Replace ThreadedExecutor with FastThreadedExecutor (#16650)

* replace ThreadedExecutor with FastThreadedExecutor
test=develop

* Fix Travise CI
test=develop

* Test FastThreadedSSAGraphExecutor
test=develop

* refine parallel_ssa_graph_executor.cc
test=develop

paddle/fluid/framework/details/parallel_ssa_graph_executor.cc

@@ -106,7 +106,7 @@ ParallelSSAGraphExecutor::ParallelSSAGraphExecutor(
   VLOG(1) << "set num_threads: " << strategy_.num_threads_
           << " to run the operators of the graph on each device.";
   for (size_t i = 0; i < places.size(); ++i) {
-    executors_.emplace_back(new details::ThreadedSSAGraphExecutor(
+    executors_.emplace_back(new details::FastThreadedSSAGraphExecutor(
         strategy_, local_scopes_, {places_[i]}, graphs_.at(i).get()));
   }
 }

paddle/fluid/framework/details/parallel_ssa_graph_executor.h

@@ -14,12 +14,12 @@
 
 #pragma once
 
+#include <memory>
 #include <string>
 #include <vector>
-
 #include "ThreadPool.h"
+#include "paddle/fluid/framework/details/fast_threaded_ssa_graph_executor.h"
 #include "paddle/fluid/framework/details/multi_devices_helper.h"
-#include "paddle/fluid/framework/details/threaded_ssa_graph_executor.h"
 #include "paddle/fluid/framework/ir/graph.h"
 
 namespace paddle {
@@ -48,7 +48,8 @@ class ParallelSSAGraphExecutor : public SSAGraphExecutor {
   std::vector<platform::Place> places_;
   std::vector<std::unique_ptr<ir::Graph>> graphs_;
 
-  std::vector<std::unique_ptr<details::ThreadedSSAGraphExecutor>> executors_;
+  std::vector<std::unique_ptr<details::FastThreadedSSAGraphExecutor>>
+      executors_;
   ExceptionHolder exception_holder_;
 };