refine pg execution

paddle/fluid/framework/details/build_strategy.cc

@@ -35,8 +35,8 @@ static inline bool SeqOnlyAllReduceOps(const BuildStrategy &strategy) {
   // Should fix the allreduce op order if scheduling
   // them in multiple threads or processes to avoid hang.
   return (!strategy.enable_sequential_execution_ &&
-          strategy.num_trainers_ > 1) ||
-         strategy.enable_parallel_graph_;
+          strategy.num_trainers_ > 1) &&
+         !strategy.enable_parallel_graph_;
 }
 
 class ParallelExecutorPassBuilder : public ir::PassBuilder {
@@ -106,7 +106,9 @@ class ParallelExecutorPassBuilder : public ir::PassBuilder {
     }
 
     // Verify that the graph is correct for multi-device executor.
-    AppendPass("multi_devices_check_pass");
+    auto multi_devices_pass = AppendPass("multi_devices_check_pass");
+    multi_devices_pass->Set<bool>(kEnablePG,
+                                  new bool(strategy.enable_parallel_graph_));
 
     if (SeqOnlyAllReduceOps(strategy)) {
       AppendPass("all_reduce_deps_pass");
@@ -180,6 +182,8 @@ std::unique_ptr<ir::Graph> BuildStrategy::Apply(
                                                     &local_scopes);
       pass->Erase(kNRanks);
       pass->Set<size_t>(kNRanks, new size_t(nranks));
+      pass->Erase(kEnablePG);
+      pass->Set<bool>(kEnablePG, new bool(true));
 
 #if defined(PADDLE_WITH_CUDA) && !defined(_WIN32)
       platform::NCCLContextMap *nctx = use_cuda ? nccl_ctxs : nullptr;

paddle/fluid/framework/details/multi_devices_graph_pass.cc

@@ -36,11 +36,6 @@ namespace framework {
 namespace details {
 
 namespace {
-// TODO(panyx0718): Clean this up as well.
-// all operators. NOTE that even we use a vector here, the operators is
-// unordered.
-typedef std::vector<OpHandleBase *> GraphOps;
-const char kGraphOps[] = "ops";
 
 bool OpHaveRole(const ir::Node &node, const framework::OpRole &role) {
   return boost::get<int>(
@@ -206,7 +201,7 @@ std::unique_ptr<ir::Graph> MultiDevSSAGraphBuilderBase::ApplyImpl(
             auto &g_name = backward_vars[i + 1];
             VLOG(10) << "Bcast " << g_name << " for parameter " << p_name;
 
-            InsertCollectiveOp(&result, p_name, g_name);
+            InsertCollectiveOp(&result, node, p_name, g_name);
           }
         } catch (boost::bad_get e) {
         }
@@ -226,7 +221,7 @@ std::unique_ptr<ir::Graph> MultiDevSSAGraphBuilderBase::ApplyImpl(
    * Only variables should be the leaves of graph.
    */
   AddOutputToLeafOps(&result);
-  result.Erase(kGraphOps);
+  // result.Erase(kGraphOps);
   return graph;
 }
 
@@ -391,20 +386,34 @@ void MultiDevSSAGraphBuilderBase::CreateComputationalOp(ir::Graph *result,
 }
 
 void MultiDevSSAGraphBuilderBase::CreateAllReduceOp(
-    ir::Graph *result, const std::string &og) const {
+    ir::Graph *result, ir::Node *node, const std::string &og) const {
+  OpHandleBase *op_handle = nullptr;
+
+  auto append_allreduce_op = [&](
+      std::vector<Scope *> &scopes,
+      std::vector<platform::Place> &places) -> OpHandleBase * {
 #if defined(PADDLE_WITH_CUDA) && !defined(_WIN32)
     result->Get<GraphOps>(kGraphOps).emplace_back(new AllReduceOpHandle(
         result->CreateEmptyNode("allreduce", ir::Node::Type::kOperation),
-      local_scopes_, places_, nccl_ctxs_));
+        scopes, places, nccl_ctxs_));
 #else
     result->Get<GraphOps>(kGraphOps).emplace_back(new AllReduceOpHandle(
         result->CreateEmptyNode("allreduce", ir::Node::Type::kOperation),
-      local_scopes_, places_));
+        scopes, places));
 #endif
-  auto *op_handle = result->Get<GraphOps>(kGraphOps).back();
+    return result->Get<GraphOps>(kGraphOps).back();
+  };
+
+  if (!strategy_.enable_parallel_graph_)
+    op_handle = append_allreduce_op(local_scopes_, places_);
 
