Merge pull request #1 from PaddlePaddle/develop

Sync with origin

Author: wangguibao

cmake/configure.cmake

@@ -62,8 +62,26 @@ if(NOT CMAKE_CROSSCOMPILING)
 endif()
 
 if(WIN32)
-  # windows stupid compile option for all targets.
+  # windows header option for all targets.
   add_definitions(-D_XKEYCHECK_H)
+  # Use symbols instead of absolute path, reduce the cmake link command length. 
+  SET(CMAKE_C_USE_RESPONSE_FILE_FOR_LIBRARIES 1)
+  SET(CMAKE_CXX_USE_RESPONSE_FILE_FOR_LIBRARIES 1)
+  SET(CMAKE_C_USE_RESPONSE_FILE_FOR_OBJECTS 1)
+  SET(CMAKE_CXX_USE_RESPONSE_FILE_FOR_OBJECTS 1)
+  SET(CMAKE_C_USE_RESPONSE_FILE_FOR_INCLUDES 1)
+  SET(CMAKE_CXX_USE_RESPONSE_FILE_FOR_INCLUDES 1)
+  SET(CMAKE_C_RESPONSE_FILE_LINK_FLAG "@")
+  SET(CMAKE_CXX_RESPONSE_FILE_LINK_FLAG "@")
+
+  # Specify the program to use when building static libraries
+  SET(CMAKE_C_CREATE_STATIC_LIBRARY "<CMAKE_AR> lib <TARGET> <LINK_FLAGS> <OBJECTS>")
+  SET(CMAKE_CXX_CREATE_STATIC_LIBRARY "<CMAKE_AR> lib <TARGET> <LINK_FLAGS> <OBJECTS>")
+
+  # set defination for the dll export
+  if (NOT MSVC)
+    message(FATAL "Windows build only support msvc. Which was binded by the nvcc compiler of NVIDIA.")
+  endif(NOT MSVC)
 endif(WIN32)
 
 if(NOT WITH_GOLANG)

cmake/flags.cmake

@@ -27,7 +27,6 @@ endfunction()
 
 CheckCompilerCXX11Flag()
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
-
 # safe_set_flag
 #
 # Set a compile flag only if compiler is support
@@ -71,6 +70,20 @@ macro(safe_set_nvflag flag_name)
     endif()
 endmacro()
 
+macro(safe_set_static_flag) # set c_flags and cxx_flags to static or shared
+    if (BUILD_SHARED_LIBS) 
+        return() # if build shared libs, the flags keep same with '/MD'
+    endif(BUILD_SHARED_LIBS)
+    foreach(flag_var
+        CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE
+        CMAKE_CXX_FLAGS_MINSIZEREL CMAKE_CXX_FLAGS_RELWITHDEBINFO
+        CMAKE_C_FLAGS CMAKE_C_FLAGS_DEBUG CMAKE_C_FLAGS_RELEASE
+        CMAKE_C_FLAGS_MINSIZEREL CMAKE_C_FLAGS_RELWITHDEBINFO)
+      if(${flag_var} MATCHES "/MD")
+        string(REGEX REPLACE "/MD" "/MT" ${flag_var} "${${flag_var}}")
+      endif(${flag_var} MATCHES "/MD")
+    endforeach(flag_var)
+endmacro()
 
 CHECK_CXX_SYMBOL_EXISTS(UINT64_MAX "stdint.h" UINT64_MAX_EXISTS)
 if(NOT UINT64_MAX_EXISTS)
@@ -97,9 +110,13 @@ SET(CMAKE_EXTRA_INCLUDE_FILES "")
 
 # Common flags. the compiler flag used for C/C++ sources whenever release or debug
 # Do not care if this flag is support for gcc.
+
+# https://github.com/PaddlePaddle/Paddle/issues/12773
+if (NOT WIN32)
 set(COMMON_FLAGS
     -fPIC
     -fno-omit-frame-pointer
+    -Werror
     -Wall
     -Wextra
     -Wnon-virtual-dtor
@@ -114,11 +131,6 @@ set(COMMON_FLAGS
     -Wno-error=terminate  # Warning in PADDLE_ENFORCE
 )
 
-# https://github.com/PaddlePaddle/Paddle/issues/12773
-if (NOT WIN32)
-list(APPEND COMMON_FLAGS -Werror)
-endif()
-
 set(GPU_COMMON_FLAGS
     -fPIC
     -fno-omit-frame-pointer
@@ -133,30 +145,53 @@ set(GPU_COMMON_FLAGS
     -Wno-error=array-bounds # Warnings in Eigen::array
 )
 
