+++ b/paddle/fluid/operators/distributed/bytebuffer_stream.cc
--- a/paddle/fluid/operators/distributed/bytebuffer_stream.h
+++ b/paddle/fluid/operators/distributed/bytebuffer_stream.h
--- a/paddle/fluid/operators/distributed/grpc_client.cc
+++ b/paddle/fluid/operators/distributed/grpc_client.cc
--- a/paddle/fluid/operators/distributed/grpc_client.h
+++ b/paddle/fluid/operators/distributed/grpc_client.h
--- a/paddle/fluid/operators/distributed/grpc_serde_test.cc
+++ b/paddle/fluid/operators/distributed/grpc_serde_test.cc
--- a/paddle/fluid/operators/distributed/grpc_server.cc
+++ b/paddle/fluid/operators/distributed/grpc_server.cc
--- a/paddle/fluid/operators/distributed/grpc_server.h
+++ b/paddle/fluid/operators/distributed/grpc_server.h
--- a/paddle/fluid/operators/distributed/grpc_service.h
+++ b/paddle/fluid/operators/distributed/grpc_service.h
--- a/paddle/fluid/operators/distributed/proto_encoder_helper.h
+++ b/paddle/fluid/operators/distributed/proto_encoder_helper.h
--- a/paddle/fluid/operators/distributed/request_handler.h
+++ b/paddle/fluid/operators/distributed/request_handler.h
--- a/paddle/fluid/operators/distributed/request_handler_impl.cc
+++ b/paddle/fluid/operators/distributed/request_handler_impl.cc
--- a/paddle/fluid/operators/distributed/request_handler_impl.h
+++ b/paddle/fluid/operators/distributed/request_handler_impl.h
--- a/paddle/fluid/operators/distributed/rpc_client.cc
+++ b/paddle/fluid/operators/distributed/rpc_client.cc
--- a/paddle/fluid/operators/distributed/rpc_client.h
+++ b/paddle/fluid/operators/distributed/rpc_client.h
--- a/paddle/fluid/operators/distributed/rpc_server.cc
+++ b/paddle/fluid/operators/distributed/rpc_server.cc
--- a/paddle/fluid/operators/distributed/rpc_server.h
+++ b/paddle/fluid/operators/distributed/rpc_server.h
--- a/paddle/fluid/operators/distributed/rpc_server_test.cc
+++ b/paddle/fluid/operators/distributed/rpc_server_test.cc
--- a/paddle/fluid/operators/distributed/send_recv.proto
+++ b/paddle/fluid/operators/distributed/send_recv.proto
--- a/paddle/fluid/operators/distributed/sendrecvop_utils.cc
+++ b/paddle/fluid/operators/distributed/sendrecvop_utils.cc
--- a/paddle/fluid/operators/distributed/sendrecvop_utils.h
+++ b/paddle/fluid/operators/distributed/sendrecvop_utils.h
--- a/paddle/fluid/operators/distributed/variable_response.cc
+++ b/paddle/fluid/operators/distributed/variable_response.cc
--- a/paddle/fluid/operators/distributed/variable_response.h
+++ b/paddle/fluid/operators/distributed/variable_response.h
--- a/paddle/fluid/operators/elementwise_add_mkldnn_op.cc
+++ b/paddle/fluid/operators/elementwise_add_mkldnn_op.cc
--- a/paddle/fluid/operators/elementwise_mul_op.cc
+++ b/paddle/fluid/operators/elementwise_mul_op.cc
--- a/paddle/fluid/operators/elementwise_op.h
+++ b/paddle/fluid/operators/elementwise_op.h
--- a/paddle/fluid/operators/fetch_barrier_op.cc
+++ b/paddle/fluid/operators/fetch_barrier_op.cc
--- a/paddle/fluid/operators/gaussian_random_batch_size_like_op.cc
+++ b/paddle/fluid/operators/gaussian_random_batch_size_like_op.cc
--- a/paddle/fluid/operators/gaussian_random_mkldnn_op.cc
+++ b/paddle/fluid/operators/gaussian_random_mkldnn_op.cc
--- a/paddle/fluid/operators/gaussian_random_op.cc
+++ b/paddle/fluid/operators/gaussian_random_op.cc
--- a/paddle/fluid/operators/gen_nccl_id_op.cc
+++ b/paddle/fluid/operators/gen_nccl_id_op.cc
--- a/paddle/fluid/operators/get_places_op.cc
+++ b/paddle/fluid/operators/get_places_op.cc
--- a/paddle/fluid/operators/layer_norm_op.cc
+++ b/paddle/fluid/operators/layer_norm_op.cc
--- a/paddle/fluid/operators/linear_chain_crf_op.cc
+++ b/paddle/fluid/operators/linear_chain_crf_op.cc
--- a/paddle/fluid/operators/listen_and_serv_op.cc
+++ b/paddle/fluid/operators/listen_and_serv_op.cc
--- a/paddle/fluid/operators/listen_and_serv_op.h
+++ b/paddle/fluid/operators/listen_and_serv_op.h
--- a/paddle/fluid/operators/load_op.cc
+++ b/paddle/fluid/operators/load_op.cc
--- a/paddle/fluid/operators/logical_op.cc
+++ b/paddle/fluid/operators/logical_op.cc
--- a/paddle/fluid/operators/lstm_op.cc
+++ b/paddle/fluid/operators/lstm_op.cc
--- a/paddle/fluid/operators/math/blas.h
+++ b/paddle/fluid/operators/math/blas.h
--- a/paddle/fluid/operators/math/blas_impl.h
+++ b/paddle/fluid/operators/math/blas_impl.h
--- a/paddle/fluid/operators/math/concat.cc
+++ b/paddle/fluid/operators/math/concat.cc
--- a/paddle/fluid/operators/math/concat.cu
+++ b/paddle/fluid/operators/math/concat.cu
--- a/paddle/fluid/operators/math/concat.h
+++ b/paddle/fluid/operators/math/concat.h
--- a/paddle/fluid/operators/math/detail/avx_functions.cc
+++ b/paddle/fluid/operators/math/detail/avx_functions.cc
--- a/paddle/fluid/operators/math/math_function.cc
+++ b/paddle/fluid/operators/math/math_function.cc
--- a/paddle/fluid/operators/math/math_function.h
+++ b/paddle/fluid/operators/math/math_function.h
--- a/paddle/fluid/operators/mean_iou_op.cc
+++ b/paddle/fluid/operators/mean_iou_op.cc
--- a/paddle/fluid/operators/mean_iou_op.cu
+++ b/paddle/fluid/operators/mean_iou_op.cu
--- a/paddle/fluid/operators/mean_iou_op.h
+++ b/paddle/fluid/operators/mean_iou_op.h
--- a/paddle/fluid/operators/mean_op.cc
+++ b/paddle/fluid/operators/mean_op.cc
--- a/paddle/fluid/operators/merge_ids_op.cc
+++ b/paddle/fluid/operators/merge_ids_op.cc
--- a/paddle/fluid/operators/merge_ids_op.h
+++ b/paddle/fluid/operators/merge_ids_op.h
--- a/paddle/fluid/operators/multiplex_op.cc
+++ b/paddle/fluid/operators/multiplex_op.cc
--- a/paddle/fluid/operators/nce_op.cc
+++ b/paddle/fluid/operators/nce_op.cc
--- a/paddle/fluid/operators/parallel_do_op.cc
+++ b/paddle/fluid/operators/parallel_do_op.cc
--- a/paddle/fluid/operators/pool_op.cc
+++ b/paddle/fluid/operators/pool_op.cc
--- a/paddle/fluid/operators/prefetch_op.cc
+++ b/paddle/fluid/operators/prefetch_op.cc
--- a/paddle/fluid/operators/print_op.cc
+++ b/paddle/fluid/operators/print_op.cc
--- a/paddle/fluid/operators/random_crop_op.cc
+++ b/paddle/fluid/operators/random_crop_op.cc
--- a/paddle/fluid/operators/random_crop_op.h
+++ b/paddle/fluid/operators/random_crop_op.h
--- a/paddle/fluid/operators/reader/CMakeLists.txt
+++ b/paddle/fluid/operators/reader/CMakeLists.txt
--- a/paddle/fluid/operators/reader/blocking_queue.h
+++ b/paddle/fluid/operators/reader/blocking_queue.h
--- a/paddle/fluid/operators/reader/create_batch_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_batch_reader_op.cc
--- a/paddle/fluid/operators/reader/create_custom_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_custom_reader_op.cc
--- a/paddle/fluid/operators/reader/create_double_buffer_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_double_buffer_reader_op.cc
--- a/paddle/fluid/operators/reader/create_multi_pass_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_multi_pass_reader_op.cc
--- a/paddle/fluid/operators/reader/create_py_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_py_reader_op.cc
--- a/paddle/fluid/operators/reader/create_recordio_file_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_recordio_file_reader_op.cc
--- a/paddle/fluid/operators/reader/create_shuffle_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_shuffle_reader_op.cc
--- a/paddle/fluid/operators/reader/create_threaded_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_threaded_reader_op.cc
--- a/paddle/fluid/operators/reader/lod_tensor_blocking_queue.h
+++ b/paddle/fluid/operators/reader/lod_tensor_blocking_queue.h
--- a/paddle/fluid/operators/reader/reader_op_registry.cc
+++ b/paddle/fluid/operators/reader/reader_op_registry.cc
--- a/paddle/fluid/operators/recurrent_op.cc
+++ b/paddle/fluid/operators/recurrent_op.cc
--- a/paddle/fluid/operators/recv_op.cc
+++ b/paddle/fluid/operators/recv_op.cc
--- a/paddle/fluid/operators/roi_pool_op.cc
+++ b/paddle/fluid/operators/roi_pool_op.cc
--- a/paddle/fluid/operators/row_conv_op.cc
+++ b/paddle/fluid/operators/row_conv_op.cc
--- a/paddle/fluid/operators/save_load_op_test.cc
+++ b/paddle/fluid/operators/save_load_op_test.cc
--- a/paddle/fluid/operators/save_op.cc
+++ b/paddle/fluid/operators/save_op.cc
--- a/paddle/fluid/operators/scale_op.cc
+++ b/paddle/fluid/operators/scale_op.cc
--- a/paddle/fluid/operators/send_barrier_op.cc
+++ b/paddle/fluid/operators/send_barrier_op.cc
--- a/paddle/fluid/operators/send_op.cc
+++ b/paddle/fluid/operators/send_op.cc
--- a/paddle/fluid/operators/send_recv_op_test.cc
+++ b/paddle/fluid/operators/send_recv_op_test.cc
--- a/paddle/fluid/operators/sequence_expand_op.h
+++ b/paddle/fluid/operators/sequence_expand_op.h
--- a/paddle/fluid/operators/shape_op.cc
+++ b/paddle/fluid/operators/shape_op.cc
--- a/paddle/fluid/operators/sigmoid_cross_entropy_with_logits_op.cc
+++ b/paddle/fluid/operators/sigmoid_cross_entropy_with_logits_op.cc
--- a/paddle/fluid/operators/slice_op.cc
+++ b/paddle/fluid/operators/slice_op.cc
