polish code

add unittest model containing while_op
remove unnecessary codes
test=develop

paddle/fluid/framework/CMakeLists.txt

@@ -72,6 +72,8 @@ cc_library(lod_tensor SRCS lod_tensor.cc DEPS ddim place tensor framework_proto
 cc_test(lod_tensor_test SRCS lod_tensor_test.cc DEPS lod_tensor memory)
 nv_test(lod_tensor_gpu_test SRCS lod_tensor_test.cu DEPS lod_tensor)
 
+cc_library(garbage_collector SRCS garbage_collector.cc DEPS device_context memory)
+
 cc_library(reader SRCS reader.cc DEPS lod_tensor ddim)
 cc_test(reader_test SRCS reader_test.cc DEPS reader)
 
@@ -164,7 +166,7 @@ cc_library(variable_helper SRCS variable_helper.cc DEPS lod_tensor)
 cc_library(naive_executor SRCS naive_executor.cc DEPS op_registry device_context scope framework_proto glog lod_rank_table feed_fetch_method graph_to_program_pass variable_helper)
 
 if(WITH_DISTRIBUTE)
-  cc_library(executor SRCS executor.cc DEPS op_registry device_context scope framework_proto glog lod_rank_table feed_fetch_method sendrecvop_grpc cares grpc++_unsecure grpc_unsecure gpr graph_to_program_pass variable_helper)
+  cc_library(executor SRCS executor.cc DEPS op_registry device_context scope framework_proto glog lod_rank_table feed_fetch_method sendrecvop_grpc cares grpc++_unsecure grpc_unsecure gpr graph_to_program_pass variable_helper garbage_collector)
   set(DISTRIBUTE_COMPILE_FLAGS "-Wno-non-virtual-dtor -Wno-error=non-virtual-dtor -Wno-error=delete-non-virtual-dtor")
   set_source_files_properties(executor.cc PROPERTIES COMPILE_FLAGS ${DISTRIBUTE_COMPILE_FLAGS})
 else()

paddle/fluid/framework/details/CMakeLists.txt

@@ -33,9 +33,10 @@ cc_library(fuse_vars_op_handle SRCS fuse_vars_op_handle.cc DEPS op_handle_base s
 
 cc_library(modify_op_lock_and_record_event_pass SRCS modify_op_lock_and_record_event_pass.cc DEPS computation_op_handle op_graph_view multi_devices_helper)
 
-cc_library(eager_deletion_op_handle SRCS eager_deletion_op_handle.cc DEPS lod_tensor selected_rows op_handle_base)
+cc_library(reference_count_pass_helper SRCS reference_count_pass_helper.cc DEPS garbage_collector computation_op_handle)
+cc_library(eager_deletion_op_handle SRCS eager_deletion_op_handle.cc DEPS lod_tensor selected_rows reference_count_pass_helper)
 cc_library(eager_deletion_pass SRCS eager_deletion_pass.cc DEPS computation_op_handle eager_deletion_op_handle graph graph_helper pass)
-cc_library(reference_count_pass SRCS reference_count_pass.cc DEPS computation_op_handle graph graph_helper pass op_graph_view)
+cc_library(reference_count_pass SRCS reference_count_pass.cc DEPS computation_op_handle graph graph_helper pass op_graph_view reference_count_pass_helper)
 
 cc_library(sequential_execution_pass SRCS sequential_execution_pass.cc DEPS graph graph_helper pass)
 cc_library(all_reduce_deps_pass SRCS all_reduce_deps_pass.cc DEPS graph graph_helper pass)

paddle/fluid/framework/details/eager_deletion_op_handle.cc

@@ -26,8 +26,8 @@ namespace details {
 
 EagerDeletionOpHandle::EagerDeletionOpHandle(
     ir::Node *node, const Scope *scope, const platform::Place &place,
-    const std::unordered_set<std::string> &var_names,
-    GarbageCollector<Tensor> *gc, AtomicReferenceCountMap *ref_cnts)
+    const std::unordered_set<std::string> &var_names, GarbageCollector *gc,
+    AtomicReferenceCountMap *ref_cnts)
     : OpHandleBase(node),
       scope_(scope),
       var_names_(var_names),
@@ -35,9 +35,9 @@ EagerDeletionOpHandle::EagerDeletionOpHandle(
       ref_cnts_(ref_cnts) {
 #ifdef PADDLE_WITH_CUDA
   if (platform::is_gpu_place(place)) {
-    dev_ctx_ = static_cast<platform::CUDADeviceContext *>(
+    dev_ctx_ = reinterpret_cast<platform::CUDADeviceContext *>(
         platform::DeviceContextPool::Instance().Get(place));
-    if (dynamic_cast<StreamGarbageCollector<Tensor> *>(gc_)) {
+    if (dynamic_cast<StreamGarbageCollector *>(gc_)) {
       platform::CUDADeviceGuard guard(
           boost::get<platform::CUDAPlace>(place).device);
       PADDLE_ENFORCE(cudaEventCreateWithFlags(&event_, cudaEventDisableTiming));
@@ -61,10 +61,11 @@ std::string EagerDeletionOpHandle::Name() const { return "eager_deletion"; }
 
 void EagerDeletionOpHandle::RunImpl() {
   auto *exec_scope = scope_->FindVar(kLocalExecScopeName)->Get<Scope *>();
-  std::vector<Tensor *> tensors;
+  std::deque<std::shared_ptr<memory::Allocation>> garbages;
   for (auto &name : var_names_) {
     auto it = ref_cnts_->find(name);
-    if (it == ref_cnts_->end()) {
+    // Var not found, not reference count has not decreased to 0
+    if (it == ref_cnts_->end() || it->second.fetch_sub(1) != 1) {
       continue;
     }
 