+else(NOT WIN32)
+set(COMMON_FLAGS
+    "/w") #disable all warnings.
+set(GPU_COMMON_FLAGS
+    "/w") #disable all warnings
+endif(NOT WIN32)
+
 if (APPLE)
     if(NOT CMAKE_CROSSCOMPILING)
         # On Mac OS X build fat binaries with x86_64 architectures by default.
         set (CMAKE_OSX_ARCHITECTURES "x86_64" CACHE STRING "Build architectures for OSX" FORCE)
     endif()
-else()
+endif(APPLE)
+
+if(LINUX)
     set(GPU_COMMON_FLAGS
         -Wall
         -Wextra
         -Werror
         ${GPU_COMMON_FLAGS})
-endif()
+endif(LINUX)
 
 if(UNIX AND NOT APPLE)
   # except apple from nix*Os family
   set(LINUX TRUE)
 endif(UNIX AND NOT APPLE)
 
-
 foreach(flag ${COMMON_FLAGS})
     safe_set_cflag(CMAKE_C_FLAGS ${flag})
     safe_set_cxxflag(CMAKE_CXX_FLAGS ${flag})
+
 endforeach()
 
 foreach(flag ${GPU_COMMON_FLAGS})
     safe_set_nvflag(${flag})
 endforeach()
+
+if(WIN32)
+# windows build turn off warnings.
+safe_set_static_flag()
+    foreach(flag_var
+        CMAKE_CXX_FLAGS CMAKE_CXX_FLAGS_DEBUG CMAKE_CXX_FLAGS_RELEASE
+        CMAKE_CXX_FLAGS_MINSIZEREL CMAKE_CXX_FLAGS_RELWITHDEBINFO
+        CMAKE_C_FLAGS CMAKE_C_FLAGS_DEBUG CMAKE_C_FLAGS_RELEASE
+        CMAKE_C_FLAGS_MINSIZEREL CMAKE_C_FLAGS_RELWITHDEBINFO)
+      if(${flag_var} MATCHES "/W3")
+        string(REGEX REPLACE "/W3" "/w" ${flag_var} "${${flag_var}}")
+      endif(${flag_var} MATCHES "/W3")
+    endforeach(flag_var)
+endif(WIN32)

doc/CMakeLists.txt

@@ -1,11 +1,9 @@
 add_custom_target(paddle_apis ALL
-                  DEPENDS paddle_v2_apis paddle_fluid_apis)
+                  DEPENDS paddle_v2_apis)
 
 add_custom_target(paddle_docs ALL
                   DEPENDS paddle_v2_docs paddle_v2_docs_cn
-                  paddle_fluid_docs paddle_fluid_docs_cn
                   paddle_mobile_docs paddle_mobile_docs_cn)
 
 add_subdirectory(v2)
-add_subdirectory(fluid)
 add_subdirectory(mobile)

paddle/contrib/float16/float16_transpiler.py

@@ -102,8 +102,8 @@ def _adjust_input(self, skip=False):
                 continue
             for input_arg in current_op.input_arg_names:
                 if input_arg in self.input_map:
-                    current_op.rename_input(input_arg,
-                                            self.input_map[input_arg])
+                    current_op._rename_input(input_arg,
+                                             self.input_map[input_arg])
 
     def _remove_unused_var(self):
         '''
@@ -187,7 +187,7 @@ def find_op(var):
                     shape=var.shape,
                     persistable=var.persistable)
                 find_op(var)
-                var.op.rename_output(var_name, tmp_var_name)
+                var.op._rename_output(var_name, tmp_var_name)
                 self.block._insert_op(
                     i,
                     type="cast",

