Feature/op_fuse_pass (#12440)

* Add Preface

* Add demo code

* Save file

* Refine code

* seems can work

* use elementwise strategy

* Use ElementwiseComputeEx

* Add comments

* extract functions from operator

* Refine code

* Follow comment

* code refine

* add op_fuse  pass

* add backward

* code refine

* use TopologySortOperations

* follow comments

* refine IsFusible

* code enhance

* fix op_fusion_pass

* refine code

* refine fuse_elemwise_act_op

* adjust the input and output

* refine logic

* add intermediate_edge

* disable inplace

* follow comments

* refine logic

* follow comments

* Remove the removable IntermediateOut

* change strategy

* code refine

* enable fuse backward

* code refine

* code refine

* rename unit test

* follow comments

Author: chengduo

paddle/fluid/framework/CMakeLists.txt

@@ -148,13 +148,13 @@ if(WITH_DISTRIBUTE)
 else()
   cc_library(executor SRCS executor.cc DEPS op_registry device_context scope framework_proto glog lod_rank_table feed_fetch_method graph_to_program_pass)
 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)
+  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)
 endif() # NOT WIN32
 
 cc_library(prune SRCS prune.cc DEPS framework_proto)

paddle/fluid/framework/details/build_strategy.h

@@ -54,6 +54,8 @@ struct BuildStrategy {
 
   std::string debug_graphviz_path_{""};
 
+  bool fuse_elewise_add_act_ops_{false};
+
   bool enable_data_balance_{false};
 };
 

paddle/fluid/framework/ir/CMakeLists.txt

@@ -37,6 +37,8 @@ pass_library(fc_lstm_fuse_pass inference)
 pass_library(fc_gru_fuse_pass inference)
 pass_library(seq_concat_fc_fuse_pass inference)
 
+cc_library(fuse_elewise_add_act_pass SRCS fuse_elewise_add_act_pass.cc DEPS pass graph_pattern_detector )
+
 set(GLOB_PASS_LIB ${PASS_LIBRARY} CACHE INTERNAL "Global PASS library")
 
 cc_test(pass_test SRCS pass_test.cc DEPS graph pass graph_helper)