   for (size_t i = 0; i < places_.size(); ++i) {
-    auto &p = places_[i];
+    auto p = places_[i];
+    std::vector<Scope *> ss{local_scopes_[i]};
+    std::vector<platform::Place> ps{p};
+    if (strategy_.enable_parallel_graph_)
+      op_handle = append_allreduce_op(ss, ps);
+
     SetCommunicationContext(op_handle, p);
     auto &vars = result->Get<GraphVars>(kGraphVars)[i][og];
     PADDLE_ENFORCE(!vars.empty());
@@ -501,13 +510,13 @@ bool MultiDevSSAGraphBuilderBase::IsSparseGradient(
 }
 
 void AllReduceSSAGraphBuilder::InsertCollectiveOp(
-    ir::Graph *result, const std::string &p_name,
+    ir::Graph *result, ir::Node *node, const std::string &p_name,
     const std::string &g_name) const {
   if (IsSparseGradient(g_name)) {
     CreateReduceOp(result, g_name, 0);
     CreateBroadcastOp(result, g_name, 0);
   } else {
-    CreateAllReduceOp(result, g_name);
+    CreateAllReduceOp(result, node, g_name);
   }
 }
 
@@ -580,7 +589,7 @@ void ReduceSSAGraphBuilder::ResetState() const {
 }
 
 void ReduceSSAGraphBuilder::InsertCollectiveOp(
-    ir::Graph *result, const std::string &p_name,
+    ir::Graph *result, ir::Node *node, const std::string &p_name,
     const std::string &g_name) const {
   size_t cur_device_id = GetAppropriateDeviceID({g_name});
   CreateReduceOp(result, g_name, cur_device_id);
@@ -900,7 +909,7 @@ int DistSSAGraphBuilder::CreateDistTrainOp(ir::Graph *result,
   return op_dev_id;
 }
 
-void DistSSAGraphBuilder::InsertCollectiveOp(ir::Graph *result,
+void DistSSAGraphBuilder::InsertCollectiveOp(ir::Graph *result, ir::Node *node,
                                              const std::string &p_name,
                                              const std::string &g_name) const {
   size_t cur_device_id = 0;
@@ -915,7 +924,7 @@ void DistSSAGraphBuilder::InsertCollectiveOp(ir::Graph *result,
         CreateReduceOp(result, g_name, 0);
         CreateBroadcastOp(result, g_name, 0);
       } else {
-        CreateAllReduceOp(result, g_name);
+        CreateAllReduceOp(result, node, g_name);
       }
       break;
     default:
@@ -966,7 +975,8 @@ static int MultiDevSSAGraphBuilderRegister(const std::string &builder_mode) {
       .RequirePassAttr(paddle::framework::details::kPlaces)                    \
       .RequirePassAttr(paddle::framework::details::kLocalScopes)               \
       .RequirePassAttr(paddle::framework::details::kStrategy)                  \
-      .RequirePassAttr(paddle::framework::details::kNRanks)
+      .RequirePassAttr(paddle::framework::details::kNRanks)                    \
+      .RequirePassAttr(paddle::framework::details::kEnablePG)
 
 REGISTER_MULTI_DEVICES_PASS(reduce_mode_multi_devices_pass,
                             paddle::framework::details::ReduceSSAGraphBuilder);

paddle/fluid/framework/details/multi_devices_graph_pass.h

@@ -36,6 +36,7 @@ constexpr char kPlaces[] = "places";
 constexpr char kLocalScopes[] = "local_scopes";
 constexpr char kStrategy[] = "strategy";
 constexpr char kNRanks[] = "nranks";
+constexpr char kEnablePG[] = "enable_pg";
 
 class MultiDevSSAGraphBuilderBase : public ir::Pass {
  protected:
@@ -46,7 +47,8 @@ class MultiDevSSAGraphBuilderBase : public ir::Pass {
 
   virtual std::vector<ir::Node *> SortOperations(const ir::Graph &graph) const;
 