--- a/paddle/fluid/operators/softmax_mkldnn_op.cc
+++ b/paddle/fluid/operators/softmax_mkldnn_op.cc
--- a/paddle/fluid/operators/softmax_op.cc
+++ b/paddle/fluid/operators/softmax_op.cc
--- a/paddle/fluid/operators/split_op.cc
+++ b/paddle/fluid/operators/split_op.cc
--- a/paddle/fluid/operators/split_op.cu.cc
+++ b/paddle/fluid/operators/split_op.cu.cc
--- a/paddle/fluid/operators/sum_mkldnn_op.cc
+++ b/paddle/fluid/operators/sum_mkldnn_op.cc
--- a/paddle/fluid/operators/sum_op.cc
+++ b/paddle/fluid/operators/sum_op.cc
--- a/paddle/fluid/operators/tensor_array_read_write_op.cc
+++ b/paddle/fluid/operators/tensor_array_read_write_op.cc
--- a/paddle/fluid/operators/tensorrt_engine_op.cc
+++ b/paddle/fluid/operators/tensorrt_engine_op.cc
--- a/paddle/fluid/operators/tensorrt_engine_op.h
+++ b/paddle/fluid/operators/tensorrt_engine_op.h
--- a/paddle/fluid/operators/tensorrt_engine_op_test.cc
+++ b/paddle/fluid/operators/tensorrt_engine_op_test.cc
--- a/paddle/fluid/operators/test_send_nccl_id.cc
+++ b/paddle/fluid/operators/test_send_nccl_id.cc
--- a/paddle/fluid/operators/uniform_random_batch_size_like_op.cc
+++ b/paddle/fluid/operators/uniform_random_batch_size_like_op.cc
--- a/paddle/fluid/operators/uniform_random_op.cc
+++ b/paddle/fluid/operators/uniform_random_op.cc
--- a/paddle/fluid/operators/while_op.cc
+++ b/paddle/fluid/operators/while_op.cc
--- a/paddle/fluid/platform/cpu_info.cc
+++ b/paddle/fluid/platform/cpu_info.cc
--- a/paddle/fluid/platform/device_tracer.cc
+++ b/paddle/fluid/platform/device_tracer.cc
--- a/paddle/fluid/platform/dynload/CMakeLists.txt
+++ b/paddle/fluid/platform/dynload/CMakeLists.txt
--- a/paddle/fluid/platform/dynload/cupti.h
+++ b/paddle/fluid/platform/dynload/cupti.h
--- a/paddle/fluid/platform/dynload/dynamic_loader.cc
+++ b/paddle/fluid/platform/dynload/dynamic_loader.cc
--- a/paddle/fluid/platform/dynload/dynamic_loader.h
+++ b/paddle/fluid/platform/dynload/dynamic_loader.h
--- a/paddle/fluid/platform/dynload/mklml.cc
+++ b/paddle/fluid/platform/dynload/mklml.cc
--- a/paddle/fluid/platform/dynload/mklml.h
+++ b/paddle/fluid/platform/dynload/mklml.h
--- a/paddle/fluid/platform/enforce.h
+++ b/paddle/fluid/platform/enforce.h
--- a/paddle/fluid/platform/mkldnn_helper.h
+++ b/paddle/fluid/platform/mkldnn_helper.h
--- a/paddle/fluid/platform/nccl_helper.h
+++ b/paddle/fluid/platform/nccl_helper.h
--- a/paddle/fluid/pybind/protobuf.cc
+++ b/paddle/fluid/pybind/protobuf.cc
--- a/paddle/fluid/pybind/pybind.cc
+++ b/paddle/fluid/pybind/pybind.cc
--- a/paddle/fluid/pybind/tensor_py.h
+++ b/paddle/fluid/pybind/tensor_py.h
--- a/paddle/fluid/string/printf.h
+++ b/paddle/fluid/string/printf.h
--- a/paddle/function/BufferArg.cpp
+++ b/paddle/function/BufferArg.cpp
--- a/paddle/function/BufferArg.h
+++ b/paddle/function/BufferArg.h
--- a/paddle/function/BufferArgTest.cpp
+++ b/paddle/function/BufferArgTest.cpp
--- a/paddle/function/ContextProjectionOp.cpp
+++ b/paddle/function/ContextProjectionOp.cpp
--- a/paddle/function/ContextProjectionOpTest.cpp
+++ b/paddle/function/ContextProjectionOpTest.cpp
--- a/paddle/function/CosSimOp.cpp
+++ b/paddle/function/CosSimOp.cpp
--- a/paddle/function/CosSimOpTest.cpp
+++ b/paddle/function/CosSimOpTest.cpp
--- a/paddle/function/CropOp.cpp
+++ b/paddle/function/CropOp.cpp
--- a/paddle/function/CrossMapNormalOp.cpp
+++ b/paddle/function/CrossMapNormalOp.cpp
--- a/paddle/function/DepthwiseConvOpGpu.cu
+++ b/paddle/function/DepthwiseConvOpGpu.cu
--- a/paddle/function/EigenGemm.cpp
+++ b/paddle/function/EigenGemm.cpp
--- a/paddle/function/Function.h
+++ b/paddle/function/Function.h
--- a/paddle/function/FunctionTest.cpp
+++ b/paddle/function/FunctionTest.cpp
--- a/paddle/function/FunctionTest.h
+++ b/paddle/function/FunctionTest.h
--- a/paddle/function/GemmConvOp.cpp
+++ b/paddle/function/GemmConvOp.cpp
--- a/paddle/function/GemmFunctor.cpp
+++ b/paddle/function/GemmFunctor.cpp
--- a/paddle/function/Im2ColTest.cpp
+++ b/paddle/function/Im2ColTest.cpp
--- a/paddle/function/MulOp.cpp
+++ b/paddle/function/MulOp.cpp
--- a/paddle/function/MulOp.h
+++ b/paddle/function/MulOp.h
--- a/paddle/function/MulOpGpu.cu
+++ b/paddle/function/MulOpGpu.cu
--- a/paddle/function/MulOpTest.cpp
+++ b/paddle/function/MulOpTest.cpp
--- a/paddle/function/PadOp.cpp
+++ b/paddle/function/PadOp.cpp
--- a/paddle/function/RowConvOp.cpp
+++ b/paddle/function/RowConvOp.cpp
--- a/paddle/function/RowConvOpGpu.cu
+++ b/paddle/function/RowConvOpGpu.cu
--- a/paddle/function/ScaleSubRegionOp.cpp
+++ b/paddle/function/ScaleSubRegionOp.cpp
--- a/paddle/function/SwitchOp.cpp
+++ b/paddle/function/SwitchOp.cpp
--- a/paddle/function/TensorType.h
+++ b/paddle/function/TensorType.h
--- a/paddle/function/neon/NeonDepthwiseConv.cpp
+++ b/paddle/function/neon/NeonDepthwiseConv.cpp
--- a/paddle/function/neon/NeonDepthwiseConvTranspose.cpp
+++ b/paddle/function/neon/NeonDepthwiseConvTranspose.cpp
--- a/paddle/function/nnpack/NNPACKConvOp.cpp
+++ b/paddle/function/nnpack/NNPACKConvOp.cpp
--- a/paddle/function/nnpack/NNPACKConvOpTest.cpp
+++ b/paddle/function/nnpack/NNPACKConvOpTest.cpp
--- a/paddle/gserver/activations/ActivationFunction.cpp
+++ b/paddle/gserver/activations/ActivationFunction.cpp
--- a/paddle/gserver/activations/MKLDNNActivation.h
+++ b/paddle/gserver/activations/MKLDNNActivation.h
--- a/paddle/gserver/dataproviders/DataProvider.h
+++ b/paddle/gserver/dataproviders/DataProvider.h
--- a/paddle/gserver/evaluators/CTCErrorEvaluator.cpp
+++ b/paddle/gserver/evaluators/CTCErrorEvaluator.cpp
--- a/paddle/gserver/evaluators/ChunkEvaluator.cpp
+++ b/paddle/gserver/evaluators/ChunkEvaluator.cpp
--- a/paddle/gserver/evaluators/DetectionMAPEvaluator.cpp
+++ b/paddle/gserver/evaluators/DetectionMAPEvaluator.cpp
--- a/paddle/gserver/evaluators/Evaluator.cpp
+++ b/paddle/gserver/evaluators/Evaluator.cpp
--- a/paddle/gserver/evaluators/Evaluator.h
+++ b/paddle/gserver/evaluators/Evaluator.h
--- a/paddle/gserver/gradientmachines/GradientMachine.h
+++ b/paddle/gserver/gradientmachines/GradientMachine.h
--- a/paddle/gserver/gradientmachines/NeuralNetwork.cpp
+++ b/paddle/gserver/gradientmachines/NeuralNetwork.cpp
--- a/paddle/gserver/gradientmachines/NeuralNetwork.h
+++ b/paddle/gserver/gradientmachines/NeuralNetwork.h
--- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp
+++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.cpp
--- a/paddle/gserver/layers/AddtoLayer.h
+++ b/paddle/gserver/layers/AddtoLayer.h
--- a/paddle/gserver/layers/AgentLayer.h
+++ b/paddle/gserver/layers/AgentLayer.h
--- a/paddle/gserver/layers/AverageLayer.h
+++ b/paddle/gserver/layers/AverageLayer.h
--- a/paddle/gserver/layers/BilinearInterpLayer.h
+++ b/paddle/gserver/layers/BilinearInterpLayer.h
--- a/paddle/gserver/layers/BlockExpandLayer.h
+++ b/paddle/gserver/layers/BlockExpandLayer.h
--- a/paddle/gserver/layers/Conv3DLayer.h
+++ b/paddle/gserver/layers/Conv3DLayer.h
--- a/paddle/gserver/layers/ConvBaseLayer.cpp
+++ b/paddle/gserver/layers/ConvBaseLayer.cpp
--- a/paddle/gserver/layers/ConvBaseLayer.h
+++ b/paddle/gserver/layers/ConvBaseLayer.h
--- a/paddle/gserver/layers/ConvBaseOperator.cpp
+++ b/paddle/gserver/layers/ConvBaseOperator.cpp
--- a/paddle/gserver/layers/ConvBaseOperator.h
+++ b/paddle/gserver/layers/ConvBaseOperator.h
--- a/paddle/gserver/layers/ConvBaseProjection.h
+++ b/paddle/gserver/layers/ConvBaseProjection.h
--- a/paddle/gserver/layers/ConvOperator.cpp
+++ b/paddle/gserver/layers/ConvOperator.cpp
--- a/paddle/gserver/layers/ConvOperator.h
+++ b/paddle/gserver/layers/ConvOperator.h
--- a/paddle/gserver/layers/ConvProjection.h
+++ b/paddle/gserver/layers/ConvProjection.h
--- a/paddle/gserver/layers/ConvShiftLayer.cpp
+++ b/paddle/gserver/layers/ConvShiftLayer.cpp
--- a/paddle/gserver/layers/ConvTransOperator.cpp
+++ b/paddle/gserver/layers/ConvTransOperator.cpp
--- a/paddle/gserver/layers/ConvTransOperator.h
+++ b/paddle/gserver/layers/ConvTransOperator.h
--- a/paddle/gserver/layers/ConvTransProjection.h
+++ b/paddle/gserver/layers/ConvTransProjection.h
--- a/paddle/gserver/layers/ConvexCombinationLayer.cpp
+++ b/paddle/gserver/layers/ConvexCombinationLayer.cpp
--- a/paddle/gserver/layers/CosSimLayer.h
+++ b/paddle/gserver/layers/CosSimLayer.h
--- a/paddle/gserver/layers/CosSimVecMatLayer.cpp
+++ b/paddle/gserver/layers/CosSimVecMatLayer.cpp
--- a/paddle/gserver/layers/CostLayer.cpp
+++ b/paddle/gserver/layers/CostLayer.cpp
--- a/paddle/gserver/layers/CrossChannelNormLayer.cpp
+++ b/paddle/gserver/layers/CrossChannelNormLayer.cpp
--- a/paddle/gserver/layers/CudnnBatchNormLayer.cpp
+++ b/paddle/gserver/layers/CudnnBatchNormLayer.cpp
--- a/paddle/gserver/layers/CudnnConvBaseLayer.h
+++ b/paddle/gserver/layers/CudnnConvBaseLayer.h
--- a/paddle/gserver/layers/CudnnPoolLayer.cpp
+++ b/paddle/gserver/layers/CudnnPoolLayer.cpp
--- a/paddle/gserver/layers/DataNormLayer.h