@@ -73,43 +74,44 @@ void EagerDeletionOpHandle::RunImpl() {
       continue;
     }
 
+    VLOG(2) << "Erase variable " << name;
+
     if (var->IsType<LoDTensor>()) {
-      if (it->second.fetch_sub(1) == 1) {
-        tensors.emplace_back(var->GetMutable<LoDTensor>());
-      }
+      garbages.emplace_back(var->GetMutable<LoDTensor>()->MoveMemory());
     } else if (var->IsType<SelectedRows>()) {
-      if (it->second.fetch_sub(1) == 1) {
-        tensors.emplace_back(var->GetMutable<SelectedRows>()->mutable_value());
-      }
+      garbages.emplace_back(
+          var->GetMutable<SelectedRows>()->mutable_value()->MoveMemory());
     } else if (var->IsType<LoDTensorArray>()) {
-      if (it->second.fetch_sub(1) == 1) {
       auto *tensor_arr = var->GetMutable<LoDTensorArray>();
       for (auto &t : *tensor_arr) {
-          tensors.emplace_back(&t);
-        }
+        garbages.emplace_back(t.MoveMemory());
       }
+    } else {
+      PADDLE_THROW("Type %s of %s is not supported eager deletion",
+                   var->Type().name(), name);
     }
   }
 
-  if (!tensors.empty()) {
-    ClearTensors(tensors);
+  if (!garbages.empty()) {
+    ClearGarbages(&garbages);
   }
 }
 
-void EagerDeletionOpHandle::ClearTensors(const std::vector<Tensor *> &tensors) {
+void EagerDeletionOpHandle::ClearGarbages(
+    std::deque<std::shared_ptr<memory::Allocation>> *garbages) {
 #ifdef PADDLE_WITH_CUDA
   if (event_) {
     auto compute_stream = dev_ctx_->stream();
     auto callback_stream =
-        static_cast<StreamGarbageCollector<Tensor> *>(gc_)->stream();
+        reinterpret_cast<StreamGarbageCollector *>(gc_)->stream();
     auto callback_func = [=]() {
       PADDLE_ENFORCE(cudaEventRecord(event_, compute_stream));
       PADDLE_ENFORCE(cudaStreamWaitEvent(callback_stream, event_, 0));
     };
-    gc_->Add(tensors, callback_func);
+    gc_->Add(std::move(*garbages), callback_func);
   } else {
 #endif
-    gc_->Add(tensors);
+    gc_->Add(std::move(*garbages));
 #ifdef PADDLE_WITH_CUDA
   }
 #endif

paddle/fluid/framework/details/eager_deletion_op_handle.h

@@ -14,8 +14,8 @@
 
 #pragma once
 
+#include <deque>
 #include <string>
-#include <vector>
 #include "paddle/fluid/framework/details/op_handle_base.h"
 #include "paddle/fluid/framework/details/reference_count_pass_helper.h"
 
@@ -30,7 +30,7 @@ class EagerDeletionOpHandle : public OpHandleBase {
   EagerDeletionOpHandle(ir::Node *node, const Scope *scope,
                         const platform::Place &place,
                         const std::unordered_set<std::string> &var_names,
-                        GarbageCollector<Tensor> *gc,
+                        GarbageCollector *gc,
                         AtomicReferenceCountMap *ref_cnts);
 
   ~EagerDeletionOpHandle();
@@ -41,11 +41,11 @@ class EagerDeletionOpHandle : public OpHandleBase {
   void RunImpl() override;
 
  private:
-  void ClearTensors(const std::vector<Tensor *> &tensors);
+  void ClearGarbages(std::deque<std::shared_ptr<memory::Allocation>> *garbages);
 
   const Scope *scope_;
   std::unordered_set<std::string> var_names_;
-  GarbageCollector<Tensor> *gc_;       // not own
+  GarbageCollector *gc_;               // not own
   AtomicReferenceCountMap *ref_cnts_;  // not own
 #ifdef PADDLE_WITH_CUDA
   platform::CUDADeviceContext *dev_ctx_{nullptr};

paddle/fluid/framework/details/eager_deletion_pass.cc

@@ -28,17 +28,21 @@ namespace details {
 
 std::unique_ptr<ir::Graph> EagerDeletionPass::ApplyImpl(
     std::unique_ptr<ir::Graph> graph) const {
-  const auto &vars = graph->Get<GraphVars>(kGraphVars);
-
   auto &ref_cnts =
       Get<std::vector<AtomicReferenceCountMap>>(kRuntimeReferenceCount);
+  PADDLE_ENFORCE(ref_cnts.empty(),
+                 "kRuntimeReferenceCount should be initialized here!");
+
+  const auto &vars = graph->Get<GraphVars>(kGraphVars);
+  ref_cnts.resize(vars.size());
+
   const auto &last_live_ops =
       Get<std::vector<LastLiveOpsOfVars>>(kLastLiveOpsOfVars);
-  auto &gcs = Get<GarbageCollectorMap>(kGarbageCollector);
+  const auto &gcs = Get<GarbageCollectorMap>(kGarbageCollector);
   const auto &places = Get<std::vector<platform::Place>>(kAllPlaces);
 