-  virtual void InsertCollectiveOp(ir::Graph *result, const std::string &p_name,
+  virtual void InsertCollectiveOp(ir::Graph *result, ir::Node *node,
+                                  const std::string &p_name,
                                   const std::string &g_name) const = 0;
 
   virtual bool DealWithSpecialOp(ir::Graph *result, ir::Node *node) const = 0;
@@ -75,7 +77,8 @@ class MultiDevSSAGraphBuilderBase : public ir::Pass {
 
   bool IsSparseGradient(const std::string &og) const;
 
-  void CreateAllReduceOp(ir::Graph *result, const std::string &og) const;
+  void CreateAllReduceOp(ir::Graph *result, ir::Node *node,
+                         const std::string &og) const;
 
   void CreateBroadcastOp(ir::Graph *result, const std::string &p_name,
                          size_t src_dev_id) const;
@@ -106,7 +109,8 @@ class MultiDevSSAGraphBuilderBase : public ir::Pass {
 
 class AllReduceSSAGraphBuilder : public MultiDevSSAGraphBuilderBase {
  protected:
-  virtual void InsertCollectiveOp(ir::Graph *result, const std::string &p_name,
+  virtual void InsertCollectiveOp(ir::Graph *result, ir::Node *node,
+                                  const std::string &p_name,
                                   const std::string &g_name) const;
 
   virtual bool DealWithSpecialOp(ir::Graph *result, ir::Node *node) const {
@@ -135,7 +139,8 @@ class ReduceSSAGraphBuilder : public BalanceVarSSAGraphBuilder {
  protected:
   virtual void Init() const;
 
-  virtual void InsertCollectiveOp(ir::Graph *result, const std::string &p_name,
+  virtual void InsertCollectiveOp(ir::Graph *result, ir::Node *node,
+                                  const std::string &p_name,
                                   const std::string &g_name) const;
 
   virtual bool DealWithSpecialOp(ir::Graph *result, ir::Node *node) const;
@@ -164,7 +169,8 @@ class DistSSAGraphBuilder : public BalanceVarSSAGraphBuilder {
 
   virtual void InsertPostprocessOps(ir::Graph *result) const;
 
-  virtual void InsertCollectiveOp(ir::Graph *result, const std::string &p_name,
+  virtual void InsertCollectiveOp(ir::Graph *result, ir::Node *node,
+                                  const std::string &p_name,
                                   const std::string &g_name) const;
 
   virtual void ResetState() const;

paddle/fluid/framework/details/multi_devices_helper.h

@@ -36,13 +36,20 @@ namespace details {
 // map from variable name to variables. The variables, who have the same name,
 // will have a differsent version. The offset in the
 // `std::vector<VarHandle*>` is the version of varaibles.
-typedef std::vector<std::unordered_map<std::string, std::vector<VarHandle*>>>
+typedef std::vector<std::unordered_map<std::string, std::vector<VarHandle *>>>
     GraphVars;
 const char kGraphVars[] = "vars";
 
 // aux variables to represent dependency. Useful to resolve data hazard.
-typedef std::unordered_set<VarHandleBase*> GraphDepVars;
+typedef std::unordered_set<VarHandleBase *> GraphDepVars;
 const char kGraphDepVars[] = "dep_vars";
+
+// TODO(panyx0718): Clean this up as well.
+// all operators. NOTE that even we use a vector here, the operators is
+// unordered.
+typedef std::vector<OpHandleBase *> GraphOps;
+const char kGraphOps[] = "ops";
+
 }  // namespace details
 }  // namespace framework
 }  // namespace paddle

paddle/fluid/framework/details/op_handle_base.h

@@ -70,6 +70,9 @@ class OpHandleBase {
     auto it = dev_ctxes_.find(place);
     return it != dev_ctxes_.end() ? it->second : nullptr;
   }
+  const std::map<platform::Place, platform::DeviceContext *> &DeviceContext() {
+    return dev_ctxes_;
+  }
 
   void SetDeviceContext(platform::Place place, platform::DeviceContext *ctx_) {
     dev_ctxes_[place] = ctx_;

paddle/fluid/framework/details/parallel_ssa_graph_executor.cc

@@ -13,11 +13,74 @@
 // limitations under the License.
 