+++ b/paddle/gserver/layers/DataNormLayer.h
--- a/paddle/gserver/layers/DeConv3DLayer.h
+++ b/paddle/gserver/layers/DeConv3DLayer.h
--- a/paddle/gserver/layers/DetectionUtil.h
+++ b/paddle/gserver/layers/DetectionUtil.h
--- a/paddle/gserver/layers/DotProdLayer.cpp
+++ b/paddle/gserver/layers/DotProdLayer.cpp
--- a/paddle/gserver/layers/ExpandConvLayer.h
+++ b/paddle/gserver/layers/ExpandConvLayer.h
--- a/paddle/gserver/layers/ExpandLayer.h
+++ b/paddle/gserver/layers/ExpandLayer.h
--- a/paddle/gserver/layers/FactorizationMachineLayer.cpp
+++ b/paddle/gserver/layers/FactorizationMachineLayer.cpp
--- a/paddle/gserver/layers/FactorizationMachineLayer.h
+++ b/paddle/gserver/layers/FactorizationMachineLayer.h
--- a/paddle/gserver/layers/FeatureMapExpandLayer.cpp
+++ b/paddle/gserver/layers/FeatureMapExpandLayer.cpp
--- a/paddle/gserver/layers/FullyConnectedLayer.cpp
+++ b/paddle/gserver/layers/FullyConnectedLayer.cpp
--- a/paddle/gserver/layers/FullyConnectedLayer.h
+++ b/paddle/gserver/layers/FullyConnectedLayer.h
--- a/paddle/gserver/layers/GatedRecurrentLayer.h
+++ b/paddle/gserver/layers/GatedRecurrentLayer.h
--- a/paddle/gserver/layers/GruCompute.cpp
+++ b/paddle/gserver/layers/GruCompute.cpp
--- a/paddle/gserver/layers/InterpolationLayer.cpp
+++ b/paddle/gserver/layers/InterpolationLayer.cpp
--- a/paddle/gserver/layers/L2DistanceLayer.h
+++ b/paddle/gserver/layers/L2DistanceLayer.h
--- a/paddle/gserver/layers/Layer.cpp
+++ b/paddle/gserver/layers/Layer.cpp
--- a/paddle/gserver/layers/Layer.h
+++ b/paddle/gserver/layers/Layer.h
--- a/paddle/gserver/layers/LinearChainCRF.h
+++ b/paddle/gserver/layers/LinearChainCRF.h
--- a/paddle/gserver/layers/LinearChainCTC.h
+++ b/paddle/gserver/layers/LinearChainCTC.h
--- a/paddle/gserver/layers/LstmLayer.cpp
+++ b/paddle/gserver/layers/LstmLayer.cpp
--- a/paddle/gserver/layers/LstmLayer.h
+++ b/paddle/gserver/layers/LstmLayer.h
--- a/paddle/gserver/layers/MDLstmLayer.cpp
+++ b/paddle/gserver/layers/MDLstmLayer.cpp
--- a/paddle/gserver/layers/MKLDNNConvLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNConvLayer.cpp
--- a/paddle/gserver/layers/MKLDNNLayer.h
+++ b/paddle/gserver/layers/MKLDNNLayer.h
--- a/paddle/gserver/layers/MKLDNNPoolLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNPoolLayer.cpp
--- a/paddle/gserver/layers/MKLPackedWeight.h
+++ b/paddle/gserver/layers/MKLPackedWeight.h
--- a/paddle/gserver/layers/MaxLayer.h
+++ b/paddle/gserver/layers/MaxLayer.h
--- a/paddle/gserver/layers/MaxOutLayer.h
+++ b/paddle/gserver/layers/MaxOutLayer.h
--- a/paddle/gserver/layers/MaxPoolWithMaskLayer.h
+++ b/paddle/gserver/layers/MaxPoolWithMaskLayer.h
--- a/paddle/gserver/layers/MultiplexLayer.cpp
+++ b/paddle/gserver/layers/MultiplexLayer.cpp
--- a/paddle/gserver/layers/NCELayer.cpp
+++ b/paddle/gserver/layers/NCELayer.cpp
--- a/paddle/gserver/layers/NormLayer.h
+++ b/paddle/gserver/layers/NormLayer.h
--- a/paddle/gserver/layers/NormProjectionLayer.h
+++ b/paddle/gserver/layers/NormProjectionLayer.h
--- a/paddle/gserver/layers/Operator.h
+++ b/paddle/gserver/layers/Operator.h
--- a/paddle/gserver/layers/OuterProdLayer.cpp
+++ b/paddle/gserver/layers/OuterProdLayer.cpp
--- a/paddle/gserver/layers/ParameterReluLayer.h
+++ b/paddle/gserver/layers/ParameterReluLayer.h
--- a/paddle/gserver/layers/Pool3DLayer.h
+++ b/paddle/gserver/layers/Pool3DLayer.h
--- a/paddle/gserver/layers/PoolLayer.h
+++ b/paddle/gserver/layers/PoolLayer.h
--- a/paddle/gserver/layers/PoolProjection.h
+++ b/paddle/gserver/layers/PoolProjection.h
--- a/paddle/gserver/layers/PoolProjectionLayer.h
+++ b/paddle/gserver/layers/PoolProjectionLayer.h
--- a/paddle/gserver/layers/PowerLayer.cpp
+++ b/paddle/gserver/layers/PowerLayer.cpp
--- a/paddle/gserver/layers/PriorBox.cpp
+++ b/paddle/gserver/layers/PriorBox.cpp
--- a/paddle/gserver/layers/Projection.h
+++ b/paddle/gserver/layers/Projection.h
--- a/paddle/gserver/layers/RecurrentLayerGroup.cpp
+++ b/paddle/gserver/layers/RecurrentLayerGroup.cpp
--- a/paddle/gserver/layers/ResizeLayer.cpp
+++ b/paddle/gserver/layers/ResizeLayer.cpp
--- a/paddle/gserver/layers/RotateLayer.h
+++ b/paddle/gserver/layers/RotateLayer.h
--- a/paddle/gserver/layers/ScalingLayer.cpp
+++ b/paddle/gserver/layers/ScalingLayer.cpp
--- a/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp
+++ b/paddle/gserver/layers/SelectiveFullyConnectedLayer.cpp
--- a/paddle/gserver/layers/SelectiveFullyConnectedLayer.h
+++ b/paddle/gserver/layers/SelectiveFullyConnectedLayer.h
--- a/paddle/gserver/layers/SequenceConcatLayer.cpp
+++ b/paddle/gserver/layers/SequenceConcatLayer.cpp
--- a/paddle/gserver/layers/SequenceLastInstanceLayer.cpp
+++ b/paddle/gserver/layers/SequenceLastInstanceLayer.cpp
--- a/paddle/gserver/layers/SequencePoolLayer.h
+++ b/paddle/gserver/layers/SequencePoolLayer.h
--- a/paddle/gserver/layers/SequenceReshapeLayer.cpp
+++ b/paddle/gserver/layers/SequenceReshapeLayer.cpp
--- a/paddle/gserver/layers/SequenceSliceLayer.cpp
+++ b/paddle/gserver/layers/SequenceSliceLayer.cpp
--- a/paddle/gserver/layers/SequenceToBatch.h
+++ b/paddle/gserver/layers/SequenceToBatch.h
--- a/paddle/gserver/layers/SlopeInterceptLayer.cpp
+++ b/paddle/gserver/layers/SlopeInterceptLayer.cpp
--- a/paddle/gserver/layers/SpatialPyramidPoolLayer.h
+++ b/paddle/gserver/layers/SpatialPyramidPoolLayer.h
--- a/paddle/gserver/layers/SubNestedSequenceLayer.cpp
+++ b/paddle/gserver/layers/SubNestedSequenceLayer.cpp
--- a/paddle/gserver/layers/SubSequenceLayer.cpp
+++ b/paddle/gserver/layers/SubSequenceLayer.cpp
--- a/paddle/gserver/layers/SumToOneNormLayer.cpp
+++ b/paddle/gserver/layers/SumToOneNormLayer.cpp
--- a/paddle/gserver/layers/TensorLayer.h
+++ b/paddle/gserver/layers/TensorLayer.h
--- a/paddle/gserver/layers/TransLayer.h
+++ b/paddle/gserver/layers/TransLayer.h
--- a/paddle/gserver/layers/UpsampleLayer.h
+++ b/paddle/gserver/layers/UpsampleLayer.h
--- a/paddle/gserver/layers/ValidationLayer.h
+++ b/paddle/gserver/layers/ValidationLayer.h
--- a/paddle/gserver/tests/CMakeLists.txt
+++ b/paddle/gserver/tests/CMakeLists.txt
--- a/paddle/gserver/tests/LayerGradUtil.h
+++ b/paddle/gserver/tests/LayerGradUtil.h
--- a/paddle/gserver/tests/MKLDNNTester.cpp
+++ b/paddle/gserver/tests/MKLDNNTester.cpp
--- a/paddle/gserver/tests/MKLDNNTester.h
+++ b/paddle/gserver/tests/MKLDNNTester.h
--- a/paddle/gserver/tests/Sequence/train.list
+++ b/paddle/gserver/tests/Sequence/train.list
--- a/paddle/gserver/tests/Sequence/train.list.nest
+++ b/paddle/gserver/tests/Sequence/train.list.nest
--- a/paddle/gserver/tests/sequence_layer_group.conf
+++ b/paddle/gserver/tests/sequence_layer_group.conf
--- a/paddle/gserver/tests/sequence_lstm.conf
+++ b/paddle/gserver/tests/sequence_lstm.conf
--- a/paddle/gserver/tests/sequence_nest_layer_group.conf
+++ b/paddle/gserver/tests/sequence_nest_layer_group.conf
--- a/paddle/gserver/tests/sequence_nest_rnn.conf
+++ b/paddle/gserver/tests/sequence_nest_rnn.conf
--- a/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf
+++ b/paddle/gserver/tests/sequence_nest_rnn_multi_input.conf
--- a/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
+++ b/paddle/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
--- a/paddle/gserver/tests/sequence_recurrent.py
+++ b/paddle/gserver/tests/sequence_recurrent.py
--- a/paddle/gserver/tests/sequence_recurrent_group.py
+++ b/paddle/gserver/tests/sequence_recurrent_group.py
--- a/paddle/gserver/tests/sequence_rnn.conf
+++ b/paddle/gserver/tests/sequence_rnn.conf
--- a/paddle/gserver/tests/sequence_rnn_matched_inputs.py
+++ b/paddle/gserver/tests/sequence_rnn_matched_inputs.py
--- a/paddle/gserver/tests/sequence_rnn_mixed_inputs.py
+++ b/paddle/gserver/tests/sequence_rnn_mixed_inputs.py
--- a/paddle/gserver/tests/sequence_rnn_multi_input.conf
+++ b/paddle/gserver/tests/sequence_rnn_multi_input.conf
--- a/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py
+++ b/paddle/gserver/tests/sequence_rnn_multi_unequalength_inputs.py
--- a/paddle/gserver/tests/test_ActivationGrad.cpp
+++ b/paddle/gserver/tests/test_ActivationGrad.cpp
--- a/paddle/gserver/tests/test_BatchNorm.cpp
+++ b/paddle/gserver/tests/test_BatchNorm.cpp
--- a/paddle/gserver/tests/test_CRFLayerGrad.cpp
+++ b/paddle/gserver/tests/test_CRFLayerGrad.cpp
--- a/paddle/gserver/tests/test_CompareSparse.cpp
+++ b/paddle/gserver/tests/test_CompareSparse.cpp
--- a/paddle/gserver/tests/test_CompareTwoNets.cpp
+++ b/paddle/gserver/tests/test_CompareTwoNets.cpp
--- a/paddle/gserver/tests/test_ConvTrans.cpp
+++ b/paddle/gserver/tests/test_ConvTrans.cpp
--- a/paddle/gserver/tests/test_ConvUnify.cpp
+++ b/paddle/gserver/tests/test_ConvUnify.cpp
--- a/paddle/gserver/tests/test_CrossEntropyOverBeamGrad.cpp
+++ b/paddle/gserver/tests/test_CrossEntropyOverBeamGrad.cpp
--- a/paddle/gserver/tests/test_KmaxSeqScore.cpp
+++ b/paddle/gserver/tests/test_KmaxSeqScore.cpp
