squash commit

cmake/cuda.cmake

@@ -169,9 +169,9 @@ set(CUDA_PROPAGATE_HOST_FLAGS OFF)
 
 # Release/Debug flags set by cmake. Such as -O3 -g -DNDEBUG etc.
 # So, don't set these flags here.
-#list(APPEND CUDA_NVCC_FLAGS "-std=c++14")
+
 if (NOT WIN32) # windows msvc2015 support c++11 natively. 
-# -std=c++11 -fPIC not recoginize by msvc, -Xcompiler will be added by cmake.
+# -std=c++11 -fPIC not recoginize by msvc
 list(APPEND CUDA_NVCC_FLAGS "-Xcompiler -fPIC")
 endif(NOT WIN32)
 
@@ -201,4 +201,4 @@ endif()
 endif(NOT WIN32)
 
 mark_as_advanced(CUDA_BUILD_CUBIN CUDA_BUILD_EMULATION CUDA_VERBOSE_BUILD)
-mark_as_advanced(CUDA_SDK_ROOT_DIR CUDA_SEPARABLE_COMPILATION)
+mark_as_advanced(CUDA_SDK_ROOT_DIR CUDA_SEPARABLE_COMPILATION)
\ No newline at end of file

cmake/flags.cmake

@@ -26,7 +26,6 @@ function(CheckCompilerCXX11Flag)
 endfunction()
 
 CheckCompilerCXX11Flag()
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
 
 # safe_set_flag
 #
@@ -136,7 +135,7 @@ else(NOT WIN32)
 set(COMMON_FLAGS
     "/w") #disable all warnings
 set(GPU_COMMON_FLAGS
-    -w) #disable all warnings
+    "") #disable all warnings
 
 endif(NOT WIN32)
 
@@ -160,7 +159,6 @@ if(UNIX AND NOT APPLE)
   set(LINUX TRUE)
 endif(UNIX AND NOT APPLE)
 
-set(GPU_COMMON_FLAGS /std:c++14 ${GPU_COMMON_FLAGS})
 foreach(flag ${COMMON_FLAGS})
     safe_set_cflag(CMAKE_C_FLAGS ${flag})
     safe_set_cxxflag(CMAKE_CXX_FLAGS ${flag})

paddle/fluid/framework/eigen.h

@@ -13,11 +13,10 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #pragma once
-// for windows compile eigen with logging
+// logging.h and windows.h conflict
 #define GLOG_NO_ABBREVIATED_SEVERITIES
 
 #include "paddle/fluid/framework/tensor.h"
-#include <math_constants.h>
 #include "unsupported/Eigen/CXX11/Tensor"
 
 namespace paddle {
@@ -49,11 +48,13 @@ struct EigenTensor {
   using ConstType =
       Eigen::TensorMap<Eigen::Tensor<const T, D, MajorType, IndexType>>;
 
-  static Type From(Tensor& tensor, DDim dims) {
+  static Type From(Tensor& tensor, DDim dims) {  // NOLINT
     return Type(tensor.data<T>(), EigenDim<D>::From(dims));
   }
 
-  static Type From(Tensor& tensor) { return From(tensor, tensor.dims_); }
+  static Type From(Tensor& tensor) {  // NOLINT
+    return From(tensor, tensor.dims_);
+  }  // NOLINT
 
   static ConstType From(const Tensor& tensor, DDim dims) {
     return ConstType(tensor.data<T>(), EigenDim<D>::From(dims));
@@ -67,7 +68,8 @@ struct EigenTensor {
 template <typename T, int MajorType = Eigen::RowMajor,
           typename IndexType = Eigen::DenseIndex>
 struct EigenMatrix : public EigenTensor<T, 2, MajorType, IndexType> {
-  static typename EigenMatrix::Type Reshape(Tensor& tensor, int num_col_dims) {
+  static typename EigenMatrix::Type Reshape(Tensor& tensor,  // NOLINT
+                                            int num_col_dims) {
     int rank = tensor.dims_.size();
     PADDLE_ENFORCE(num_col_dims > 0 && num_col_dims < rank,
                    "`num_col_dims` must be between (0, rank_of_tensor).");
@@ -89,11 +91,12 @@ template <typename T, int MajorType = Eigen::RowMajor,
           typename IndexType = Eigen::DenseIndex>
 struct EigenVector : public EigenTensor<T, 1, MajorType, IndexType> {
   // Flatten reshapes a Tensor into an EigenVector.
-  static typename EigenVector::Type Flatten(Tensor& tensor) {
+  static typename EigenVector::Type Flatten(Tensor& tensor) {  // NOLINT
     return EigenVector::From(tensor, {product(tensor.dims_)});
   }
 