 #include "paddle/fluid/framework/details/parallel_ssa_graph_executor.h"
+#include "paddle/fluid/framework/ir/graph_helper.h"
 
 namespace paddle {
 namespace framework {
 namespace details {
 
+std::vector<std::unique_ptr<ir::Graph>> SeparateMultiDevicesGraph(
+    const std::vector<platform::Place> &places,
+    std::unique_ptr<ir::Graph> graph) {
+  std::vector<std::unique_ptr<ir::Graph>> graphs;
+  graphs.reserve(places.size());
+  for (size_t i = 0; i < places.size(); ++i) {
+    ProgramDesc empty;
+    graphs.emplace_back(std::unique_ptr<ir::Graph>(new ir::Graph(empty)));
+    auto &g = graphs.back();
+    g->Set(kGraphVars, new GraphVars(1UL));
+    g->Set(kGraphDepVars, new GraphDepVars);
+    g->Set(kGraphOps, new GraphOps);
+  }
+
+  for (auto &op : graph->Get<GraphOps>(kGraphOps)) {
+    auto &dev_ctx = op->DeviceContext();
+    auto &p = dev_ctx.begin()->first;
+#ifdef PADDLE_WITH_CUDA
+    int dev_id = boost::get<platform::CUDAPlace>(p).device;
+    auto &dev_ops = graphs[dev_id]->Get<GraphOps>(kGraphOps);
+    auto &dev_dummys = graphs[dev_id]->Get<GraphDepVars>(kGraphDepVars);
+    dev_ops.emplace_back(op);
+    graphs[dev_id]->AddNode(graph->ReleaseNode(op->Node()).release());
+
+    for (auto &var : op->Inputs()) {
+      auto dummy_ptr = dynamic_cast<DummyVarHandle *>(var);
+      if (dummy_ptr) {
+        dev_dummys.insert(var);
+        if (graph->Nodes().count(var->Node()))
+          graphs[dev_id]->AddNode(graph->ReleaseNode(var->Node()).release());
+      }
+    }
+    for (auto &var : op->Outputs()) {
+      auto dummy_ptr = dynamic_cast<DummyVarHandle *>(var);
+      if (dummy_ptr) {
+        dev_dummys.insert(var);
+        if (graph->Nodes().count(var->Node()))
+          graphs[dev_id]->AddNode(graph->ReleaseNode(var->Node()).release());
+      }
+    }
+#else
+    PADDLE_THROW("Parallel Graph Execution only support CUDAPlace.");
+#endif
+  }
+
+  for (size_t dev_id = 0; dev_id < places.size(); ++dev_id) {
+    auto &dev_vars = graphs[dev_id]->Get<GraphVars>(kGraphVars)[0];
+    auto &origin_vars = graph->Get<GraphVars>(kGraphVars)[dev_id];
+    for (auto &name_pair : origin_vars) {
+      dev_vars.emplace(name_pair.first, name_pair.second);
+      for (auto &version_pair : name_pair.second) {
+        if (graph->Nodes().count(version_pair->Node())) {
+          graphs[dev_id]->AddNode(
+              graph->ReleaseNode(version_pair->Node()).release());
+        }
+      }
+    }
+  }
+
+  return graphs;
+}
+
 ParallelSSAGraphExecutor::ParallelSSAGraphExecutor(
     const ExecutionStrategy &strategy, const std::vector<Scope *> &local_scopes,
     const std::vector<platform::Place> &places,
@@ -37,7 +100,7 @@ ParallelSSAGraphExecutor::ParallelSSAGraphExecutor(
           << " 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(
-        strategy_, {local_scopes_[i]}, {places_[i]}, std::move(graphs_[i])));
+        strategy_, local_scopes_, {places_[i]}, std::move(graphs_.at(i))));
   }
 }
 