--- a/paddle/gserver/tests/test_LayerGrad.cpp
+++ b/paddle/gserver/tests/test_LayerGrad.cpp
--- a/paddle/gserver/tests/test_LinearChainCRF.cpp
+++ b/paddle/gserver/tests/test_LinearChainCRF.cpp
--- a/paddle/gserver/tests/test_MKLDNN.cpp
+++ b/paddle/gserver/tests/test_MKLDNN.cpp
--- a/paddle/gserver/tests/test_MaxPoolingWithMaskOutput.cpp
+++ b/paddle/gserver/tests/test_MaxPoolingWithMaskOutput.cpp
--- a/paddle/gserver/tests/test_MultinomialSampler.cpp
+++ b/paddle/gserver/tests/test_MultinomialSampler.cpp
--- a/paddle/gserver/tests/test_NetworkCompare.cpp
+++ b/paddle/gserver/tests/test_NetworkCompare.cpp
--- a/paddle/gserver/tests/test_PyDataProvider.cpp
+++ b/paddle/gserver/tests/test_PyDataProvider.cpp
--- a/paddle/gserver/tests/test_PyDataProvider2.cpp
+++ b/paddle/gserver/tests/test_PyDataProvider2.cpp
--- a/paddle/gserver/tests/test_RecurrentGradientMachine.cpp
+++ b/paddle/gserver/tests/test_RecurrentGradientMachine.cpp
--- a/paddle/gserver/tests/test_RecurrentLayer.cpp
+++ b/paddle/gserver/tests/test_RecurrentLayer.cpp
--- a/paddle/gserver/tests/test_SelectiveFCLayer.cpp
+++ b/paddle/gserver/tests/test_SelectiveFCLayer.cpp
--- a/paddle/gserver/tests/test_SeqSliceLayerGrad.cpp
+++ b/paddle/gserver/tests/test_SeqSliceLayerGrad.cpp
--- a/paddle/gserver/tests/test_Upsample.cpp
+++ b/paddle/gserver/tests/test_Upsample.cpp
--- a/paddle/gserver/tests/test_WarpCTCLayer.cpp
+++ b/paddle/gserver/tests/test_WarpCTCLayer.cpp
--- a/paddle/cuda/CMakeLists.txt
+++ b/paddle/cuda/CMakeLists.txt
--- a/paddle/cuda/include/hl_activation_functions.h
+++ b/paddle/cuda/include/hl_activation_functions.h
--- a/paddle/cuda/include/hl_aggregate.h
+++ b/paddle/cuda/include/hl_aggregate.h
--- a/paddle/cuda/include/hl_avx_functions.h
+++ b/paddle/cuda/include/hl_avx_functions.h
--- a/paddle/legacy/cuda/include/hl_base.h
+++ b/paddle/legacy/cuda/include/hl_base.h
--- a/paddle/cuda/include/hl_batch_norm.h
+++ b/paddle/cuda/include/hl_batch_norm.h
--- a/paddle/cuda/include/hl_batch_transpose.h
+++ b/paddle/cuda/include/hl_batch_transpose.h
--- a/paddle/cuda/include/hl_cnn.h
+++ b/paddle/cuda/include/hl_cnn.h
--- a/paddle/cuda/include/hl_cpu_gru.cuh
+++ b/paddle/cuda/include/hl_cpu_gru.cuh
--- a/paddle/cuda/include/hl_cpu_lstm.cuh
+++ b/paddle/cuda/include/hl_cpu_lstm.cuh
--- a/paddle/cuda/include/hl_cpu_matrix_kernel.cuh
+++ b/paddle/cuda/include/hl_cpu_matrix_kernel.cuh
--- a/paddle/cuda/include/hl_cpu_matrix_kernel_detail.cuh
+++ b/paddle/cuda/include/hl_cpu_matrix_kernel_detail.cuh
--- a/paddle/cuda/include/hl_cpu_scalar.cuh
+++ b/paddle/cuda/include/hl_cpu_scalar.cuh
--- a/paddle/cuda/include/hl_cpu_simd_neon.cuh
+++ b/paddle/cuda/include/hl_cpu_simd_neon.cuh
--- a/paddle/cuda/include/hl_cpu_simd_sse.cuh
+++ b/paddle/cuda/include/hl_cpu_simd_sse.cuh
--- a/paddle/cuda/include/hl_cuda.h
+++ b/paddle/cuda/include/hl_cuda.h
--- a/paddle/cuda/include/hl_cuda.ph
+++ b/paddle/cuda/include/hl_cuda.ph
--- a/paddle/cuda/include/hl_cuda_cublas.h
+++ b/paddle/cuda/include/hl_cuda_cublas.h
--- a/paddle/cuda/include/hl_cuda_cudnn.h
+++ b/paddle/cuda/include/hl_cuda_cudnn.h
--- a/paddle/cuda/include/hl_cuda_cudnn.ph
+++ b/paddle/cuda/include/hl_cuda_cudnn.ph
--- a/paddle/cuda/include/hl_device_functions.cuh
+++ b/paddle/cuda/include/hl_device_functions.cuh
--- a/paddle/cuda/include/hl_functions.h
+++ b/paddle/cuda/include/hl_functions.h
--- a/paddle/cuda/include/hl_gpu.h
+++ b/paddle/cuda/include/hl_gpu.h
--- a/paddle/cuda/include/hl_gpu_functions.cuh
+++ b/paddle/cuda/include/hl_gpu_functions.cuh
--- a/paddle/cuda/include/hl_gpu_gru.cuh
+++ b/paddle/cuda/include/hl_gpu_gru.cuh
--- a/paddle/cuda/include/hl_gpu_lstm.cuh
+++ b/paddle/cuda/include/hl_gpu_lstm.cuh
--- a/paddle/cuda/include/hl_gpu_matrix_kernel.cuh
+++ b/paddle/cuda/include/hl_gpu_matrix_kernel.cuh
--- a/paddle/cuda/include/hl_gru_ops.cuh
+++ b/paddle/cuda/include/hl_gru_ops.cuh
--- a/paddle/cuda/include/hl_lstm.h
+++ b/paddle/cuda/include/hl_lstm.h
--- a/paddle/cuda/include/hl_lstm_ops.cuh
+++ b/paddle/cuda/include/hl_lstm_ops.cuh
--- a/paddle/cuda/include/hl_matrix.h
+++ b/paddle/cuda/include/hl_matrix.h
--- a/paddle/cuda/include/hl_matrix_apply.cuh
+++ b/paddle/cuda/include/hl_matrix_apply.cuh
--- a/paddle/cuda/include/hl_matrix_base.cuh
+++ b/paddle/cuda/include/hl_matrix_base.cuh
--- a/paddle/cuda/include/hl_matrix_base_detail.cuh
+++ b/paddle/cuda/include/hl_matrix_base_detail.cuh
--- a/paddle/cuda/include/hl_matrix_ops.cuh
+++ b/paddle/cuda/include/hl_matrix_ops.cuh
--- a/paddle/cuda/include/hl_matrix_type.cuh
+++ b/paddle/cuda/include/hl_matrix_type.cuh
--- a/paddle/cuda/include/hl_perturbation_util.cuh
+++ b/paddle/cuda/include/hl_perturbation_util.cuh
--- a/paddle/cuda/include/hl_recurrent_apply.cuh
+++ b/paddle/cuda/include/hl_recurrent_apply.cuh
--- a/paddle/cuda/include/hl_sequence.h
+++ b/paddle/cuda/include/hl_sequence.h
--- a/paddle/cuda/include/hl_sparse.h
+++ b/paddle/cuda/include/hl_sparse.h
--- a/paddle/cuda/include/hl_sparse.ph
+++ b/paddle/cuda/include/hl_sparse.ph
--- a/paddle/cuda/include/hl_table_apply.h
+++ b/paddle/cuda/include/hl_table_apply.h
--- a/paddle/cuda/include/hl_tensor_ops.h
+++ b/paddle/cuda/include/hl_tensor_ops.h
--- a/paddle/cuda/include/hl_thread.ph
+++ b/paddle/cuda/include/hl_thread.ph
--- a/paddle/cuda/include/hl_time.h
+++ b/paddle/cuda/include/hl_time.h
--- a/paddle/cuda/include/hl_top_k.h
+++ b/paddle/cuda/include/hl_top_k.h
--- a/paddle/cuda/include/hl_warpctc_wrap.h
+++ b/paddle/cuda/include/hl_warpctc_wrap.h
--- a/paddle/cuda/include/stub/hl_aggregate_stub.h
+++ b/paddle/cuda/include/stub/hl_aggregate_stub.h
--- a/paddle/cuda/include/stub/hl_cnn_stub.h
+++ b/paddle/cuda/include/stub/hl_cnn_stub.h
--- a/paddle/cuda/include/stub/hl_cuda_cublas_stub.h
+++ b/paddle/cuda/include/stub/hl_cuda_cublas_stub.h
--- a/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h
+++ b/paddle/cuda/include/stub/hl_cuda_cudnn_stub.h
--- a/paddle/cuda/include/stub/hl_cuda_stub.h
+++ b/paddle/cuda/include/stub/hl_cuda_stub.h
--- a/paddle/cuda/include/stub/hl_lstm_stub.h
+++ b/paddle/cuda/include/stub/hl_lstm_stub.h
--- a/paddle/cuda/include/stub/hl_matrix_stub.h
+++ b/paddle/cuda/include/stub/hl_matrix_stub.h
--- a/paddle/cuda/include/stub/hl_sequence_stub.h
+++ b/paddle/cuda/include/stub/hl_sequence_stub.h
--- a/paddle/cuda/include/stub/hl_sparse_stub.h
+++ b/paddle/cuda/include/stub/hl_sparse_stub.h
--- a/paddle/cuda/src/avx_mathfun.h
+++ b/paddle/cuda/src/avx_mathfun.h
--- a/paddle/cuda/src/hl_avx_functions.cc
+++ b/paddle/cuda/src/hl_avx_functions.cc
--- a/paddle/cuda/src/hl_batch_norm.cu
+++ b/paddle/cuda/src/hl_batch_norm.cu
--- a/paddle/cuda/src/hl_batch_transpose.cu
+++ b/paddle/cuda/src/hl_batch_transpose.cu
--- a/paddle/cuda/src/hl_cpu_functions.cc
+++ b/paddle/cuda/src/hl_cpu_functions.cc
--- a/paddle/cuda/src/hl_cuda_aggregate.cu
+++ b/paddle/cuda/src/hl_cuda_aggregate.cu
--- a/paddle/cuda/src/hl_cuda_cnn.cu
+++ b/paddle/cuda/src/hl_cuda_cnn.cu
--- a/paddle/cuda/src/hl_cuda_cublas.cc
+++ b/paddle/cuda/src/hl_cuda_cublas.cc
--- a/paddle/cuda/src/hl_cuda_cudnn.cc
+++ b/paddle/cuda/src/hl_cuda_cudnn.cc
--- a/paddle/cuda/src/hl_cuda_device.cc
+++ b/paddle/cuda/src/hl_cuda_device.cc
--- a/paddle/cuda/src/hl_cuda_lstm.cu
+++ b/paddle/cuda/src/hl_cuda_lstm.cu
--- a/paddle/cuda/src/hl_cuda_matrix.cu
+++ b/paddle/cuda/src/hl_cuda_matrix.cu
--- a/paddle/cuda/src/hl_cuda_sequence.cu
+++ b/paddle/cuda/src/hl_cuda_sequence.cu
--- a/paddle/cuda/src/hl_cuda_sparse.cu
+++ b/paddle/cuda/src/hl_cuda_sparse.cu
--- a/paddle/cuda/src/hl_cuda_sparse.cuh
+++ b/paddle/cuda/src/hl_cuda_sparse.cuh
--- a/paddle/cuda/src/hl_math.cc
+++ b/paddle/cuda/src/hl_math.cc
--- a/paddle/cuda/src/hl_perturbation_util.cu
+++ b/paddle/cuda/src/hl_perturbation_util.cu
--- a/paddle/cuda/src/hl_table_apply.cu
+++ b/paddle/cuda/src/hl_table_apply.cu
--- a/paddle/cuda/src/hl_time.cc
+++ b/paddle/cuda/src/hl_time.cc
--- a/paddle/legacy/cuda/src/hl_top_k.cu
+++ b/paddle/legacy/cuda/src/hl_top_k.cu
--- a/paddle/cuda/src/hl_warpctc_wrap.cc
+++ b/paddle/cuda/src/hl_warpctc_wrap.cc
--- a/paddle/function/BlockExpandOp.cpp
+++ b/paddle/function/BlockExpandOp.cpp
--- a/paddle/function/BlockExpandOpTest.cpp
+++ b/paddle/function/BlockExpandOpTest.cpp
--- a/paddle/legacy/function/BufferArg.cpp
+++ b/paddle/legacy/function/BufferArg.cpp
--- a/paddle/legacy/function/BufferArg.h
+++ b/paddle/legacy/function/BufferArg.h
--- a/paddle/legacy/function/BufferArgTest.cpp
+++ b/paddle/legacy/function/BufferArgTest.cpp
--- a/paddle/function/CMakeLists.txt
+++ b/paddle/function/CMakeLists.txt
--- a/paddle/legacy/function/ContextProjectionOp.cpp
+++ b/paddle/legacy/function/ContextProjectionOp.cpp
--- a/paddle/function/ContextProjectionOp.h
+++ b/paddle/function/ContextProjectionOp.h
--- a/paddle/function/ContextProjectionOpGpu.cu
+++ b/paddle/function/ContextProjectionOpGpu.cu
