Merge remote-tracking branch 'ups/develop' into fix

Author: tensor-tang

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 |

cmake/inference_lib.cmake

@@ -39,7 +39,7 @@ 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)
@@ -155,6 +155,15 @@ copy(inference_lib DEPS paddle_fluid_shared paddle_fluid
   DSTS ${dst_dir}/${module} ${dst_dir}/${module}
 )
 
+if(WITH_CONTRIB)
+   set(contrib_dst_dir "${FLUID_INSTALL_DIR}/contrib/inference")
+   copy(contrib_inference_lib DEPS paddle_inference_api
+        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}
+   )
+endif()
+
 set(module "platform")
 copy(platform_lib DEPS profiler_py_proto
   SRCS ${src_dir}/${module}/*.h ${src_dir}/${module}/dynload/*.h ${src_dir}/${module}/details/*.h

doc/fluid/api/layers.rst

@@ -342,6 +342,12 @@ conv2d
 ..  autofunction:: paddle.fluid.layers.conv2d
     :noindex:
 
+conv3d
+------
+
+..  autofunction:: paddle.fluid.layers.conv3d
+    :noindex:
+
 sequence_pool
 -------------
 
@@ -366,6 +372,12 @@ pool2d
 ..  autofunction:: paddle.fluid.layers.pool2d
     :noindex:
 
+pool3d
+------
+
+..  autofunction:: paddle.fluid.layers.pool3d
+    :noindex:
+
 batch_norm
 ----------
 
@@ -384,6 +396,13 @@ conv2d_transpose
 ..  autofunction:: paddle.fluid.layers.conv2d_transpose
     :noindex:
 
+conv3d_transpose
+----------------
+
+..  autofunction:: paddle.fluid.layers.conv2d_transpose
+    :noindex:
+
+
 sequence_expand
 ---------------
 

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）
 

paddle/contrib/inference/CMakeLists.txt

@@ -50,7 +50,7 @@ cc_test(test_paddle_inference_api
 inference_api_test(test_paddle_inference_api_impl
                     ARGS test_word2vec test_image_classification)
 
-if (WITH_ANAKIN)
+if (WITH_ANAKIN AND WITH_TESTING) # only needed in CI
     # Due to Anakin do not have official library releases and the versions of protobuf and cuda do not match Paddle's,
     # so anakin library will not be merged to our official inference library. To use anakin prediction API, one need to
     # compile the libinference_anakin_api.a and compile with anakin.so.

paddle/contrib/tape/CMakeLists.txt

@@ -17,7 +17,7 @@ if(APPLE)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-error=pessimizing-move")
 endif(APPLE)
 
-cc_library(tape_variable SRCS variable.cc DEPS ${FLUID_CORE_MODULES})
+cc_library(tape_variable SRCS variable.cc DEPS ${FLUID_CORE_MODULES} device_context framework_proto proto_desc operator)
 cc_library(tape SRCS tape.cc DEPS ${FLUID_CORE_MODULES} ${GLOB_OP_LIB} tape_variable)
 
 cc_test(test_tape

paddle/fluid/framework/init.cc

@@ -18,6 +18,7 @@ limitations under the License. */
 
 #include "paddle/fluid/framework/init.h"
 #include "paddle/fluid/framework/operator.h"
+#include "paddle/fluid/operators/math/blas.h"
 #include "paddle/fluid/platform/device_context.h"
 #include "paddle/fluid/platform/place.h"
 #include "paddle/fluid/string/piece.h"
@@ -113,6 +114,9 @@ void InitDevices(bool init_p2p, const std::vector<int> devices) {
   }
   places.emplace_back(platform::CPUPlace());
   platform::DeviceContextPool::Init(places);
+#ifndef PADDLE_WITH_MKLDNN
+  operators::math::SetNumThreads(1);
+#endif
 }
 
 void InitGLOG(const std::string &prog_name) {

paddle/fluid/inference/tensorrt/convert/op_converter.h

@@ -64,7 +64,8 @@ class OpConverter {
     (*it)(op, scope, test_mode);
   }
 
-  // convert fluid block to tensorrt network
+  // Convert a fluid block to tensorrt network, NOTE it just convert operators,
+  // the INetwork's inputs and outputs should specified in some other modules.
   void ConvertBlock(const framework::proto::BlockDesc& block,
                     const std::unordered_set<std::string>& parameters,
                     const framework::Scope& scope, TensorRTEngine* engine) {

paddle/fluid/inference/tensorrt/engine.h

@@ -51,11 +51,12 @@ class TensorRTEngine : public EngineBase {
     nvinfer1::Weights w_;
   };
 