paddle/fluid/framework/details/parallel_ssa_graph_executor.h

@@ -14,16 +14,24 @@
 
 #pragma once
 
+#include <fstream>
+#include <sstream>
 #include <string>
 #include <vector>
 
 #include "ThreadPool.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 {
 namespace framework {
 namespace details {
 
+std::vector<std::unique_ptr<ir::Graph>> SeparateMultiDevicesGraph(
+    const std::vector<platform::Place> &places,
+    std::unique_ptr<ir::Graph> graph);
+
 class ParallelSSAGraphExecutor : public SSAGraphExecutor {
  public:
   ParallelSSAGraphExecutor(const ExecutionStrategy &strategy,
@@ -31,11 +39,14 @@ class ParallelSSAGraphExecutor : public SSAGraphExecutor {
                            const std::vector<platform::Place> &places,
                            std::vector<std::unique_ptr<ir::Graph>> &&graphs);
   ~ParallelSSAGraphExecutor() final = default;
+
   const ir::Graph &Graph() const override { return *graphs_[0]; }
 
   FeedFetchList Run(const std::vector<std::string> &fetch_tensors) override;
 
  private:
+  // std::vector<std::unique_ptr<ir::Graph>> SeparateMultiDevicesGraph();
+
   ExecutionStrategy strategy_;
   std::vector<Scope *> local_scopes_;
   std::unique_ptr<::ThreadPool> pool_{nullptr};

paddle/fluid/framework/details/threaded_ssa_graph_executor.cc

@@ -56,10 +56,10 @@ FeedFetchList ThreadedSSAGraphExecutor::Run(
       }
     }
   }
+
   for (auto &var : graph_->Get<details::GraphDepVars>(details::kGraphDepVars)) {
     InsertPendingVar(&pending_vars, ready_vars.get(), var);
   }
-
   for (auto &op : ir::FilterByNodeWrapper<OpHandleBase>(*graph_)) {
     if (op->Inputs().empty()) {  // Special case, Op has no input.
       ready_ops.insert(op);
@@ -219,7 +219,7 @@ void ThreadedSSAGraphExecutor::RunOp(
       VLOG(10) << op << " " << op->Name() << " Done ";
       running_ops_--;
       ready_var_q->Extend(op->Outputs());
-      VLOG(10) << op << " " << op->Name() << "Signal posted";
+      VLOG(10) << op << " " << op->Name() << " Signal posted";
     } catch (...) {
       exception_holder_.Catch(std::current_exception());
     }

paddle/fluid/framework/ir/graph.h

@@ -167,6 +167,14 @@ class Graph {
     return ret;
   }
 
+  std::unique_ptr<ir::Node> ReleaseNode(ir::Node *node) {
+    std::unique_ptr<ir::Node> ret;
+    ret.reset(nodes_.at(node).release());
+    nodes_.erase(node);
+    node_set_.erase(node);
+    return ret;
+  }
+
   void RemoveNode(ir::Node *node) {
     PADDLE_ENFORCE(node_set_.find(node) != node_set_.end());
     node_set_.erase(node);
@@ -183,13 +191,6 @@ class Graph {
     return nullptr;
   }
 
-  void ResolveHazard(
-      const std::map<std::string, std::vector<ir::Node *>> &var_nodes);
-
- private:
-  std::map<std::string, std::vector<ir::Node *>> InitFromProgram(
-      const ProgramDesc &program);
-
   // This method takes ownership of `node`.
   ir::Node *AddNode(ir::Node *node) {
     PADDLE_ENFORCE(node_set_.find(node) == node_set_.end());
@@ -198,6 +199,17 @@ class Graph {
     return node;
   }
 