--- a/paddle/legacy/function/ContextProjectionOpTest.cpp
+++ b/paddle/legacy/function/ContextProjectionOpTest.cpp
--- a/paddle/function/ConvOp.h
+++ b/paddle/function/ConvOp.h
--- a/paddle/function/ConvOpTest.h
+++ b/paddle/function/ConvOpTest.h
--- a/paddle/legacy/function/CosSimOp.cpp
+++ b/paddle/legacy/function/CosSimOp.cpp
--- a/paddle/function/CosSimOp.h
+++ b/paddle/function/CosSimOp.h
--- a/paddle/function/CosSimOpGpu.cu
+++ b/paddle/function/CosSimOpGpu.cu
--- a/paddle/legacy/function/CosSimOpTest.cpp
+++ b/paddle/legacy/function/CosSimOpTest.cpp
--- a/paddle/legacy/function/CropOp.cpp
+++ b/paddle/legacy/function/CropOp.cpp
--- a/paddle/function/CropOp.h
+++ b/paddle/function/CropOp.h
--- a/paddle/function/CropOpGpu.cu
+++ b/paddle/function/CropOpGpu.cu
--- a/paddle/function/CropOpTest.cpp
+++ b/paddle/function/CropOpTest.cpp
--- a/paddle/legacy/function/CrossMapNormalOp.cpp
+++ b/paddle/legacy/function/CrossMapNormalOp.cpp
--- a/paddle/function/CrossMapNormalOp.h
+++ b/paddle/function/CrossMapNormalOp.h
--- a/paddle/function/CrossMapNormalOpGpu.cu
+++ b/paddle/function/CrossMapNormalOpGpu.cu
--- a/paddle/function/CrossMapNormalOpTest.cpp
+++ b/paddle/function/CrossMapNormalOpTest.cpp
--- a/paddle/function/DepthwiseConvOp.cpp
+++ b/paddle/function/DepthwiseConvOp.cpp
--- a/paddle/function/DepthwiseConvOp.h
+++ b/paddle/function/DepthwiseConvOp.h
--- a/paddle/legacy/function/DepthwiseConvOpGpu.cu
+++ b/paddle/legacy/function/DepthwiseConvOpGpu.cu
--- a/paddle/function/DepthwiseConvOpTest.cpp
+++ b/paddle/function/DepthwiseConvOpTest.cpp
--- a/paddle/legacy/function/EigenGemm.cpp
+++ b/paddle/legacy/function/EigenGemm.cpp
--- a/paddle/function/EigenThreadDevice.h
+++ b/paddle/function/EigenThreadDevice.h
--- a/paddle/function/Function.cpp
+++ b/paddle/function/Function.cpp
--- a/paddle/legacy/function/Function.h
+++ b/paddle/legacy/function/Function.h
--- a/paddle/legacy/function/FunctionTest.cpp
+++ b/paddle/legacy/function/FunctionTest.cpp
--- a/paddle/legacy/function/FunctionTest.h
+++ b/paddle/legacy/function/FunctionTest.h
--- a/paddle/legacy/function/GemmConvOp.cpp
+++ b/paddle/legacy/function/GemmConvOp.cpp
--- a/paddle/function/GemmConvOpTest.cpp
+++ b/paddle/function/GemmConvOpTest.cpp
--- a/paddle/legacy/function/GemmFunctor.cpp
+++ b/paddle/legacy/function/GemmFunctor.cpp
--- a/paddle/function/GemmFunctor.h
+++ b/paddle/function/GemmFunctor.h
--- a/paddle/function/GruFunctor.h
+++ b/paddle/function/GruFunctor.h
--- a/paddle/function/Im2Col.h
+++ b/paddle/function/Im2Col.h
--- a/paddle/function/Im2ColOp.cpp
+++ b/paddle/function/Im2ColOp.cpp
--- a/paddle/function/Im2ColOpGpu.cu
+++ b/paddle/function/Im2ColOpGpu.cu
--- a/paddle/legacy/function/Im2ColTest.cpp
+++ b/paddle/legacy/function/Im2ColTest.cpp
--- a/paddle/legacy/function/MulOp.cpp
+++ b/paddle/legacy/function/MulOp.cpp
--- a/paddle/legacy/function/MulOp.h
+++ b/paddle/legacy/function/MulOp.h
--- a/paddle/legacy/function/MulOpGpu.cu
+++ b/paddle/legacy/function/MulOpGpu.cu
--- a/paddle/legacy/function/MulOpTest.cpp
+++ b/paddle/legacy/function/MulOpTest.cpp
--- a/paddle/function/NaiveConvOp.cpp
+++ b/paddle/function/NaiveConvOp.cpp
--- a/paddle/legacy/function/PadOp.cpp
+++ b/paddle/legacy/function/PadOp.cpp
--- a/paddle/function/PadOp.h
+++ b/paddle/function/PadOp.h
--- a/paddle/function/PadOpGpu.cu
+++ b/paddle/function/PadOpGpu.cu
--- a/paddle/function/PadOpTest.cpp
+++ b/paddle/function/PadOpTest.cpp
--- a/paddle/legacy/function/RowConvOp.cpp
+++ b/paddle/legacy/function/RowConvOp.cpp
--- a/paddle/function/RowConvOp.h
+++ b/paddle/function/RowConvOp.h
--- a/paddle/legacy/function/RowConvOpGpu.cu
+++ b/paddle/legacy/function/RowConvOpGpu.cu
--- a/paddle/function/RowConvOpTest.cpp
+++ b/paddle/function/RowConvOpTest.cpp
--- a/paddle/legacy/function/ScaleSubRegionOp.cpp
+++ b/paddle/legacy/function/ScaleSubRegionOp.cpp
--- a/paddle/function/ScaleSubRegionOp.h
+++ b/paddle/function/ScaleSubRegionOp.h
--- a/paddle/function/ScaleSubRegionOpGpu.cu
+++ b/paddle/function/ScaleSubRegionOpGpu.cu
--- a/paddle/function/ScaleSubRegionOpTest.cpp
+++ b/paddle/function/ScaleSubRegionOpTest.cpp
--- a/paddle/legacy/function/SwitchOp.cpp
+++ b/paddle/legacy/function/SwitchOp.cpp
--- a/paddle/function/SwitchOp.h
+++ b/paddle/function/SwitchOp.h
--- a/paddle/function/SwitchOpGpu.cu
+++ b/paddle/function/SwitchOpGpu.cu
--- a/paddle/function/SwitchOpTest.cpp
+++ b/paddle/function/SwitchOpTest.cpp
--- a/paddle/function/TensorShape.h
+++ b/paddle/function/TensorShape.h
--- a/paddle/function/TensorShapeTest.cpp
+++ b/paddle/function/TensorShapeTest.cpp
--- a/paddle/legacy/function/TensorType.h
+++ b/paddle/legacy/function/TensorType.h
--- a/paddle/function/TensorTypeTest.cpp
+++ b/paddle/function/TensorTypeTest.cpp
--- a/paddle/legacy/function/neon/NeonDepthwiseConv.cpp
+++ b/paddle/legacy/function/neon/NeonDepthwiseConv.cpp
--- a/paddle/function/neon/NeonDepthwiseConv.h
+++ b/paddle/function/neon/NeonDepthwiseConv.h
--- a/paddle/legacy/function/neon/NeonDepthwiseConvTranspose.cpp
+++ b/paddle/legacy/function/neon/NeonDepthwiseConvTranspose.cpp
--- a/paddle/function/neon/neon_util.h
+++ b/paddle/function/neon/neon_util.h
--- a/paddle/legacy/function/nnpack/NNPACKConvOp.cpp
+++ b/paddle/legacy/function/nnpack/NNPACKConvOp.cpp
--- a/paddle/legacy/function/nnpack/NNPACKConvOpTest.cpp
+++ b/paddle/legacy/function/nnpack/NNPACKConvOpTest.cpp
--- a/paddle/gserver/CMakeLists.txt
+++ b/paddle/gserver/CMakeLists.txt
--- a/paddle/legacy/gserver/activations/ActivationFunction.cpp
+++ b/paddle/legacy/gserver/activations/ActivationFunction.cpp
--- a/paddle/gserver/activations/ActivationFunction.h
+++ b/paddle/gserver/activations/ActivationFunction.h
--- a/paddle/gserver/activations/MKLDNNActivation.cpp
+++ b/paddle/gserver/activations/MKLDNNActivation.cpp
--- a/paddle/legacy/gserver/activations/MKLDNNActivation.h
+++ b/paddle/legacy/gserver/activations/MKLDNNActivation.h
--- a/paddle/gserver/dataproviders/DataProvider.cpp
+++ b/paddle/gserver/dataproviders/DataProvider.cpp
--- a/paddle/legacy/gserver/dataproviders/DataProvider.h
+++ b/paddle/legacy/gserver/dataproviders/DataProvider.h
--- a/paddle/gserver/dataproviders/DataProviderGroup.h
+++ b/paddle/gserver/dataproviders/DataProviderGroup.h
--- a/paddle/gserver/dataproviders/MultiDataProvider.cpp
+++ b/paddle/gserver/dataproviders/MultiDataProvider.cpp
--- a/paddle/gserver/dataproviders/MultiDataProvider.h
+++ b/paddle/gserver/dataproviders/MultiDataProvider.h
--- a/paddle/gserver/dataproviders/ProtoReader.h
+++ b/paddle/gserver/dataproviders/ProtoReader.h
--- a/paddle/gserver/dataproviders/PyDataProvider.cpp
+++ b/paddle/gserver/dataproviders/PyDataProvider.cpp
--- a/paddle/gserver/dataproviders/PyDataProvider.h
+++ b/paddle/gserver/dataproviders/PyDataProvider.h
--- a/paddle/gserver/dataproviders/PyDataProvider2.cpp
+++ b/paddle/gserver/dataproviders/PyDataProvider2.cpp
--- a/paddle/legacy/gserver/evaluators/CTCErrorEvaluator.cpp
+++ b/paddle/legacy/gserver/evaluators/CTCErrorEvaluator.cpp
--- a/paddle/legacy/gserver/evaluators/ChunkEvaluator.cpp
+++ b/paddle/legacy/gserver/evaluators/ChunkEvaluator.cpp
--- a/paddle/legacy/gserver/evaluators/DetectionMAPEvaluator.cpp
+++ b/paddle/legacy/gserver/evaluators/DetectionMAPEvaluator.cpp
--- a/paddle/legacy/gserver/evaluators/Evaluator.cpp
+++ b/paddle/legacy/gserver/evaluators/Evaluator.cpp
--- a/paddle/legacy/gserver/evaluators/Evaluator.h
+++ b/paddle/legacy/gserver/evaluators/Evaluator.h
--- a/paddle/gserver/gradientmachines/GradientMachine.cpp
+++ b/paddle/gserver/gradientmachines/GradientMachine.cpp
--- a/paddle/legacy/gserver/gradientmachines/GradientMachine.h
+++ b/paddle/legacy/gserver/gradientmachines/GradientMachine.h
--- a/paddle/gserver/gradientmachines/GradientMachineMode.cpp
+++ b/paddle/gserver/gradientmachines/GradientMachineMode.cpp
--- a/paddle/gserver/gradientmachines/GradientMachineMode.h
+++ b/paddle/gserver/gradientmachines/GradientMachineMode.h
--- a/paddle/gserver/gradientmachines/MultiGradientMachine.cpp
+++ b/paddle/gserver/gradientmachines/MultiGradientMachine.cpp
--- a/paddle/gserver/gradientmachines/MultiGradientMachine.h
+++ b/paddle/gserver/gradientmachines/MultiGradientMachine.h
--- a/paddle/gserver/gradientmachines/MultiNetwork.cpp
+++ b/paddle/gserver/gradientmachines/MultiNetwork.cpp
--- a/paddle/gserver/gradientmachines/MultiNetwork.h
+++ b/paddle/gserver/gradientmachines/MultiNetwork.h
--- a/paddle/legacy/gserver/gradientmachines/NeuralNetwork.cpp
+++ b/paddle/legacy/gserver/gradientmachines/NeuralNetwork.cpp
--- a/paddle/legacy/gserver/gradientmachines/NeuralNetwork.h
+++ b/paddle/legacy/gserver/gradientmachines/NeuralNetwork.h
--- a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp
+++ b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.cpp
--- a/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h