-  ref_cnts = std::vector<AtomicReferenceCountMap>(vars.size());
-
+  // a reverse map of last_live_ops
+  //   i.e., last op --> variable names which can be deleted.
   std::unordered_map<ComputationOpHandle *, std::unordered_set<std::string>>
       op_vars_map;
 
@@ -58,8 +62,8 @@ std::unique_ptr<ir::Graph> EagerDeletionPass::ApplyImpl(
     auto *eager_deletion_node =
         graph->CreateEmptyNode("eager_deletion", ir::Node::Type::kOperation);
     auto *eager_deletion_op = new EagerDeletionOpHandle(
-        eager_deletion_node, op->GetScope(), op->GetPlace(),
-        std::move(var_names), gcs.at(places[op->GetScopeIdx()]).get(),
+        eager_deletion_node, op->GetScope(), op->GetPlace(), var_names,
+        gcs.at(places[op->GetScopeIdx()]).get(),
         &(ref_cnts[op->GetScopeIdx()]));
 
     auto it = std::find_if(

paddle/fluid/framework/details/op_graph_view.cc

@@ -42,6 +42,7 @@ void OpGraphView::Build(const std::vector<OpHandleBase *> &ops) {
 
 std::unordered_set<OpHandleBase *> OpGraphView::AllOps() const {
   std::unordered_set<OpHandleBase *> ret;
+  ret.reserve(preceding_ops_.size());
   for (auto &pair : preceding_ops_) {
     ret.insert(pair.first);
   }

paddle/fluid/framework/details/reference_count_pass.cc

@@ -29,15 +29,17 @@ namespace paddle {
 namespace framework {
 namespace details {
 
-class OpRelationDetector {
- public:
+// A functor to shrink/remove operators who depend on other operators in a set
+class ShrinkDepsOpFunctor {
+ private:
   enum RelationShip { kSame = 0, kNoDeps = 1, kBefore = 2, kAfter = 3 };
 
-  explicit OpRelationDetector(const std::vector<OpHandleBase *> &all_ops)
+ public:
+  explicit ShrinkDepsOpFunctor(const std::vector<OpHandleBase *> &all_ops)
       : graph_(all_ops) {}
 
   template <typename OpSet>
-  OpSet MaxNoDepOps(const OpSet &op_set) const {
+  OpSet operator()(const OpSet &op_set) const {
     using KeyType = typename OpSet::key_type;
     static_assert(
         std::is_base_of<OpHandleBase,
@@ -51,7 +53,7 @@ class OpRelationDetector {
     auto not_before = [](RelationShip r) { return r != kBefore; };
     for (size_t i = 0; i < rels.size(); ++i) {
       if (std::all_of(rels[i].begin(), rels[i].end(), not_before)) {
-        ret.insert(static_cast<KeyType>(ops[i]));
+        ret.emplace(static_cast<KeyType>(ops[i]));
       }
     }
     return ret;
@@ -59,7 +61,7 @@ class OpRelationDetector {
 
  private:
   std::vector<std::vector<RelationShip>> GetRelations(
-      const std::vector<OpHandleBase *> ops) const {
+      const std::vector<OpHandleBase *> &ops) const {
     std::unordered_map<OpHandleBase *, size_t> op_to_idx;
     for (size_t i = 0; i < ops.size(); ++i) {
       PADDLE_ENFORCE(graph_.HasOp(ops[i]), "Op does not exist in graph");
@@ -112,6 +114,10 @@ class OpRelationDetector {
   const OpGraphView graph_;
 };
 
+/**
+ * Find the nearest downstream computation op handle. If the op is a
+ * computation op, just return itself.
+ */
 static ComputationOpHandle *FindNextComputationOpHandleOrReturnItself(
     OpHandleBase *op, size_t scope_idx) {
   std::queue<OpHandleBase *> q;
@@ -134,33 +140,87 @@ static ComputationOpHandle *FindNextComputationOpHandleOrReturnItself(
   return nullptr;
 }
 