+  bool ContainNode(ir::Node *node) {
+    return node_set_.find(node) != node_set_.end();
+  }
+
+  void ResolveHazard(
+      const std::map<std::string, std::vector<ir::Node *>> &var_nodes);
+
+ private:
+  std::map<std::string, std::vector<ir::Node *>> InitFromProgram(
+      const ProgramDesc &program);
+
   // NOTE: program_ shouldn't be exposed to user.
   const ProgramDesc program_;
   std::map<std::string, boost::any> attrs_;

paddle/fluid/framework/ir/graph_helper.h

@@ -59,7 +59,9 @@ template <typename T>
 std::vector<T *> FilterByNodeWrapper(const Graph &graph) {
   std::vector<T *> ret;
   for (ir::Node *n : graph.Nodes()) {
-    if (n->IsWrappedBy<T>()) ret.push_back(&n->Wrapper<T>());
+    if (n->IsWrappedBy<T>()) {
+      ret.push_back(&n->Wrapper<T>());
+    }
   }
   return ret;
 }

paddle/fluid/framework/parallel_executor.cc

@@ -26,6 +26,7 @@ limitations under the License. */
 #include "paddle/fluid/framework/details/parallel_ssa_graph_executor.h"
 #include "paddle/fluid/framework/details/reference_count_pass_helper.h"
 #include "paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h"
+#include "paddle/fluid/framework/details/sequential_execution_pass.h"
 #include "paddle/fluid/framework/details/threaded_ssa_graph_executor.h"
 #include "paddle/fluid/platform/profiler.h"
 
@@ -201,7 +202,6 @@ ParallelExecutor::ParallelExecutor(
   member_->use_all_reduce_ =
       build_strategy.reduce_ == BuildStrategy::ReduceStrategy::kAllReduce;
   member_->nranks_ = build_strategy.num_trainers_ * places.size();
-
   if (!member_->use_all_reduce_) {
     PADDLE_ENFORCE(places.size() > 1,
                    "If you set build_strategy.reduce with 'Reduce',"
@@ -229,9 +229,10 @@ ParallelExecutor::ParallelExecutor(
   // choice the execution strategy.
   build_strategy.enable_parallel_graph_ =
       EnableParallelGraphExecution(main_program, exec_strategy, build_strategy);
-
-  VLOG(1) << "Enable ParallelGraph Execution: "
-          << build_strategy.enable_parallel_graph_;
+  if (build_strategy.enable_parallel_graph_)
+    VLOG(0) << "The Executor would execute the graph by ParallelGraph "
+               "Execution which can get better performance,"
+            << "you can force it off by env FLAGS_enable_parallel_graph=0";
 
   if (member_->use_cuda_) {
 // Bcast Parameters to all GPUs
@@ -265,40 +266,25 @@ ParallelExecutor::ParallelExecutor(
 
   // Step 2. Convert main_program to SSA form and dependency graph. Also, insert
   // ncclOp
-  std::vector<std::unique_ptr<ir::Graph>> graphs;
+  std::unique_ptr<ir::Graph> graph;
 #if defined(PADDLE_WITH_CUDA) && !defined(_WIN32)
-  if (build_strategy.enable_parallel_graph_) {
-    for (size_t i = 0; i < member_->places_.size(); ++i) {
-      std::unique_ptr<ir::Graph> graph = build_strategy.Apply(
-          main_program, {member_->places_[i]}, loss_var_name,
-          {member_->local_scopes_[i]}, member_->nranks_, member_->use_cuda_,
-          member_->nccl_ctxs_.get());
-      graphs.push_back(std::move(graph));
-    }
-  } else {
-    std::unique_ptr<ir::Graph> graph = build_strategy.Apply(
-        main_program, member_->places_, loss_var_name, member_->local_scopes_,
-        member_->nranks_, member_->use_cuda_, member_->nccl_ctxs_.get());
-    graphs.push_back(std::move(graph));
-  }
+  graph = build_strategy.Apply(main_program, member_->places_, loss_var_name,
+                               member_->local_scopes_, member_->nranks_,
+                               member_->use_cuda_, member_->nccl_ctxs_.get());
 #else
-  std::unique_ptr<ir::Graph> graph = build_strategy.Apply(
-      main_program, member_->places_, loss_var_name, member_->local_scopes_,
-      member_->nranks_, member_->use_cuda_);
-  graphs.push_back(std::move(graph));
+  graph = build_strategy.Apply(main_program, member_->places_, loss_var_name,
+                               member_->local_scopes_, member_->nranks_,
+                               member_->use_cuda_);
 #endif
   auto max_memory_size = GetEagerDeletionThreshold();
   if (max_memory_size >= 0) {
-    for (size_t i = 0; i < graphs.size(); ++i) {
-      graphs[i] = member_->PrepareGCAndRefCnts(
-          std::move(graphs[i]), static_cast<size_t>(max_memory_size));
-    }
+    graph = member_->PrepareGCAndRefCnts(std::move(graph),
+                                         static_cast<size_t>(max_memory_size));
   }
 