+++ b/paddle/gserver/gradientmachines/ParallelNeuralNetwork.h
--- a/paddle/legacy/gserver/gradientmachines/RecurrentGradientMachine.cpp
+++ b/paddle/legacy/gserver/gradientmachines/RecurrentGradientMachine.cpp
--- a/paddle/gserver/gradientmachines/RecurrentGradientMachine.h
+++ b/paddle/gserver/gradientmachines/RecurrentGradientMachine.h
--- a/paddle/gserver/layers/AddtoLayer.cpp
+++ b/paddle/gserver/layers/AddtoLayer.cpp
--- a/paddle/legacy/gserver/layers/AddtoLayer.h
+++ b/paddle/legacy/gserver/layers/AddtoLayer.h
--- a/paddle/gserver/layers/AgentLayer.cpp
+++ b/paddle/gserver/layers/AgentLayer.cpp
--- a/paddle/legacy/gserver/layers/AgentLayer.h
+++ b/paddle/legacy/gserver/layers/AgentLayer.h
--- a/paddle/gserver/layers/AverageLayer.cpp
+++ b/paddle/gserver/layers/AverageLayer.cpp
--- a/paddle/legacy/gserver/layers/AverageLayer.h
+++ b/paddle/legacy/gserver/layers/AverageLayer.h
--- a/paddle/gserver/layers/BatchNormBaseLayer.cpp
+++ b/paddle/gserver/layers/BatchNormBaseLayer.cpp
--- a/paddle/gserver/layers/BatchNormBaseLayer.h
+++ b/paddle/gserver/layers/BatchNormBaseLayer.h
--- a/paddle/gserver/layers/BatchNormalizationLayer.cpp
+++ b/paddle/gserver/layers/BatchNormalizationLayer.cpp
--- a/paddle/gserver/layers/BatchNormalizationLayer.h
+++ b/paddle/gserver/layers/BatchNormalizationLayer.h
--- a/paddle/gserver/layers/BilinearInterpLayer.cpp
+++ b/paddle/gserver/layers/BilinearInterpLayer.cpp
--- a/paddle/legacy/gserver/layers/BilinearInterpLayer.h
+++ b/paddle/legacy/gserver/layers/BilinearInterpLayer.h
--- a/paddle/gserver/layers/BlockExpandLayer.cpp
+++ b/paddle/gserver/layers/BlockExpandLayer.cpp
--- a/paddle/legacy/gserver/layers/BlockExpandLayer.h
+++ b/paddle/legacy/gserver/layers/BlockExpandLayer.h
--- a/paddle/gserver/layers/CRFDecodingLayer.cpp
+++ b/paddle/gserver/layers/CRFDecodingLayer.cpp
--- a/paddle/gserver/layers/CRFDecodingLayer.h
+++ b/paddle/gserver/layers/CRFDecodingLayer.h
--- a/paddle/gserver/layers/CRFLayer.cpp
+++ b/paddle/gserver/layers/CRFLayer.cpp
--- a/paddle/gserver/layers/CRFLayer.h
+++ b/paddle/gserver/layers/CRFLayer.h
--- a/paddle/gserver/layers/CTCLayer.cpp
+++ b/paddle/gserver/layers/CTCLayer.cpp
--- a/paddle/gserver/layers/CTCLayer.h
+++ b/paddle/gserver/layers/CTCLayer.h
--- a/paddle/gserver/layers/ClipLayer.cpp
+++ b/paddle/gserver/layers/ClipLayer.cpp
--- a/paddle/gserver/layers/ConcatenateLayer.cpp
+++ b/paddle/gserver/layers/ConcatenateLayer.cpp
--- a/paddle/gserver/layers/ContextProjection.cpp
+++ b/paddle/gserver/layers/ContextProjection.cpp
--- a/paddle/gserver/layers/ContextProjection.h
+++ b/paddle/gserver/layers/ContextProjection.h
--- a/paddle/gserver/layers/Conv3DLayer.cpp
+++ b/paddle/gserver/layers/Conv3DLayer.cpp
--- a/paddle/legacy/gserver/layers/Conv3DLayer.h
+++ b/paddle/legacy/gserver/layers/Conv3DLayer.h
--- a/paddle/legacy/gserver/layers/ConvBaseLayer.cpp
+++ b/paddle/legacy/gserver/layers/ConvBaseLayer.cpp
--- a/paddle/legacy/gserver/layers/ConvBaseLayer.h
+++ b/paddle/legacy/gserver/layers/ConvBaseLayer.h
--- a/paddle/legacy/gserver/layers/ConvBaseOperator.cpp
+++ b/paddle/legacy/gserver/layers/ConvBaseOperator.cpp
--- a/paddle/legacy/gserver/layers/ConvBaseOperator.h
+++ b/paddle/legacy/gserver/layers/ConvBaseOperator.h
--- a/paddle/gserver/layers/ConvBaseProjection.cpp
+++ b/paddle/gserver/layers/ConvBaseProjection.cpp
--- a/paddle/legacy/gserver/layers/ConvBaseProjection.h
+++ b/paddle/legacy/gserver/layers/ConvBaseProjection.h
--- a/paddle/legacy/gserver/layers/ConvOperator.cpp
+++ b/paddle/legacy/gserver/layers/ConvOperator.cpp
--- a/paddle/legacy/gserver/layers/ConvOperator.h
+++ b/paddle/legacy/gserver/layers/ConvOperator.h
--- a/paddle/gserver/layers/ConvProjection.cpp
+++ b/paddle/gserver/layers/ConvProjection.cpp
--- a/paddle/legacy/gserver/layers/ConvProjection.h
+++ b/paddle/legacy/gserver/layers/ConvProjection.h
--- a/paddle/legacy/gserver/layers/ConvShiftLayer.cpp
+++ b/paddle/legacy/gserver/layers/ConvShiftLayer.cpp
--- a/paddle/legacy/gserver/layers/ConvTransOperator.cpp
+++ b/paddle/legacy/gserver/layers/ConvTransOperator.cpp
--- a/paddle/legacy/gserver/layers/ConvTransOperator.h
+++ b/paddle/legacy/gserver/layers/ConvTransOperator.h
--- a/paddle/gserver/layers/ConvTransProjection.cpp
+++ b/paddle/gserver/layers/ConvTransProjection.cpp
--- a/paddle/legacy/gserver/layers/ConvTransProjection.h
+++ b/paddle/legacy/gserver/layers/ConvTransProjection.h
--- a/paddle/legacy/gserver/layers/ConvexCombinationLayer.cpp
+++ b/paddle/legacy/gserver/layers/ConvexCombinationLayer.cpp
--- a/paddle/gserver/layers/CosSimLayer.cpp
+++ b/paddle/gserver/layers/CosSimLayer.cpp
--- a/paddle/legacy/gserver/layers/CosSimLayer.h
+++ b/paddle/legacy/gserver/layers/CosSimLayer.h
--- a/paddle/legacy/gserver/layers/CosSimVecMatLayer.cpp
+++ b/paddle/legacy/gserver/layers/CosSimVecMatLayer.cpp
--- a/paddle/legacy/gserver/layers/CostLayer.cpp
+++ b/paddle/legacy/gserver/layers/CostLayer.cpp
--- a/paddle/gserver/layers/CostLayer.h
+++ b/paddle/gserver/layers/CostLayer.h
--- a/paddle/gserver/layers/CropLayer.cpp
+++ b/paddle/gserver/layers/CropLayer.cpp
--- a/paddle/gserver/layers/CropLayer.h
+++ b/paddle/gserver/layers/CropLayer.h
--- a/paddle/legacy/gserver/layers/CrossChannelNormLayer.cpp
+++ b/paddle/legacy/gserver/layers/CrossChannelNormLayer.cpp
--- a/paddle/gserver/layers/CrossEntropyOverBeam.cpp
+++ b/paddle/gserver/layers/CrossEntropyOverBeam.cpp
--- a/paddle/gserver/layers/CrossEntropyOverBeam.h
+++ b/paddle/gserver/layers/CrossEntropyOverBeam.h
--- a/paddle/legacy/gserver/layers/CudnnBatchNormLayer.cpp
+++ b/paddle/legacy/gserver/layers/CudnnBatchNormLayer.cpp
--- a/paddle/gserver/layers/CudnnBatchNormLayer.h
+++ b/paddle/gserver/layers/CudnnBatchNormLayer.h
--- a/paddle/gserver/layers/CudnnConvBaseLayer.cpp
+++ b/paddle/gserver/layers/CudnnConvBaseLayer.cpp
--- a/paddle/legacy/gserver/layers/CudnnConvBaseLayer.h
+++ b/paddle/legacy/gserver/layers/CudnnConvBaseLayer.h
--- a/paddle/legacy/gserver/layers/CudnnPoolLayer.cpp
+++ b/paddle/legacy/gserver/layers/CudnnPoolLayer.cpp
--- a/paddle/gserver/layers/CudnnPoolLayer.h
+++ b/paddle/gserver/layers/CudnnPoolLayer.h
--- a/paddle/gserver/layers/DataLayer.cpp
+++ b/paddle/gserver/layers/DataLayer.cpp
--- a/paddle/gserver/layers/DataLayer.h
+++ b/paddle/gserver/layers/DataLayer.h
--- a/paddle/gserver/layers/DataNormLayer.cpp
+++ b/paddle/gserver/layers/DataNormLayer.cpp
--- a/paddle/legacy/gserver/layers/DataNormLayer.h
+++ b/paddle/legacy/gserver/layers/DataNormLayer.h
--- a/paddle/gserver/layers/DeConv3DLayer.cpp
+++ b/paddle/gserver/layers/DeConv3DLayer.cpp
--- a/paddle/legacy/gserver/layers/DeConv3DLayer.h
+++ b/paddle/legacy/gserver/layers/DeConv3DLayer.h
--- a/paddle/gserver/layers/DetectionOutputLayer.cpp
+++ b/paddle/gserver/layers/DetectionOutputLayer.cpp
--- a/paddle/gserver/layers/DetectionOutputLayer.h
+++ b/paddle/gserver/layers/DetectionOutputLayer.h
--- a/paddle/gserver/layers/DetectionUtil.cpp
+++ b/paddle/gserver/layers/DetectionUtil.cpp
--- a/paddle/legacy/gserver/layers/DetectionUtil.h
+++ b/paddle/legacy/gserver/layers/DetectionUtil.h
--- a/paddle/gserver/layers/DotMulOperator.cpp
+++ b/paddle/gserver/layers/DotMulOperator.cpp
--- a/paddle/gserver/layers/DotMulProjection.cpp
+++ b/paddle/gserver/layers/DotMulProjection.cpp
--- a/paddle/legacy/gserver/layers/DotProdLayer.cpp
+++ b/paddle/legacy/gserver/layers/DotProdLayer.cpp
--- a/paddle/gserver/layers/EosIdCheckLayer.cpp
+++ b/paddle/gserver/layers/EosIdCheckLayer.cpp
--- a/paddle/gserver/layers/ExpandConvLayer.cpp
+++ b/paddle/gserver/layers/ExpandConvLayer.cpp
--- a/paddle/legacy/gserver/layers/ExpandConvLayer.h
+++ b/paddle/legacy/gserver/layers/ExpandConvLayer.h
--- a/paddle/gserver/layers/ExpandLayer.cpp
+++ b/paddle/gserver/layers/ExpandLayer.cpp
--- a/paddle/legacy/gserver/layers/ExpandLayer.h
+++ b/paddle/legacy/gserver/layers/ExpandLayer.h
--- a/paddle/legacy/gserver/layers/FactorizationMachineLayer.cpp
+++ b/paddle/legacy/gserver/layers/FactorizationMachineLayer.cpp
--- a/paddle/legacy/gserver/layers/FactorizationMachineLayer.h
+++ b/paddle/legacy/gserver/layers/FactorizationMachineLayer.h
--- a/paddle/legacy/gserver/layers/FeatureMapExpandLayer.cpp
+++ b/paddle/legacy/gserver/layers/FeatureMapExpandLayer.cpp
--- a/paddle/gserver/layers/FullMatrixProjection.cpp
+++ b/paddle/gserver/layers/FullMatrixProjection.cpp
--- a/paddle/gserver/layers/FullMatrixProjection.h
+++ b/paddle/gserver/layers/FullMatrixProjection.h
--- a/paddle/legacy/gserver/layers/FullyConnectedLayer.cpp
+++ b/paddle/legacy/gserver/layers/FullyConnectedLayer.cpp
--- a/paddle/legacy/gserver/layers/FullyConnectedLayer.h
+++ b/paddle/legacy/gserver/layers/FullyConnectedLayer.h
--- a/paddle/gserver/layers/GatedRecurrentLayer.cpp
+++ b/paddle/gserver/layers/GatedRecurrentLayer.cpp
--- a/paddle/legacy/gserver/layers/GatedRecurrentLayer.h