+static std::unordered_set<ComputationOpHandle *>
+ExtractComputationOpFromLastLivedVar(VarHandle *var, size_t scope_idx,
+                                     const ShrinkDepsOpFunctor &shrink_func,
+                                     bool *ok) {
+  // stage one. Get last op for variable.
+  std::unordered_set<OpHandleBase *> candidates;
+  {
+    if (var->PendingOps().empty() && var->GeneratedOp()) {
+      // No operator depends on this variable. So the last operator is the op
+      // who generates this variable.
+      candidates.emplace(var->GeneratedOp());
+    } else {
+      candidates = var->PendingOps();
+    }
+
+    // No pending ops or generated op is nullptr
+    if (candidates.empty()) {
+      *ok = false;
+      return {};
+    }
+  }
+
+  // stage two. Try to cast them to computation op.
+  // return (*ok=false) when failed.
+  //
+  // The reason why we cannot make any types of op handle to be the last lived
+  // op is:
+  //    some op handle may operate on many DeviceContext, however, our garbage
+  //    collector can only wait one DeviceContext for now. So currently, we wait
+  //    the nearest compute op.
+  std::unordered_set<ComputationOpHandle *> computation_op;
+  {
+    for (auto *op : candidates) {
+      auto *compute_op =
+          FindNextComputationOpHandleOrReturnItself(op, scope_idx);
+      if (compute_op == nullptr) {
+        *ok = false;
+        return {};
+      }
+      computation_op.emplace(compute_op);
+    }
+  }
+
+  // stage three. Try to shrink computation op if they depend on each other.
+  // Get the smallest set of the most ops.
+  *ok = true;
+  return shrink_func(computation_op);
+}
+
+static VarDesc *TryGetLatestVarDesc(const std::vector<VarHandle *> &vars) {
+  VarDesc *var_desc = nullptr;
+  std::find_if(vars.rbegin(), vars.rend(), [&](VarHandle *var_handle) -> bool {
+    var_desc = var_handle->Node()->Var();
+    return var_desc != nullptr;
+  });
+  return var_desc;
+}
+
 std::unique_ptr<ir::Graph> ReferenceCountPass::ApplyImpl(
     std::unique_ptr<ir::Graph> graph) const {
-  auto &vars = graph->Get<GraphVars>(kGraphVars);
   auto &ref_cnts = Get<std::vector<ReferenceCountMap>>(kGlobalReferenceCount);
   auto &last_live_ops_of_vars =
       Get<std::vector<LastLiveOpsOfVars>>(kLastLiveOpsOfVars);
 
-  last_live_ops_of_vars = std::vector<LastLiveOpsOfVars>(vars.size());
-  ref_cnts = std::vector<ReferenceCountMap>(vars.size());
+  PADDLE_ENFORCE(last_live_ops_of_vars.empty() && ref_cnts.empty(),
+                 "Last Live Ops and Reference Counts of vars should be "
+                 "initialized at here.");
 
-  OpRelationDetector detector(ir::FilterByNodeWrapper<OpHandleBase>(*graph));
+  const auto &vars = graph->Get<GraphVars>(kGraphVars);
 
-  for (size_t i = 0; i < vars.size(); ++i) {
-    for (auto &name_var_pair : vars[i]) {
-      if (name_var_pair.second.empty()) {
-        continue;
-      }
+  last_live_ops_of_vars.resize(vars.size());
+  ref_cnts.resize(vars.size());
 
-      const std::string &var_name = name_var_pair.first;
-      auto *last_ver_var = name_var_pair.second.back();
+  ShrinkDepsOpFunctor shrink_func(
+      ir::FilterByNodeWrapper<OpHandleBase>(*graph));
 
-      VarDesc *var_desc = nullptr;
-      std::find_if(name_var_pair.second.rbegin(), name_var_pair.second.rend(),
-                   [&](VarHandle *var_handle) -> bool {
-                     var_desc = var_handle->Node()->Var();
-                     return var_desc != nullptr;
-                   });
+  for (size_t i = 0; i < vars.size(); ++i) {
+    for (auto &name_var_pair : vars[i]) {
+      // Whether this variable can be reused or deleted? If not, we do not
+      // compute reference counts and dependencies.
+      VarDesc *var_desc = TryGetLatestVarDesc(name_var_pair.second);
 
       if (var_desc == nullptr || var_desc->Persistable()) {
         continue;
@@ -170,50 +230,20 @@ std::unique_ptr<ir::Graph> ReferenceCountPass::ApplyImpl(
       if (var_type != proto::VarType::LOD_TENSOR &&
           var_type != proto::VarType::SELECTED_ROWS &&
           var_type != proto::VarType::LOD_TENSOR_ARRAY) {
+        // Var type cannot be deleted
         continue;
       }
 
-      std::unordered_set<ComputationOpHandle *> last_live_op;
-      auto add_last_live_op = [&](OpHandleBase *op) -> bool {
-        auto *compute_op = FindNextComputationOpHandleOrReturnItself(op, i);
-        if (compute_op) {
-          last_live_op.insert(compute_op);
-          return true;
-        } else {
-          return false;
-        }
-      };
-
-      bool can_delete = false;
-      auto &pending_ops = last_ver_var->PendingOps();
-      if (pending_ops.empty()) {
-        auto *generated_op = last_ver_var->GeneratedOp();
-        if (generated_op && add_last_live_op(generated_op)) {
-          can_delete = true;
-        }
-      } else {
-        can_delete = true;
-        for (auto *pending_op : pending_ops) {
-          if (!add_last_live_op(pending_op)) {
-            can_delete = false;
-            break;
-          }
-        }
-      }
-
-      if (can_delete) {
-        size_t original_size = last_live_op.size();
-        last_live_op = detector.MaxNoDepOps(last_live_op);
-        if (last_live_op.size() != original_size) {
-          VLOG(10) << "Shrink last living op number of " << var_name << " from "
-                   << original_size << " to " << last_live_op.size();
-        }
-
-        PADDLE_ENFORCE(!last_live_op.empty(),
-                       "Last living ops of %s cannot be empty", var_name);
+      bool ok;
+      auto result = ExtractComputationOpFromLastLivedVar(
+          name_var_pair.second.back(), i, shrink_func, &ok);
 