   // Step 3. Create vars in each scope. Passes may also create new vars.
   //         skip control vars and empty vars
   std::vector<details::VariableInfo> var_infos;
-  for (auto &graph : graphs) {
   for (auto &node : graph->Nodes()) {
     if (node->IsVar() && !node->IsCtrlVar() && node->Var()) {
       var_infos.emplace_back();
@@ -307,16 +293,15 @@ ParallelExecutor::ParallelExecutor(
       var_infos.back().persistable_ = node->Var()->Persistable();
     }
   }
-  }
 
   // If the loss_var_name is given, the number of graph should be only one.
   if (loss_var_name.size()) {
-    size_t graph_num = ir::GraphNum(*graphs[0]);
+    size_t graph_num = ir::GraphNum(*graph);
     if (graph_num > 1) {
       LOG(WARNING)
           << "The number of graph should be only one, "
              "but the current graph has "
-          << ir::GraphNum(*graphs[0])
+          << ir::GraphNum(*graph)
           << " sub_graphs. If you want to see the nodes of the "
              "sub_graphs, you should use 'FLAGS_print_sub_graph_dir' "
              "to specify the output dir. NOTES: if you not do training, "
@@ -325,18 +310,30 @@ ParallelExecutor::ParallelExecutor(
   }
 
   if (build_strategy.enable_parallel_graph_) {
+    auto parallel_graph =
+        details::SeparateMultiDevicesGraph(member_->places_, std::move(graph));
+    auto seq_allreduce_pass =
+        ir::PassRegistry::Instance().Get("all_reduce_deps_pass");
+    seq_allreduce_pass->Erase(details::kAllOpDescs);
+    seq_allreduce_pass->Set<const std::vector<OpDesc *>>(
+        details::kAllOpDescs,
+        new std::vector<OpDesc *>(main_program.Block(0).AllOps()));
+    for (size_t i = 0; i < parallel_graph.size(); ++i) {
+      parallel_graph[i] =
+          seq_allreduce_pass->Apply(std::move(parallel_graph[i]));
+    }
     member_->executor_.reset(new details::ParallelSSAGraphExecutor(
         exec_strategy, member_->local_scopes_, member_->places_,
-        std::move(graphs)));
+        std::move(parallel_graph)));
   } else {
     if (exec_strategy.type_ == ExecutionStrategy::kDefault) {
       member_->executor_.reset(new details::ThreadedSSAGraphExecutor(
           exec_strategy, member_->local_scopes_, member_->places_,
-          std::move(graphs[0])));
+          std::move(graph)));
     } else {
       member_->executor_.reset(new details::FastThreadedSSAGraphExecutor(
           exec_strategy, member_->local_scopes_, member_->places_,
-          std::move(graphs[0])));
+          std::move(graph)));
     }
   }
 
@@ -487,8 +484,8 @@ bool ParallelExecutor::EnableParallelGraphExecution(
     }
   }
 