+++ b/paddle/legacy/gserver/layers/GatedRecurrentLayer.h
--- a/paddle/gserver/layers/GetOutputLayer.cpp
+++ b/paddle/gserver/layers/GetOutputLayer.cpp
--- a/paddle/legacy/gserver/layers/GruCompute.cpp
+++ b/paddle/legacy/gserver/layers/GruCompute.cpp
--- a/paddle/gserver/layers/GruCompute.cu
+++ b/paddle/gserver/layers/GruCompute.cu
--- a/paddle/gserver/layers/GruCompute.h
+++ b/paddle/gserver/layers/GruCompute.h
--- a/paddle/gserver/layers/GruStepLayer.cpp
+++ b/paddle/gserver/layers/GruStepLayer.cpp
--- a/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp
+++ b/paddle/gserver/layers/HierarchicalSigmoidLayer.cpp
--- a/paddle/gserver/layers/HierarchicalSigmoidLayer.h
+++ b/paddle/gserver/layers/HierarchicalSigmoidLayer.h
--- a/paddle/gserver/layers/IdentityProjection.cpp
+++ b/paddle/gserver/layers/IdentityProjection.cpp
--- a/paddle/legacy/gserver/layers/InterpolationLayer.cpp
+++ b/paddle/legacy/gserver/layers/InterpolationLayer.cpp
--- a/paddle/gserver/layers/KmaxSeqScoreLayer.cpp
+++ b/paddle/gserver/layers/KmaxSeqScoreLayer.cpp
--- a/paddle/gserver/layers/L2DistanceLayer.cpp
+++ b/paddle/gserver/layers/L2DistanceLayer.cpp
--- a/paddle/legacy/gserver/layers/L2DistanceLayer.h
+++ b/paddle/legacy/gserver/layers/L2DistanceLayer.h
--- a/paddle/legacy/gserver/layers/Layer.cpp
+++ b/paddle/legacy/gserver/layers/Layer.cpp
--- a/paddle/legacy/gserver/layers/Layer.h
+++ b/paddle/legacy/gserver/layers/Layer.h
--- a/paddle/gserver/layers/LinearChainCRF.cpp
+++ b/paddle/gserver/layers/LinearChainCRF.cpp
--- a/paddle/legacy/gserver/layers/LinearChainCRF.h
+++ b/paddle/legacy/gserver/layers/LinearChainCRF.h
--- a/paddle/gserver/layers/LinearChainCTC.cpp
+++ b/paddle/gserver/layers/LinearChainCTC.cpp
--- a/paddle/legacy/gserver/layers/LinearChainCTC.h
+++ b/paddle/legacy/gserver/layers/LinearChainCTC.h
--- a/paddle/gserver/layers/LstmCompute.cpp
+++ b/paddle/gserver/layers/LstmCompute.cpp
--- a/paddle/gserver/layers/LstmCompute.cu
+++ b/paddle/gserver/layers/LstmCompute.cu
--- a/paddle/gserver/layers/LstmCompute.h
+++ b/paddle/gserver/layers/LstmCompute.h
--- a/paddle/legacy/gserver/layers/LstmLayer.cpp
+++ b/paddle/legacy/gserver/layers/LstmLayer.cpp
--- a/paddle/legacy/gserver/layers/LstmLayer.h
+++ b/paddle/legacy/gserver/layers/LstmLayer.h
--- a/paddle/gserver/layers/LstmStepLayer.cpp
+++ b/paddle/gserver/layers/LstmStepLayer.cpp
--- a/paddle/legacy/gserver/layers/MDLstmLayer.cpp
+++ b/paddle/legacy/gserver/layers/MDLstmLayer.cpp
--- a/paddle/gserver/layers/MKLDNNAddtoLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNAddtoLayer.cpp
--- a/paddle/gserver/layers/MKLDNNAddtoLayer.h
+++ b/paddle/gserver/layers/MKLDNNAddtoLayer.h
--- a/paddle/gserver/layers/MKLDNNBase.h
+++ b/paddle/gserver/layers/MKLDNNBase.h
--- a/paddle/gserver/layers/MKLDNNBatchNormLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNBatchNormLayer.cpp
--- a/paddle/gserver/layers/MKLDNNBatchNormLayer.h
+++ b/paddle/gserver/layers/MKLDNNBatchNormLayer.h
--- a/paddle/gserver/layers/MKLDNNConcatLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNConcatLayer.cpp
--- a/paddle/gserver/layers/MKLDNNConcatLayer.h
+++ b/paddle/gserver/layers/MKLDNNConcatLayer.h
--- a/paddle/legacy/gserver/layers/MKLDNNConvLayer.cpp
+++ b/paddle/legacy/gserver/layers/MKLDNNConvLayer.cpp
--- a/paddle/gserver/layers/MKLDNNConvLayer.h
+++ b/paddle/gserver/layers/MKLDNNConvLayer.h
--- a/paddle/gserver/layers/MKLDNNFcLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNFcLayer.cpp
--- a/paddle/gserver/layers/MKLDNNFcLayer.h
+++ b/paddle/gserver/layers/MKLDNNFcLayer.h
--- a/paddle/gserver/layers/MKLDNNLRNLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNLRNLayer.cpp
--- a/paddle/gserver/layers/MKLDNNLRNLayer.h
+++ b/paddle/gserver/layers/MKLDNNLRNLayer.h
--- a/paddle/gserver/layers/MKLDNNLayer.cpp
+++ b/paddle/gserver/layers/MKLDNNLayer.cpp
--- a/paddle/legacy/gserver/layers/MKLDNNLayer.h
+++ b/paddle/legacy/gserver/layers/MKLDNNLayer.h
--- a/paddle/legacy/gserver/layers/MKLDNNPoolLayer.cpp
+++ b/paddle/legacy/gserver/layers/MKLDNNPoolLayer.cpp
--- a/paddle/gserver/layers/MKLDNNPoolLayer.h
+++ b/paddle/gserver/layers/MKLDNNPoolLayer.h
--- a/paddle/gserver/layers/MKLPackedRecurrentLayer.cpp
+++ b/paddle/gserver/layers/MKLPackedRecurrentLayer.cpp
--- a/paddle/gserver/layers/MKLPackedRecurrentLayer.h
+++ b/paddle/gserver/layers/MKLPackedRecurrentLayer.h
--- a/paddle/legacy/gserver/layers/MKLPackedWeight.h
+++ b/paddle/legacy/gserver/layers/MKLPackedWeight.h
--- a/paddle/gserver/layers/MaxIdLayer.cpp
+++ b/paddle/gserver/layers/MaxIdLayer.cpp
--- a/paddle/gserver/layers/MaxLayer.cpp
+++ b/paddle/gserver/layers/MaxLayer.cpp
--- a/paddle/legacy/gserver/layers/MaxLayer.h
+++ b/paddle/legacy/gserver/layers/MaxLayer.h
--- a/paddle/gserver/layers/MaxOutLayer.cpp
+++ b/paddle/gserver/layers/MaxOutLayer.cpp
--- a/paddle/legacy/gserver/layers/MaxOutLayer.h
+++ b/paddle/legacy/gserver/layers/MaxOutLayer.h
--- a/paddle/gserver/layers/MaxPoolWithMaskLayer.cpp
+++ b/paddle/gserver/layers/MaxPoolWithMaskLayer.cpp
--- a/paddle/legacy/gserver/layers/MaxPoolWithMaskLayer.h
+++ b/paddle/legacy/gserver/layers/MaxPoolWithMaskLayer.h
--- a/paddle/gserver/layers/MixedLayer.cpp
+++ b/paddle/gserver/layers/MixedLayer.cpp
--- a/paddle/gserver/layers/MixedLayer.h
+++ b/paddle/gserver/layers/MixedLayer.h
--- a/paddle/gserver/layers/MultiBoxLossLayer.cpp
+++ b/paddle/gserver/layers/MultiBoxLossLayer.cpp
--- a/paddle/gserver/layers/MultiBoxLossLayer.h
+++ b/paddle/gserver/layers/MultiBoxLossLayer.h
--- a/paddle/gserver/layers/MultinomialSampler.cpp
+++ b/paddle/gserver/layers/MultinomialSampler.cpp
--- a/paddle/gserver/layers/MultinomialSampler.h
+++ b/paddle/gserver/layers/MultinomialSampler.h
--- a/paddle/legacy/gserver/layers/MultiplexLayer.cpp
+++ b/paddle/legacy/gserver/layers/MultiplexLayer.cpp
--- a/paddle/legacy/gserver/layers/NCELayer.cpp
+++ b/paddle/legacy/gserver/layers/NCELayer.cpp
--- a/paddle/gserver/layers/NormLayer.cpp
+++ b/paddle/gserver/layers/NormLayer.cpp
--- a/paddle/legacy/gserver/layers/NormLayer.h
+++ b/paddle/legacy/gserver/layers/NormLayer.h
--- a/paddle/gserver/layers/NormProjectionLayer.cpp
+++ b/paddle/gserver/layers/NormProjectionLayer.cpp
--- a/paddle/legacy/gserver/layers/NormProjectionLayer.h
+++ b/paddle/legacy/gserver/layers/NormProjectionLayer.h
--- a/paddle/gserver/layers/Operator.cpp
+++ b/paddle/gserver/layers/Operator.cpp
--- a/paddle/legacy/gserver/layers/Operator.h
+++ b/paddle/legacy/gserver/layers/Operator.h
--- a/paddle/legacy/gserver/layers/OuterProdLayer.cpp
+++ b/paddle/legacy/gserver/layers/OuterProdLayer.cpp
--- a/paddle/gserver/layers/PadLayer.cpp
+++ b/paddle/gserver/layers/PadLayer.cpp
--- a/paddle/gserver/layers/PadLayer.h
+++ b/paddle/gserver/layers/PadLayer.h
--- a/paddle/gserver/layers/ParameterReluLayer.cpp
+++ b/paddle/gserver/layers/ParameterReluLayer.cpp
--- a/paddle/legacy/gserver/layers/ParameterReluLayer.h
+++ b/paddle/legacy/gserver/layers/ParameterReluLayer.h
--- a/paddle/gserver/layers/Pool3DLayer.cpp
+++ b/paddle/gserver/layers/Pool3DLayer.cpp
--- a/paddle/legacy/gserver/layers/Pool3DLayer.h
+++ b/paddle/legacy/gserver/layers/Pool3DLayer.h
--- a/paddle/gserver/layers/PoolLayer.cpp
+++ b/paddle/gserver/layers/PoolLayer.cpp
--- a/paddle/legacy/gserver/layers/PoolLayer.h
+++ b/paddle/legacy/gserver/layers/PoolLayer.h
--- a/paddle/gserver/layers/PoolProjection.cpp
+++ b/paddle/gserver/layers/PoolProjection.cpp
--- a/paddle/legacy/gserver/layers/PoolProjection.h
+++ b/paddle/legacy/gserver/layers/PoolProjection.h
--- a/paddle/gserver/layers/PoolProjectionLayer.cpp
+++ b/paddle/gserver/layers/PoolProjectionLayer.cpp
--- a/paddle/legacy/gserver/layers/PoolProjectionLayer.h
+++ b/paddle/legacy/gserver/layers/PoolProjectionLayer.h
--- a/paddle/legacy/gserver/layers/PowerLayer.cpp
+++ b/paddle/legacy/gserver/layers/PowerLayer.cpp
--- a/paddle/gserver/layers/PrintLayer.cpp
+++ b/paddle/gserver/layers/PrintLayer.cpp
--- a/paddle/legacy/gserver/layers/PriorBox.cpp
+++ b/paddle/legacy/gserver/layers/PriorBox.cpp
--- a/paddle/gserver/layers/Projection.cpp
+++ b/paddle/gserver/layers/Projection.cpp
--- a/paddle/legacy/gserver/layers/Projection.h
+++ b/paddle/legacy/gserver/layers/Projection.h
--- a/paddle/gserver/layers/ROIPoolLayer.cpp
+++ b/paddle/gserver/layers/ROIPoolLayer.cpp
--- a/paddle/gserver/layers/ROIPoolLayer.h
+++ b/paddle/gserver/layers/ROIPoolLayer.h
--- a/paddle/gserver/layers/RecurrentLayer.cpp
+++ b/paddle/gserver/layers/RecurrentLayer.cpp
--- a/paddle/gserver/layers/RecurrentLayer.h
+++ b/paddle/gserver/layers/RecurrentLayer.h
--- a/paddle/legacy/gserver/layers/RecurrentLayerGroup.cpp
+++ b/paddle/legacy/gserver/layers/RecurrentLayerGroup.cpp
--- a/paddle/legacy/gserver/layers/ResizeLayer.cpp
+++ b/paddle/legacy/gserver/layers/ResizeLayer.cpp