-        ref_cnts[i].emplace(var_name, last_live_op.size());
-        last_live_ops_of_vars[i].emplace(var_name, std::move(last_live_op));
+      if (ok) {
+        auto &var_name = name_var_pair.first;
+        PADDLE_ENFORCE(!result.empty(), "Last living ops of %s cannot be empty",
+                       var_name);
+        ref_cnts[i].emplace(var_name, result.size());
+        last_live_ops_of_vars[i].emplace(var_name, std::move(result));
       }
     }
   }

paddle/fluid/framework/details/reference_count_pass_helper.cc

+// 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.
+
+#include "paddle/fluid/framework/details/reference_count_pass_helper.h"
+
+namespace paddle {
+namespace framework {
+namespace details {}  // namespace details
+}  // namespace framework
+}  // namespace paddle

paddle/fluid/framework/details/reference_count_pass_helper.h

@@ -18,10 +18,10 @@
 #include <map>
 #include <string>
 #include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 #include "paddle/fluid/framework/garbage_collector.h"
-#include "paddle/fluid/framework/tensor.h"
 
 namespace paddle {
 namespace framework {
@@ -35,7 +35,7 @@ using AtomicReferenceCountMap =
     std::unordered_map<std::string, std::atomic<size_t>>;
 
 using GarbageCollectorMap =
-    std::map<platform::Place, std::unique_ptr<GarbageCollector<Tensor>>>;
+    std::map<platform::Place, std::unique_ptr<GarbageCollector>>;
 
 const char kGlobalReferenceCount[] = "global_reference_count";
 const char kRuntimeReferenceCount[] = "runtime_reference_count";

paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.cc

@@ -30,20 +30,7 @@ ScopeBufferedSSAGraphExecutor::ScopeBufferedSSAGraphExecutor(
       underlying_executor_(std::move(underlying_executor)),
       local_scopes_(std::move(local_scopes)),
       var_infos_(std::move(var_infos)),
-      places_(std::move(places)) {
-  if (Graph().Has(details::kGarbageCollector)) {
-    gc_ = &(Graph().Get<GarbageCollectorMap>(details::kGarbageCollector));
-  }
-}
-
-void ScopeBufferedSSAGraphExecutor::WaitAllGarbageCollectors() {
-  if (gc_) {
-    for (auto &gc_pair : *gc_) {
-      gc_pair.second->Wait();
-      gc_pair.second->Reset();
-    }
-  }
-}
+      places_(std::move(places)) {}
 
 FeedFetchList ScopeBufferedSSAGraphExecutor::Run(
     const std::vector<std::string> &fetch_tensors) {
@@ -83,19 +70,15 @@ FeedFetchList ScopeBufferedSSAGraphExecutor::Run(
       drop_scope_counter_ == strategy_.num_iteration_per_drop_scope_) {
     drop_scope_counter_ = 0;
     // Wait All computational streams
-    for (auto &p : places_) {
+    for (auto p : places_) {
       platform::DeviceContextPool::Instance().Get(p)->Wait();
     }
-    WaitAllGarbageCollectors();
     for (auto &scope : local_scopes_) {
       auto &local_scope =
           *scope->Var(details::kLocalExecScopeName)->GetMutable<Scope *>();
       scope->DeleteScope(local_scope);
     }
-  } else {
-    WaitAllGarbageCollectors();
   }
-
   if (eptr) {
     std::rethrow_exception(eptr);
   } else {

paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h

@@ -21,11 +21,9 @@
 #include "paddle/fluid/framework/details/var_handle.h"
 
 #include "paddle/fluid/framework/details/execution_strategy.h"
-#include "paddle/fluid/framework/details/reference_count_pass_helper.h"
 #include "paddle/fluid/framework/details/ssa_graph_executor.h"
 #include "paddle/fluid/framework/scope.h"
 #include "paddle/fluid/platform/place.h"
-
 namespace paddle {
 namespace framework {
 namespace details {
@@ -50,8 +48,6 @@ class ScopeBufferedSSAGraphExecutor : public SSAGraphExecutor {
   FeedFetchList Run(const std::vector<std::string>& fetch_tensors) override;
 
  private:
-  void WaitAllGarbageCollectors();
-
   size_t drop_scope_counter_{0};
 
   ExecutionStrategy strategy_;
@@ -59,8 +55,6 @@ class ScopeBufferedSSAGraphExecutor : public SSAGraphExecutor {
   std::vector<Scope*> local_scopes_;
   std::vector<VariableInfo> var_infos_;
   std::vector<platform::Place> places_;
-
-  GarbageCollectorMap* gc_{nullptr};
 };
 }  // namespace details
 }  // namespace framework

paddle/fluid/framework/executor.cc

@@ -13,6 +13,7 @@ See the License for the specific language governing permissions and
 limitations under the License. */
 