-  if (!member_->use_all_reduce_ || !member_->use_cuda_)
-    enable_parallel_graph = false;
+  // if (!member_->use_all_reduce_ || !member_->use_cuda_)
+  if (!member_->use_all_reduce_) enable_parallel_graph = false;
 
   if (build_strategy.enable_sequential_execution_ ||
       exec_strategy.type_ == ExecutionStrategy::ExecutorType::kExperimental)

python/paddle/fluid/tests/unittests/parallel_executor_test_base.py

@@ -72,6 +72,7 @@ class TestParallelExecutorBase(unittest.TestCase):
             exe.run(startup)
             exec_strategy = fluid.ExecutionStrategy()
             exec_strategy.allow_op_delay = allow_op_delay
+            exec_strategy.num_threads = 1
             if use_fast_executor:
                 exec_strategy.use_experimental_executor = True
             build_strategy = fluid.BuildStrategy()
@@ -99,7 +100,7 @@ class TestParallelExecutorBase(unittest.TestCase):
             first_loss, = run_executor(
                 exe=exe, binary=binary, feed=feed_dict, fetch_list=[loss.name])
 
-            for i in range(iter):
+            for _ in range(iter):
                 run_executor(
                     exe=exe, binary=binary, feed=feed_dict, fetch_list=[])
 

python/paddle/fluid/tests/unittests/test_parallel_executor_pg.py

+# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+from __future__ import print_function
+
+import unittest
+
+import numpy as np
+import os
+os.environ['FLAGS_enable_parallel_graph'] = str(1)
+import paddle.fluid.core as core
+import os
+import paddle.fluid as fluid
+from parallel_executor_test_base import TestParallelExecutorBase
+
+
+def simple_fc_net(use_feed):
+    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
+    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    hidden = img
+    for _ in range(4):
+        hidden = fluid.layers.fc(
+            hidden,
+            size=200,
+            act='tanh',
+            bias_attr=fluid.ParamAttr(
+                initializer=fluid.initializer.Constant(value=1.0)))
+    prediction = fluid.layers.fc(hidden, size=10, act='softmax')
+    loss = fluid.layers.cross_entropy(input=prediction, label=label)
+    loss = fluid.layers.mean(loss)
+    return loss
+
+
+class TestMNIST(TestParallelExecutorBase):
+    @classmethod
+    def setUpClass(cls):
+        os.environ['CPU_NUM'] = str(4)
+
+    def _init_data(self):
+        np.random.seed(5)
+        img = np.random.random(size=[32, 784]).astype(np.float32)
+        label = np.ones(shape=[32, 1], dtype='int64')
+        return img, label
+
+    # simple_fc
+    def check_simple_fc_convergence(self, use_cuda, use_reduce=False):
+        if use_cuda and not core.is_compiled_with_cuda():
+            return
+
+        img, label = self._init_data()
+
+        self.check_network_convergence(
+            simple_fc_net,
+            feed_dict={"image": img,
+                       "label": label},
+            use_cuda=use_cuda,
+            use_reduce=use_reduce)
+
+    def test_simple_fc(self):
+        # use_cuda
+        self.check_simple_fc_convergence(True)
+
+    def check_simple_fc_parallel_accuracy(self, use_cuda):
+        if use_cuda and not core.is_compiled_with_cuda():
+            return
+
+        img, label = self._init_data()
+
+        single_first_loss, single_last_loss = self.check_network_convergence(
+            method=simple_fc_net,
+            seed=1,
+            feed_dict={"image": img,
+                       "label": label},
+            use_cuda=use_cuda,
+            use_parallel_executor=False)
+        parallel_first_loss, parallel_last_loss = self.check_network_convergence(
+            method=simple_fc_net,
+            seed=1,
+            feed_dict={"image": img,
+                       "label": label},
+            use_cuda=use_cuda,
+            use_parallel_executor=True)
+
+        self.assertAlmostEquals(
+            np.mean(parallel_first_loss),
+            single_first_loss,
+            delta=1e-6, )
+        self.assertAlmostEquals(
+            np.mean(parallel_last_loss), single_last_loss, delta=1e-6)
+
+    def test_simple_fc_parallel_accuracy(self):
+        self.check_simple_fc_parallel_accuracy(True)
+
+
+if __name__ == '__main__':
+    unittest.main()