-  static typename EigenVector::ConstType Flatten(const Tensor& tensor) {
+  static typename EigenVector::ConstType Flatten(
+      const Tensor& tensor) {  // NOLINT
     return EigenVector::From(tensor, {product(tensor.dims_)});
   }
 };
@@ -107,7 +110,7 @@ struct EigenScalar {
   using ConstType = Eigen::TensorMap<
       Eigen::TensorFixedSize<const T, Eigen::Sizes<>, MajorType, IndexType>>;
 
-  static Type From(Tensor& tensor) { return Type(tensor.data<T>()); }
+  static Type From(Tensor& tensor) { return Type(tensor.data<T>()); }  // NOLINT
 
   static ConstType From(const Tensor& tensor) {
     return ConstType(tensor.data<T>());

paddle/fluid/framework/ir/attention_lstm_fuse_pass.cc

@@ -11,6 +11,7 @@
 // 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 <string>
 
 #include "paddle/fluid/framework/ir/attention_lstm_fuse_pass.h"
 #include "paddle/fluid/framework/ir/graph_pattern_detector.h"
@@ -215,12 +216,12 @@ void PrepareLSTMWeight(const LoDTensor& W_forget_w0,
   VLOG(3) << "LSTMWeight resized to " << out->dims();
 
   float* out_data = out->mutable_data<float>(platform::CPUPlace());
-  std::array<const float*, 4> tensors(
-      {{W_forget_w0.data<float>(), W_input_w0.data<float>(),
-        W_output_w0.data<float>(), W_cell_w0.data<float>()}});
-  std::array<const float*, 4> tensors1(
-      {{W_forget_w1.data<float>(), W_input_w1.data<float>(),
-        W_output_w1.data<float>(), W_cell_w1.data<float>()}});
+  std::array<const float*, 4> tensors = {
+      W_forget_w0.data<float>(), W_input_w0.data<float>(),
+      W_output_w0.data<float>(), W_cell_w0.data<float>()};
+  std::array<const float*, 4> tensors1 = {
+      W_forget_w1.data<float>(), W_input_w1.data<float>(),
+      W_output_w1.data<float>(), W_cell_w1.data<float>()};
 
   for (int row = 0; row < D; row++) {
     for (int col = 0; col < 4; col++) {
@@ -242,9 +243,9 @@ void PrepareLSTMWeight(const LoDTensor& W_forget_w0,
 void PrepareLSTMBias(const LoDTensor& B_forget, const LoDTensor& B_input,
                      const LoDTensor& B_output, const LoDTensor& B_cell,
                      LoDTensor* out) {
-  std::array<const float*, 4> tensors(
-      {{B_forget.data<float>(), B_input.data<float>(), B_output.data<float>(),
-        B_cell.data<float>()}});
+  std::array<const float*, 4> tensors = {
+      B_forget.data<float>(), B_input.data<float>(), B_output.data<float>(),
+      B_cell.data<float>()};
 
   PADDLE_ENFORCE_EQ(B_forget.dims().size(), 1);
   int D = B_forget.dims()[0];

paddle/fluid/inference/analysis/CMakeLists.txt

@@ -21,6 +21,7 @@ cc_binary(inference_analyzer SRCS analyzer_main.cc DEPS analysis paddle_fluid)
 
 set(PYTHON_TESTS_DIR ${PADDLE_BINARY_DIR}/python/paddle/fluid/tests)
 