 #include "paddle/fluid/framework/executor.h"
+#include <deque>
 
 #include "paddle/fluid/framework/feed_fetch_method.h"
 #include "paddle/fluid/framework/lod_rank_table.h"
@@ -83,31 +84,37 @@ ExecutorPrepareContext::~ExecutorPrepareContext() {
 }
 
 static void DeleteUnusedTensors(
-    const Scope& scope, const OperatorBase* op, GarbageCollector<Tensor>* gc,
+    const Scope& scope, const OperatorBase* op, GarbageCollector* gc,
     std::unordered_map<std::string, size_t>* ref_cnts) {
-  std::unordered_set<Tensor*> erase_tensors;
+  std::deque<std::shared_ptr<memory::Allocation>> garbages;
 
   auto handler = [&](const VariableNameMap& name_map) {
     for (auto& name_pair : name_map) {
       for (auto& name : name_pair.second) {
         auto it = ref_cnts->find(name);
         if (it == ref_cnts->end()) continue;
-        if (--(it->second) == 0) {
+        if (--(it->second) != 0) {
+          continue;
+        }
         auto* var = scope.FindVar(name);
         if (var != nullptr) {
-            VLOG(2) << "Erase tensor \'" << name << "\'";
+          continue;
+        }
+
+        VLOG(2) << "Erase variable " << name;
         if (var->IsType<LoDTensor>()) {
-              erase_tensors.insert(var->GetMutable<LoDTensor>());
+          garbages.emplace_back(var->GetMutable<LoDTensor>()->MoveMemory());
         } else if (var->IsType<SelectedRows>()) {
-              erase_tensors.insert(
-                  var->GetMutable<SelectedRows>()->mutable_value());
+          garbages.emplace_back(
+              var->GetMutable<SelectedRows>()->mutable_value()->MoveMemory());
         } else if (var->IsType<LoDTensorArray>()) {
           auto* lod_tensor_arr = var->GetMutable<LoDTensorArray>();
           for (auto& t : *lod_tensor_arr) {
-                erase_tensors.insert(&t);
-              }
-            }
+            garbages.emplace_back(t.MoveMemory());
           }
+        } else {
+          PADDLE_THROW("Type %s of %s is not supported eager deletion",
+                       var->Type().name(), name);
         }
       }
     }
@@ -116,8 +123,8 @@ static void DeleteUnusedTensors(
   handler(op->Inputs());
   handler(op->Outputs());
 
-  if (!erase_tensors.empty()) {
-    gc->Add(erase_tensors);
+  if (!garbages.empty()) {
+    gc->Add(std::move(garbages));
   }
 }
 
@@ -411,22 +418,22 @@ void Executor::RunPreparedContext(ExecutorPrepareContext* ctx, Scope* scope,
   }
 
   int64_t max_memory_size = GetEagerDeletionThreshold();
-  std::unique_ptr<GarbageCollector<Tensor>> gc;
+  std::unique_ptr<GarbageCollector> gc;
   if (max_memory_size >= 0) {
     ctx->ResetReferenceCount();
 #ifdef PADDLE_WITH_CUDA
     if (platform::is_gpu_place(place_)) {
       if (IsFastEagerDeletionModeEnabled()) {
-        gc.reset(new UnsafeFastGPUGarbageCollector<Tensor>(
+        gc.reset(new UnsafeFastGPUGarbageCollector(
             boost::get<platform::CUDAPlace>(place_), max_memory_size));
       } else {
-        gc.reset(new DefaultStreamGarbageCollector<Tensor>(
+        gc.reset(new DefaultStreamGarbageCollector(
             boost::get<platform::CUDAPlace>(place_), max_memory_size));
       }
     } else if (platform::is_cpu_place(place_)) {
 #endif
-      gc.reset(new CPUGarbageCollector<Tensor>(
-          boost::get<platform::CPUPlace>(place_), max_memory_size));
+      gc.reset(new CPUGarbageCollector(boost::get<platform::CPUPlace>(place_),
+                                       max_memory_size));
 #ifdef PADDLE_WITH_CUDA
     }
 #endif
@@ -442,7 +449,6 @@ void Executor::RunPreparedContext(ExecutorPrepareContext* ctx, Scope* scope,
   }
 
   platform::DeviceContextPool::Instance().Get(place_)->Wait();
-  if (gc) gc->Wait();
 
   if (local_scope != scope) {
     scope->DeleteScope(local_scope);