paddle/fluid/API.spec

@@ -6,26 +6,9 @@ paddle.fluid.Program.global_block ArgSpec(args=['self'], varargs=None, keywords=
 paddle.fluid.Program.list_vars ArgSpec(args=['self'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.Program.parse_from_string ArgSpec(args=['binary_str'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.Program.to_string ArgSpec(args=['self', 'throw_on_error', 'with_details'], varargs=None, keywords=None, defaults=(False,))
-paddle.fluid.Operator.__init__ ArgSpec(args=['self', 'block', 'desc', 'type', 'inputs', 'outputs', 'attrs'], varargs=None, keywords=None, defaults=(None, None, None, None))
-paddle.fluid.Operator.all_attrs ArgSpec(args=['self'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.attr ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.attr_type ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.block_attr ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.block_attr_id ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.blocks_attr ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.blocks_attr_ids ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.has_attr ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.has_kernel ArgSpec(args=['self', 'op_type'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.input ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.output ArgSpec(args=['self', 'name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.rename_input ArgSpec(args=['self', 'old_name', 'new_name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.rename_output ArgSpec(args=['self', 'old_name', 'new_name'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.set_attr ArgSpec(args=['self', 'name', 'val'], varargs=None, keywords=None, defaults=None)
-paddle.fluid.Operator.to_string ArgSpec(args=['self', 'throw_on_error'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.default_startup_program ArgSpec(args=[], varargs=None, keywords=None, defaults=None)
 paddle.fluid.default_main_program ArgSpec(args=[], varargs=None, keywords=None, defaults=None)
 paddle.fluid.program_guard ArgSpec(args=[], varargs='args', keywords='kwds', defaults=None)
-paddle.fluid.get_var ArgSpec(args=['name', 'program'], varargs=None, keywords=None, defaults=(None,))
 paddle.fluid.name_scope ArgSpec(args=[], varargs='args', keywords='kwds', defaults=None)
 paddle.fluid.Executor.__init__ ArgSpec(args=['self', 'place'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.Executor.close ArgSpec(args=['self'], varargs=None, keywords=None, defaults=None)
@@ -170,6 +153,13 @@ paddle.fluid.layers.elementwise_mul ArgSpec(args=['x', 'y', 'out', 'axis', 'use_
 paddle.fluid.layers.elementwise_max ArgSpec(args=['x', 'y', 'out', 'axis', 'use_mkldnn', 'act', 'name'], varargs=None, keywords=None, defaults=(None, -1, False, None, None))
 paddle.fluid.layers.elementwise_min ArgSpec(args=['x', 'y', 'out', 'axis', 'use_mkldnn', 'act', 'name'], varargs=None, keywords=None, defaults=(None, -1, False, None, None))
 paddle.fluid.layers.elementwise_pow ArgSpec(args=['x', 'y', 'out', 'axis', 'use_mkldnn', 'act', 'name'], varargs=None, keywords=None, defaults=(None, -1, False, None, None))
+paddle.fluid.layers.uniform_random_batch_size_like ArgSpec(args=['input', 'shape', 'dtype', 'input_dim_idx', 'output_dim_idx', 'min', 'max', 'seed'], varargs=None, keywords=None, defaults=('float32', 0, 0, -1.0, 1.0, 0))
+paddle.fluid.layers.gaussian_random ArgSpec(args=['shape', 'mean', 'std', 'seed', 'dtype', 'use_mkldnn'], varargs=None, keywords=None, defaults=(0.0, 1.0, 0, 'float32', False))
+paddle.fluid.layers.sampling_id ArgSpec(args=['x', 'min', 'max', 'seed', 'dtype'], varargs=None, keywords=None, defaults=(0.0, 1.0, 0, 'float32'))
+paddle.fluid.layers.gaussian_random_batch_size_like ArgSpec(args=['input', 'shape', 'input_dim_idx', 'output_dim_idx', 'mean', 'std', 'seed', 'dtype'], varargs=None, keywords=None, defaults=(0, 0, 0.0, 1.0, 0, 'float32'))
+paddle.fluid.layers.sum ArgSpec(args=['x', 'use_mkldnn'], varargs=None, keywords=None, defaults=(False,))
+paddle.fluid.layers.slice ArgSpec(args=['input', 'axes', 'starts', 'ends'], varargs=None, keywords=None, defaults=None)
+paddle.fluid.layers.shape ArgSpec(args=['input'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.layers.data ArgSpec(args=['name', 'shape', 'append_batch_size', 'dtype', 'lod_level', 'type', 'stop_gradient'], varargs=None, keywords=None, defaults=(True, 'float32', 0, VarType.LOD_TENSOR, True))
 paddle.fluid.layers.open_files ArgSpec(args=['filenames', 'shapes', 'lod_levels', 'dtypes', 'thread_num', 'buffer_size', 'pass_num', 'is_test'], varargs=None, keywords=None, defaults=(None, None, 1, None))
 paddle.fluid.layers.read_file ArgSpec(args=['reader'], varargs=None, keywords=None, defaults=None)
@@ -241,13 +231,6 @@ paddle.fluid.layers.logical_and ArgSpec(args=[], varargs='args', keywords='kwarg
 paddle.fluid.layers.logical_or ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.logical_xor ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.logical_not ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.uniform_random_batch_size_like ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.gaussian_random ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.sampling_id ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.gaussian_random_batch_size_like ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.sum ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.slice ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
-paddle.fluid.layers.shape ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.maxout ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.sigmoid ArgSpec(args=['x', 'name'], varargs=None, keywords=None, defaults=(None,))
 paddle.fluid.layers.logsigmoid ArgSpec(args=['x', 'name'], varargs=None, keywords=None, defaults=(None,))
@@ -286,7 +269,7 @@ paddle.fluid.layers.iou_similarity ArgSpec(args=[], varargs='args', keywords='kw
 paddle.fluid.layers.box_coder ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.polygon_box_transform ArgSpec(args=[], varargs='args', keywords='kwargs', defaults=None)
 paddle.fluid.layers.accuracy ArgSpec(args=['input', 'label', 'k', 'correct', 'total'], varargs=None, keywords=None, defaults=(1, None, None))
-paddle.fluid.layers.auc ArgSpec(args=['input', 'label', 'curve', 'num_thresholds', 'topk'], varargs=None, keywords=None, defaults=('ROC', 4095, 1))
+paddle.fluid.layers.auc ArgSpec(args=['input', 'label', 'curve', 'num_thresholds', 'topk', 'slide_steps'], varargs=None, keywords=None, defaults=('ROC', 4095, 1, 1))
 paddle.fluid.layers.exponential_decay ArgSpec(args=['learning_rate', 'decay_steps', 'decay_rate', 'staircase'], varargs=None, keywords=None, defaults=(False,))
 paddle.fluid.layers.natural_exp_decay ArgSpec(args=['learning_rate', 'decay_steps', 'decay_rate', 'staircase'], varargs=None, keywords=None, defaults=(False,))
 paddle.fluid.layers.inverse_time_decay ArgSpec(args=['learning_rate', 'decay_steps', 'decay_rate', 'staircase'], varargs=None, keywords=None, defaults=(False,))
@@ -315,6 +298,7 @@ paddle.fluid.contrib.BeamSearchDecoder.early_stop ArgSpec(args=['self'], varargs
 paddle.fluid.contrib.BeamSearchDecoder.read_array ArgSpec(args=['self', 'init', 'is_ids', 'is_scores'], varargs=None, keywords=None, defaults=(False, False))
 paddle.fluid.contrib.BeamSearchDecoder.update_array ArgSpec(args=['self', 'array', 'value'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.contrib.memory_usage ArgSpec(args=['program', 'batch_size'], varargs=None, keywords=None, defaults=None)
+paddle.fluid.contrib.op_freq_statistic ArgSpec(args=['program'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.transpiler.DistributeTranspiler.__init__ ArgSpec(args=['self', 'config'], varargs=None, keywords=None, defaults=(None,))
 paddle.fluid.transpiler.DistributeTranspiler.get_pserver_program ArgSpec(args=['self', 'endpoint'], varargs=None, keywords=None, defaults=None)
 paddle.fluid.transpiler.DistributeTranspiler.get_pserver_programs ArgSpec(args=['self', 'endpoint'], varargs=None, keywords=None, defaults=None)