--- a/paddle/gserver/layers/RotateLayer.cpp
+++ b/paddle/gserver/layers/RotateLayer.cpp
--- a/paddle/legacy/gserver/layers/RotateLayer.h
+++ b/paddle/legacy/gserver/layers/RotateLayer.h
--- a/paddle/gserver/layers/RowConvLayer.cpp
+++ b/paddle/gserver/layers/RowConvLayer.cpp
--- a/paddle/gserver/layers/RowConvLayer.h
+++ b/paddle/gserver/layers/RowConvLayer.h
--- a/paddle/gserver/layers/RowL2NormLayer.cpp
+++ b/paddle/gserver/layers/RowL2NormLayer.cpp
--- a/paddle/gserver/layers/SamplingIdLayer.cpp
+++ b/paddle/gserver/layers/SamplingIdLayer.cpp
--- a/paddle/gserver/layers/ScaleShiftLayer.cpp
+++ b/paddle/gserver/layers/ScaleShiftLayer.cpp
--- a/paddle/gserver/layers/ScaleSubRegionLayer.cpp
+++ b/paddle/gserver/layers/ScaleSubRegionLayer.cpp
--- a/paddle/gserver/layers/ScaleSubRegionLayer.h
+++ b/paddle/gserver/layers/ScaleSubRegionLayer.h
--- a/paddle/legacy/gserver/layers/ScalingLayer.cpp
+++ b/paddle/legacy/gserver/layers/ScalingLayer.cpp
--- a/paddle/gserver/layers/ScalingProjection.cpp
+++ b/paddle/gserver/layers/ScalingProjection.cpp
--- a/paddle/legacy/gserver/layers/SelectiveFullyConnectedLayer.cpp
+++ b/paddle/legacy/gserver/layers/SelectiveFullyConnectedLayer.cpp
--- a/paddle/legacy/gserver/layers/SelectiveFullyConnectedLayer.h
+++ b/paddle/legacy/gserver/layers/SelectiveFullyConnectedLayer.h
--- a/paddle/legacy/gserver/layers/SequenceConcatLayer.cpp
+++ b/paddle/legacy/gserver/layers/SequenceConcatLayer.cpp
--- a/paddle/legacy/gserver/layers/SequenceLastInstanceLayer.cpp
+++ b/paddle/legacy/gserver/layers/SequenceLastInstanceLayer.cpp
--- a/paddle/gserver/layers/SequencePoolLayer.cpp
+++ b/paddle/gserver/layers/SequencePoolLayer.cpp
--- a/paddle/legacy/gserver/layers/SequencePoolLayer.h
+++ b/paddle/legacy/gserver/layers/SequencePoolLayer.h
--- a/paddle/legacy/gserver/layers/SequenceReshapeLayer.cpp
+++ b/paddle/legacy/gserver/layers/SequenceReshapeLayer.cpp
--- a/paddle/legacy/gserver/layers/SequenceSliceLayer.cpp
+++ b/paddle/legacy/gserver/layers/SequenceSliceLayer.cpp
--- a/paddle/gserver/layers/SequenceToBatch.cpp
+++ b/paddle/gserver/layers/SequenceToBatch.cpp
--- a/paddle/legacy/gserver/layers/SequenceToBatch.h
+++ b/paddle/legacy/gserver/layers/SequenceToBatch.h
--- a/paddle/gserver/layers/SliceProjection.cpp
+++ b/paddle/gserver/layers/SliceProjection.cpp
--- a/paddle/legacy/gserver/layers/SlopeInterceptLayer.cpp
+++ b/paddle/legacy/gserver/layers/SlopeInterceptLayer.cpp
--- a/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp
+++ b/paddle/gserver/layers/SpatialPyramidPoolLayer.cpp
--- a/paddle/legacy/gserver/layers/SpatialPyramidPoolLayer.h
+++ b/paddle/legacy/gserver/layers/SpatialPyramidPoolLayer.h
--- a/paddle/legacy/gserver/layers/SubNestedSequenceLayer.cpp
+++ b/paddle/legacy/gserver/layers/SubNestedSequenceLayer.cpp
--- a/paddle/legacy/gserver/layers/SubSequenceLayer.cpp
+++ b/paddle/legacy/gserver/layers/SubSequenceLayer.cpp
--- a/paddle/legacy/gserver/layers/SumToOneNormLayer.cpp
+++ b/paddle/legacy/gserver/layers/SumToOneNormLayer.cpp
--- a/paddle/gserver/layers/SwitchOrderLayer.cpp
+++ b/paddle/gserver/layers/SwitchOrderLayer.cpp
--- a/paddle/gserver/layers/SwitchOrderLayer.h
+++ b/paddle/gserver/layers/SwitchOrderLayer.h
--- a/paddle/gserver/layers/TableProjection.cpp
+++ b/paddle/gserver/layers/TableProjection.cpp
--- a/paddle/gserver/layers/TableProjection.h
+++ b/paddle/gserver/layers/TableProjection.h
--- a/paddle/gserver/layers/TensorLayer.cpp
+++ b/paddle/gserver/layers/TensorLayer.cpp
--- a/paddle/legacy/gserver/layers/TensorLayer.h
+++ b/paddle/legacy/gserver/layers/TensorLayer.h
--- a/paddle/gserver/layers/TransLayer.cpp
+++ b/paddle/gserver/layers/TransLayer.cpp
--- a/paddle/legacy/gserver/layers/TransLayer.h
+++ b/paddle/legacy/gserver/layers/TransLayer.h
--- a/paddle/gserver/layers/TransposedFullMatrixProjection.cpp
+++ b/paddle/gserver/layers/TransposedFullMatrixProjection.cpp
--- a/paddle/gserver/layers/UpsampleLayer.cpp
+++ b/paddle/gserver/layers/UpsampleLayer.cpp
--- a/paddle/legacy/gserver/layers/UpsampleLayer.h
+++ b/paddle/legacy/gserver/layers/UpsampleLayer.h
--- a/paddle/gserver/layers/ValidationLayer.cpp
+++ b/paddle/gserver/layers/ValidationLayer.cpp
--- a/paddle/legacy/gserver/layers/ValidationLayer.h
+++ b/paddle/legacy/gserver/layers/ValidationLayer.h
--- a/paddle/gserver/layers/WarpCTCLayer.cpp
+++ b/paddle/gserver/layers/WarpCTCLayer.cpp
--- a/paddle/gserver/layers/WarpCTCLayer.h
+++ b/paddle/gserver/layers/WarpCTCLayer.h
--- a/paddle/gserver/tests/.gitignore
+++ b/paddle/gserver/tests/.gitignore
--- a/paddle/legacy/gserver/tests/CMakeLists.txt
+++ b/paddle/legacy/gserver/tests/CMakeLists.txt
--- a/paddle/gserver/tests/LayerGradUtil.cpp
+++ b/paddle/gserver/tests/LayerGradUtil.cpp
--- a/paddle/legacy/gserver/tests/LayerGradUtil.h
+++ b/paddle/legacy/gserver/tests/LayerGradUtil.h
--- a/paddle/legacy/gserver/tests/MKLDNNTester.cpp
+++ b/paddle/legacy/gserver/tests/MKLDNNTester.cpp
--- a/paddle/legacy/gserver/tests/MKLDNNTester.h
+++ b/paddle/legacy/gserver/tests/MKLDNNTester.h
--- a/paddle/gserver/tests/Sequence/dummy.list
+++ b/paddle/gserver/tests/Sequence/dummy.list
--- a/paddle/gserver/tests/Sequence/tour_dict_phrase.dict
+++ b/paddle/gserver/tests/Sequence/tour_dict_phrase.dict
--- a/paddle/gserver/tests/Sequence/tour_train_wdseg
+++ b/paddle/gserver/tests/Sequence/tour_train_wdseg
--- a/paddle/gserver/tests/Sequence/tour_train_wdseg.nest
+++ b/paddle/gserver/tests/Sequence/tour_train_wdseg.nest
--- a/paddle/legacy/gserver/tests/Sequence/train.list
+++ b/paddle/legacy/gserver/tests/Sequence/train.list
--- a/paddle/legacy/gserver/tests/Sequence/train.list.nest
+++ b/paddle/legacy/gserver/tests/Sequence/train.list.nest
--- a/paddle/gserver/tests/__init__.py
+++ b/paddle/gserver/tests/__init__.py
--- a/paddle/gserver/tests/concat_dotmul_a.conf
+++ b/paddle/gserver/tests/concat_dotmul_a.conf
--- a/paddle/gserver/tests/concat_dotmul_b.conf
+++ b/paddle/gserver/tests/concat_dotmul_b.conf
--- a/paddle/gserver/tests/concat_fullmatrix_a.conf
+++ b/paddle/gserver/tests/concat_fullmatrix_a.conf
--- a/paddle/gserver/tests/concat_fullmatrix_b.conf
+++ b/paddle/gserver/tests/concat_fullmatrix_b.conf
--- a/paddle/gserver/tests/concat_slice_a.conf
+++ b/paddle/gserver/tests/concat_slice_a.conf
--- a/paddle/gserver/tests/concat_slice_b.conf
+++ b/paddle/gserver/tests/concat_slice_b.conf
--- a/paddle/gserver/tests/concat_table_a.conf
+++ b/paddle/gserver/tests/concat_table_a.conf
--- a/paddle/gserver/tests/concat_table_b.conf
+++ b/paddle/gserver/tests/concat_table_b.conf
--- a/paddle/gserver/tests/img_conv_a.conf
+++ b/paddle/gserver/tests/img_conv_a.conf
--- a/paddle/gserver/tests/img_conv_b.conf
+++ b/paddle/gserver/tests/img_conv_b.conf
--- a/paddle/gserver/tests/img_conv_c.conf
+++ b/paddle/gserver/tests/img_conv_c.conf
--- a/paddle/gserver/tests/img_conv_cudnn.py
+++ b/paddle/gserver/tests/img_conv_cudnn.py
--- a/paddle/gserver/tests/img_conv_exconv.py
+++ b/paddle/gserver/tests/img_conv_exconv.py
--- a/paddle/gserver/tests/img_pool_a.conf
+++ b/paddle/gserver/tests/img_pool_a.conf
--- a/paddle/gserver/tests/img_pool_b.conf
+++ b/paddle/gserver/tests/img_pool_b.conf
--- a/paddle/gserver/tests/mkldnn_branch_net.conf
+++ b/paddle/gserver/tests/mkldnn_branch_net.conf
--- a/paddle/gserver/tests/mkldnn_simple_net.conf
+++ b/paddle/gserver/tests/mkldnn_simple_net.conf
--- a/paddle/gserver/tests/pyDataProvider.py
+++ b/paddle/gserver/tests/pyDataProvider.py
--- a/paddle/gserver/tests/pyDataProvider/pyDataProviderList
+++ b/paddle/gserver/tests/pyDataProvider/pyDataProviderList
--- a/paddle/gserver/tests/pyDataProvider/trainer.conf
+++ b/paddle/gserver/tests/pyDataProvider/trainer.conf
--- a/paddle/gserver/tests/rnn_data_provider.py
+++ b/paddle/gserver/tests/rnn_data_provider.py
--- a/paddle/gserver/tests/sequenceGen.py
+++ b/paddle/gserver/tests/sequenceGen.py
--- a/paddle/legacy/gserver/tests/sequence_layer_group.conf
+++ b/paddle/legacy/gserver/tests/sequence_layer_group.conf
--- a/paddle/legacy/gserver/tests/sequence_lstm.conf
+++ b/paddle/legacy/gserver/tests/sequence_lstm.conf
--- a/paddle/legacy/gserver/tests/sequence_nest_layer_group.conf
+++ b/paddle/legacy/gserver/tests/sequence_nest_layer_group.conf
--- a/paddle/legacy/gserver/tests/sequence_nest_rnn.conf
+++ b/paddle/legacy/gserver/tests/sequence_nest_rnn.conf
--- a/paddle/legacy/gserver/tests/sequence_nest_rnn_multi_input.conf
+++ b/paddle/legacy/gserver/tests/sequence_nest_rnn_multi_input.conf
--- a/paddle/legacy/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
+++ b/paddle/legacy/gserver/tests/sequence_nest_rnn_multi_unequalength_inputs.py
--- a/paddle/legacy/gserver/tests/sequence_recurrent.py
+++ b/paddle/legacy/gserver/tests/sequence_recurrent.py
--- a/paddle/legacy/gserver/tests/sequence_recurrent_group.py
+++ b/paddle/legacy/gserver/tests/sequence_recurrent_group.py