paddle/fluid/framework/garbage_collector.cc

+// 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.
+
+#include <algorithm>
+#ifdef PADDLE_WITH_CUDA
+#include "paddle/fluid/platform/cuda_device_guard.h"
+#endif
+#include "paddle/fluid/framework/garbage_collector.h"
+
+namespace paddle {
+namespace framework {
+
+GarbageCollector::GarbageCollector(const platform::Place &place,
+                                   size_t max_memory_size)
+    : max_memory_size_((std::max)(max_memory_size, static_cast<size_t>(1))) {
+  garbages_.reset(new GarbageQueue());
+  dev_ctx_ = platform::DeviceContextPool::Instance().Get(place);
+}
+
+CPUGarbageCollector::CPUGarbageCollector(const platform::CPUPlace &place,
+                                         size_t max_memory_size)
+    : GarbageCollector(place, max_memory_size) {}
+
+void CPUGarbageCollector::ClearCallback(const std::function<void()> &callback) {
+  callback();
+}
+
+#ifdef PADDLE_WITH_CUDA
+UnsafeFastGPUGarbageCollector::UnsafeFastGPUGarbageCollector(
+    const platform::CUDAPlace &place, size_t max_memory_size)
+    : GarbageCollector(place, max_memory_size) {}
+
+void UnsafeFastGPUGarbageCollector::ClearCallback(
+    const std::function<void()> &callback) {
+  callback();
+}
+
+DefaultStreamGarbageCollector::DefaultStreamGarbageCollector(
+    const platform::CUDAPlace &place, size_t max_memory_size)
+    : GarbageCollector(place, max_memory_size) {}
+
+void DefaultStreamGarbageCollector::Wait() const {
+  static_cast<platform::CUDADeviceContext *>(this->dev_ctx_)
+      ->WaitStreamCallback();
+}
+
+void DefaultStreamGarbageCollector::ClearCallback(
+    const std::function<void()> &callback) {
+  static_cast<platform::CUDADeviceContext *>(this->dev_ctx_)
+      ->AddStreamCallback(callback);
+}
+
+StreamGarbageCollector::StreamGarbageCollector(const platform::CUDAPlace &place,
+                                               size_t max_memory_size)
+    : GarbageCollector(place, max_memory_size) {
+  platform::CUDADeviceGuard guard(place.device);
+  PADDLE_ENFORCE(cudaStreamCreate(&stream_));
+  callback_manager_.reset(new platform::StreamCallbackManager(stream_));
+}
+
+StreamGarbageCollector::~StreamGarbageCollector() {
+  auto place = boost::get<platform::CUDAPlace>(this->dev_ctx_->GetPlace());
+  platform::CUDADeviceGuard guard(place.device);
+  PADDLE_ENFORCE(cudaStreamSynchronize(stream_));
+  PADDLE_ENFORCE(cudaStreamDestroy(stream_));
+}
+
+cudaStream_t StreamGarbageCollector::stream() const { return stream_; }
+
+void StreamGarbageCollector::Wait() const { callback_manager_->Wait(); }
+
+void StreamGarbageCollector::ClearCallback(
+    const std::function<void()> &callback) {
+  callback_manager_->AddCallback(callback);
+}
+#endif
+}  // namespace framework
+}  // namespace paddle

paddle/fluid/framework/garbage_collector.h

@@ -14,160 +14,83 @@
 
 #pragma once
 
-#include <algorithm>
 #include <deque>
 #include <functional>
 #include <memory>
 #include <mutex>  // NOLINT
-#ifdef PADDLE_WITH_CUDA
-#include "paddle/fluid/platform/cuda_device_guard.h"
-#endif
 #include "paddle/fluid/platform/device_context.h"
 
 namespace paddle {
 namespace framework {
 
-// T should have memory_size() and clear() method
-template <typename T>
 class GarbageCollector {
  public:
-  GarbageCollector(const platform::Place &place, size_t max_memory_size)
-      : max_memory_size_((std::max)(max_memory_size, static_cast<size_t>(1))) {
-    garbages_.reset(new std::deque<T *>());
-    dev_ctx_ = platform::DeviceContextPool::Instance().Get(place);
-  }
+  using GarbageQueue = std::deque<std::shared_ptr<memory::Allocation>>;
 
-  virtual ~GarbageCollector() {}
+  GarbageCollector(const platform::Place &place, size_t max_memory_size);
 
-  size_t NumOfGarbages() const {
-    std::lock_guard<std::mutex> guard(mutex_);
-    return garbages_->size();
-  }
+  virtual ~GarbageCollector() = default;
 
-  void Reset() {
-    std::lock_guard<std::mutex> guard(mutex_);
-    garbages_.reset(new std::deque<T *>());
-    cur_memory_size_ = 0;
-  }
+  virtual void Wait() const {}
 
   template <typename Container>
-  void Add(const Container &objs) {
-    Add(objs, []() {});
-  }
+  void Add(Container &&objs);
 
   template <typename Container, typename Callback>
-  void Add(const Container &objs, Callback &&callback) {
-    std::deque<T *> *clear_deque = nullptr;
-    {
-      std::lock_guard<std::mutex> guard(mutex_);
-      for (auto *obj : objs) {
-        garbages_->push_back(obj);
-        cur_memory_size_ += obj->memory_size();
-      }
-      if (cur_memory_size_ >= max_memory_size_) {
-        cur_memory_size_ = 0;
-        clear_deque = garbages_.release();
-        garbages_.reset(new std::deque<T *>());
-      }
-    }
-
-    if (clear_deque != nullptr) {
-      callback();
-      ClearCallback([clear_deque]() {
-        for (auto *obj : *clear_deque) obj->clear();
-        delete clear_deque;
-      });
-    }
-  }
-
-  virtual void Wait() const {}
+  void Add(Container &&objs, Callback &&callback);
 
  protected:
   virtual void ClearCallback(const std::function<void()> &callback) = 0;
 
   platform::DeviceContext *dev_ctx_;
-  std::unique_ptr<std::deque<T *>> garbages_;
+  std::unique_ptr<GarbageQueue> garbages_;
   mutable std::mutex mutex_;
   const size_t max_memory_size_;
-  size_t cur_memory_size_ = 0;
+  size_t cur_memory_size_{0};
 };
 
-template <typename T>
-class CPUGarbageCollector : public GarbageCollector<T> {
+class CPUGarbageCollector : public GarbageCollector {
  public:
-  CPUGarbageCollector(const platform::CPUPlace &place, size_t max_memory_size)
-      : GarbageCollector<T>(place, max_memory_size) {}
+  CPUGarbageCollector(const platform::CPUPlace &place, size_t max_memory_size);
 
  protected:
-  void ClearCallback(const std::function<void()> &callback) override {
-    callback();
-  }
+  void ClearCallback(const std::function<void()> &callback) override;
 };
 