paddle/fluid/CMakeLists.txt

@@ -13,3 +13,5 @@ if(WITH_INFERENCE)
   # NOTE: please add subdirectory inference at last.
   add_subdirectory(inference)
 endif()
+
+add_subdirectory(train)

paddle/fluid/framework/CMakeLists.txt

@@ -150,11 +150,10 @@ else()
 endif()
 
 if (NOT WIN32)
-  cc_library(parallel_executor SRCS parallel_executor.cc DEPS
-          threaded_ssa_graph_executor scope_buffered_ssa_graph_executor
-          graph graph_viz_pass multi_devices_graph_pass
-          multi_devices_graph_print_pass multi_devices_graph_check_pass
-          fast_threaded_ssa_graph_executor fuse_elewise_add_act_pass)
+cc_library(parallel_executor SRCS parallel_executor.cc DEPS
+        threaded_ssa_graph_executor scope_buffered_ssa_graph_executor
+        graph build_strategy
+        fast_threaded_ssa_graph_executor)
 endif() # NOT WIN32
 
 cc_library(prune SRCS prune.cc DEPS framework_proto)

paddle/fluid/framework/details/CMakeLists.txt

@@ -54,3 +54,8 @@ cc_library(scope_buffered_ssa_graph_executor SRCS scope_buffered_ssa_graph_execu
 #        device_context reduce_op_handle )
 cc_library(fast_threaded_ssa_graph_executor SRCS fast_threaded_ssa_graph_executor.cc
         DEPS fetch_op_handle ssa_graph_executor scope simple_threadpool device_context)
+
+cc_library(build_strategy SRCS build_strategy.cc DEPS
+        graph_viz_pass multi_devices_graph_pass
+        multi_devices_graph_print_pass multi_devices_graph_check_pass
+        fuse_elewise_add_act_pass)