-  TensorRTEngine(int max_batch, int max_workspace, cudaStream_t* stream,
+  TensorRTEngine(int max_batch, int max_workspace,
+                 cudaStream_t* stream = nullptr,
                  nvinfer1::ILogger& logger = NaiveLogger::Global())
       : max_batch_(max_batch),
         max_workspace_(max_workspace),
-        stream_(stream),
+        stream_(stream ? stream : &default_stream_),
         logger_(logger) {}
 
   virtual ~TensorRTEngine();
@@ -121,6 +122,8 @@ class TensorRTEngine : public EngineBase {
   // the max memory size the engine uses
   int max_workspace_;
   cudaStream_t* stream_;
+  // If stream_ is not set from outside, hold its own stream.
+  cudaStream_t default_stream_;
   nvinfer1::ILogger& logger_;
 
   std::vector<Buffer> buffers_;
@@ -165,20 +168,31 @@ class TensorRTEngine : public EngineBase {
  */
 class TRT_EngineManager {
  public:
-  TensorRTEngine* Create(int max_batch, int max_workspace,
-                         cudaStream_t* stream) {
-    engines_.emplace_back(new TensorRTEngine(max_batch, max_workspace, stream));
-    return engines_.back().get();
+  bool HasEngine(const std::string& name) const {
+    return engines_.count(name) != 0;
+  }
+
+  // Get an engine called `name`.
+  TensorRTEngine* Get(const std::string& name) const {
+    return engines_.at(name).get();
+  }
+
+  // Create or get an engine called `name`
+  TensorRTEngine* Create(int max_batch, int max_workspace, cudaStream_t* stream,
+                         const std::string& name) {
+    auto* p = new TensorRTEngine(max_batch, max_workspace, stream);
+    engines_[name].reset(p);
+    return p;
   }
 
   void DeleteALl() {
-    for (auto& ptr : engines_) {
-      ptr.reset(nullptr);
+    for (auto& item : engines_) {
+      item.second.reset(nullptr);
     }
   }
 
  private:
-  std::vector<std::unique_ptr<TensorRTEngine>> engines_;
+  std::unordered_map<std::string, std::unique_ptr<TensorRTEngine>> engines_;
 };
 
 }  // namespace tensorrt

paddle/fluid/operators/activation_op.cc

@@ -252,15 +252,14 @@ class SoftShrinkOpMaker : public framework::OpProtoAndCheckerMaker {
     AddOutput("Out", "Output of Softshrink operator");
     AddAttr<float>("lambda", "non-negative offset").SetDefault(0.5f);
     AddComment(R"DOC(
-Softshrink Activation Operator.
+:strong:`Softshrink Activation Operator`
 
-$$
-out = \begin{cases} 
-    x - \lambda, \text{if } x > \lambda \\
-    x + \lambda, \text{if } x < -\lambda \\
-    0,  \text{otherwise}
-    \end{cases}
-$$
+..  math::
+    out = \begin{cases} 
+         x - \lambda, \text{if } x > \lambda \\
+         x + \lambda, \text{if } x < -\lambda \\
+         0,  \text{otherwise}
+         \end{cases}
 
 )DOC");
   }
@@ -271,18 +270,18 @@ class HardShrinkOpMaker : public framework::OpProtoAndCheckerMaker {
   void Make() override {
     AddInput("X", "Input of HardShrink operator");
     AddOutput("Out", "Output of HardShrink operator");
-    AddAttr<float>("threshold", "The value of threshold for HardShrink")
+    AddAttr<float>("threshold",
+                   "The value of threshold for HardShrink. [default: 0.5]")
         .SetDefault(0.5f);
     AddComment(R"DOC(
-HardShrink Activation Operator.
+:strong:`HardShrink activation operator`
 
-$$
-out = \begin{cases} 
-    x, \text{if } x > \lambda \\
-    x, \text{if } x < -\lambda \\
-    0,  \text{otherwise}
-    \end{cases}
-$$
+..  math::
+    out = \begin{cases}
+            x, \text{if } x > \lambda \\
+            x, \text{if } x < -\lambda \\
+            0,  \text{otherwise}
+          \end{cases}
 
 )DOC");
   }
@@ -394,18 +393,18 @@ class ThresholdedReluOpMaker : public framework::OpProtoAndCheckerMaker {
   void Make() override {
     AddInput("X", "Input of ThresholdedRelu operator");
     AddOutput("Out", "Output of ThresholdedRelu operator");
-    AddAttr<float>("threshold", "The threshold location of activation")
+    AddAttr<float>("threshold",
+                   "The threshold location of activation. [default 1.0].")
         .SetDefault(1.0f);
     AddComment(R"DOC(
-ThresholdedRelu Activation Operator.
+:strong:`ThresholdedRelu activation operator`
 
-$$
-out = \begin{cases} 
-    x, \text{if } x > threshold \\
-    0,  \text{otherwise}
-    \end{cases}
-$$
+..  math::
 
+    out = \begin{cases}
+             x,  \text{if } x > threshold \\
+             0,  \text{otherwise}
+          \end{cases}
 )DOC");
   }
 };