+if (NOT WIN32)
 function (inference_analysis_test TARGET)
     if(WITH_TESTING)
         set(options "")
@@ -98,3 +99,4 @@ inference_analysis_test(test_chinese_ner SRCS chinese_ner_tester.cc
     ARGS --inference_model_dir=${PYTHON_TESTS_DIR}/book/word2vec.inference.model
         --infer_model=${CHINESE_NER_INSTALL_DIR}/model
         --infer_data=${CHINESE_NER_INSTALL_DIR}/data.txt)
+endif(NOT WIN32)

paddle/fluid/operators/CMakeLists.txt

@@ -87,7 +87,7 @@ function(op_library TARGET)
     if (WIN32)
     # no nccl, no avx instructions ops.
     foreach(windows_unsupport_op "nccl_op" "gen_nccl_id_op" "warpctc_op" "hierarchical_sigmoid_op"
-     "crf_decoding_op" "select_op" "lstmp_op" "gru_op" "lstm_op" "fusion_lstm_op")
+     "crf_decoding_op" "select_op" "lstmp_op" "gru_op" "lstm_op" "fusion_lstm_op" "cumsum_op")
         if ("${TARGET}" STREQUAL "${windows_unsupport_op}")
           return()
         endif()

paddle/fluid/operators/cum_op.h

@@ -85,17 +85,14 @@ class CumKernel : public framework::OpKernel<typename Functor::ELEMENT_TYPE> {
   template <typename Device, typename Dim, typename X, typename Out>
   void ComputeImp(Device d, const Dim& dims, X x, Out out, int axis,
                   bool reverse, bool exclusive) const {
-    Functor func();
     if (!reverse) {
-      out.reshape(dims).device(d) =
-          func.apply(x.reshape(dims), axis, exclusive);
+      out.reshape(dims).device(d) = Functor()(x.reshape(dims), axis, exclusive);
     } else {
       std::array<bool, Dim::count> rev;
       rev.fill(false);
       rev[axis] = reverse;
       out.reshape(dims).device(d) =
-          func.apply(x.reshape(dims).reverse(rev), axis, exclusive)
-              .reverse(rev);
+          Functor()(x.reshape(dims).reverse(rev), axis, exclusive).reverse(rev);
     }
   }
 };
@@ -104,7 +101,8 @@ template <typename T>
 struct CumsumFunctor {
   using ELEMENT_TYPE = T;
   template <typename X>
-  const typename X::TensorScanSumOp apply(X x, int axis, bool exclusive) const {
+  const typename X::TensorScanSumOp operator()(X x, int axis,
+                                               bool exclusive) const {
     return x.cumsum(axis, exclusive);
   }
 };

paddle/fluid/operators/math/math_function.cc

@@ -13,6 +13,15 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #include "paddle/fluid/operators/math/math_function.h"
+
+#ifdef PADDLE_WITH_MKLML
+#include "paddle/fluid/platform/dynload/mklml.h"
+#endif
+
+#ifdef PADDLE_USE_OPENBLAS
+#include <cblas.h>
+#endif
+
 #include <vector>
 #include "paddle/fluid/framework/data_type.h"
 #include "paddle/fluid/operators/math/math_function_impl.h"

paddle/fluid/operators/math/math_function.h

@@ -13,18 +13,6 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #pragma once
-#ifdef PADDLE_WITH_MKLML
-#include "paddle/fluid/platform/dynload/mklml.h"
-#endif
-
-#ifdef PADDLE_USE_OPENBLAS
-#include <cblas.h>
-// remove typedef in openblas
-#undef FLOAT
-#undef INT
-#undef SIZE
-#endif
-
 #include <cmath>
 #include <vector>
 

paddle/fluid/operators/math/selected_rows_functor.cu

@@ -16,6 +16,7 @@ limitations under the License. */
 #include <vector>
 
 #include "paddle/fluid/operators/math/math_function.h"
+#include "paddle/fluid/operators/math/math_function_impl.h"
 #include "paddle/fluid/operators/math/selected_rows_functor.h"
 #include "paddle/fluid/platform/cuda_primitives.h"
 

paddle/fluid/operators/math/sequence_pooling.cu

@@ -16,13 +16,12 @@ limitations under the License. */
 #include "paddle/fluid/operators/math/math_function.h"
 #include "paddle/fluid/operators/math/sequence_pooling.h"
 #include "paddle/fluid/platform/cuda_primitives.h"
+#include "paddle/fluid/platform/macros.h"
 
 namespace paddle {
 namespace operators {
 namespace math {
 
-#define FLT_MAX __FLT_MAX__
-
 template <typename T>
 struct MaxPoolFunctor {
   HOSTDEVICE void operator()(const T* input, const size_t start,