 #ifdef PADDLE_WITH_CUDA
-template <typename T>
-class UnsafeFastGPUGarbageCollector : public GarbageCollector<T> {
+class UnsafeFastGPUGarbageCollector : public GarbageCollector {
  public:
   UnsafeFastGPUGarbageCollector(const platform::CUDAPlace &place,
-                                size_t max_memory_size)
-      : GarbageCollector<T>(place, max_memory_size) {}
+                                size_t max_memory_size);
 
  protected:
-  void ClearCallback(const std::function<void()> &callback) override {
-    callback();
-  }
+  void ClearCallback(const std::function<void()> &callback) override;
 };
 
-template <typename T>
-class DefaultStreamGarbageCollector : public GarbageCollector<T> {
+class DefaultStreamGarbageCollector : public GarbageCollector {
  public:
   DefaultStreamGarbageCollector(const platform::CUDAPlace &place,
-                                size_t max_memory_size)
-      : GarbageCollector<T>(place, max_memory_size) {}
-
-  cudaStream_t stream() const {
-    return static_cast<const platform::CUDADeviceContext *>(this->dev_ctx_)
-        ->stream();
-  }
+                                size_t max_memory_size);
 
-  void Wait() const override {
-    static_cast<platform::CUDADeviceContext *>(this->dev_ctx_)
-        ->WaitStreamCallback();
-  }
+  void Wait() const override;
 
  protected:
-  void ClearCallback(const std::function<void()> &callback) override {
-    static_cast<platform::CUDADeviceContext *>(this->dev_ctx_)
-        ->AddStreamCallback(callback);
-  }
+  void ClearCallback(const std::function<void()> &callback) override;
 };
 
-template <typename T>
-class StreamGarbageCollector : public GarbageCollector<T> {
+class StreamGarbageCollector : public GarbageCollector {
  public:
   StreamGarbageCollector(const platform::CUDAPlace &place,
-                         size_t max_memory_size)
-      : GarbageCollector<T>(place, max_memory_size) {
-    platform::CUDADeviceGuard guard(place.device);
-    PADDLE_ENFORCE(cudaStreamCreate(&stream_));
-    callback_manager_.reset(new platform::StreamCallbackManager(stream_));
-  }
+                         size_t max_memory_size);
 
-  ~StreamGarbageCollector() {
-    auto place = boost::get<platform::CUDAPlace>(this->dev_ctx_->GetPlace());
-    platform::CUDADeviceGuard guard(place.device);
-    PADDLE_ENFORCE(cudaStreamSynchronize(stream_));
-    PADDLE_ENFORCE(cudaStreamDestroy(stream_));
-  }
+  ~StreamGarbageCollector();
 
-  void Wait() const override { callback_manager_->Wait(); }
+  void Wait() const override;
 
-  cudaStream_t stream() const { return stream_; }
+  cudaStream_t stream() const;
 
  protected:
-  void ClearCallback(const std::function<void()> &callback) override {
-    callback_manager_->AddCallback(callback);
-  }
+  void ClearCallback(const std::function<void()> &callback) override;
 
  private:
   cudaStream_t stream_;
@@ -175,5 +98,33 @@ class StreamGarbageCollector : public GarbageCollector<T> {
 };
 #endif
 
+template <typename Container>
+void GarbageCollector::Add(Container &&objs) {
+  Add(std::forward<Container>(objs), []() {});
+}
+
+template <typename Container, typename Callback>
+void GarbageCollector::Add(Container &&objs, Callback &&callback) {
+  GarbageQueue *garbage_queue = nullptr;
+  {
+    std::lock_guard<std::mutex> guard(mutex_);
+    for (auto &obj : objs) {
+      if (!obj) continue;
+      cur_memory_size_ += obj->size();
+      garbages_->push_back(std::move(obj));
+    }
+    if (cur_memory_size_ >= max_memory_size_) {
+      cur_memory_size_ = 0;
+      garbage_queue = garbages_.release();
+      garbages_.reset(new GarbageQueue());
+    }
+  }
+
+  if (garbage_queue) {
+    callback();
+    ClearCallback([garbage_queue]() { delete garbage_queue; });
+  }
+}
+
 }  // namespace framework
 }  // namespace paddle

paddle/fluid/framework/parallel_executor.cc

@@ -97,29 +97,31 @@ std::unique_ptr<ir::Graph> ParallelExecutorPrivate::PrepareGCAndRefCnts(
     if (gcs_.count(place) > 0) {
       continue;
     }
-    GarbageCollector<Tensor> *gc = nullptr;
+    std::unique_ptr<GarbageCollector> gc;
 #ifdef PADDLE_WITH_CUDA
     if (platform::is_gpu_place(place)) {
       if (IsFastEagerDeletionModeEnabled()) {
-        gc = new UnsafeFastGPUGarbageCollector<Tensor>(
-            boost::get<platform::CUDAPlace>(place), max_memory_size);
+        gc.reset(new UnsafeFastGPUGarbageCollector(
+            boost::get<platform::CUDAPlace>(place), max_memory_size));
       } else {
-        gc = new StreamGarbageCollector<Tensor>(
-            boost::get<platform::CUDAPlace>(place), max_memory_size);
+        gc.reset(new StreamGarbageCollector(
+            boost::get<platform::CUDAPlace>(place), max_memory_size));
       }
       VLOG(10) << "Created " << i << "-th GarbageCollector at " << place;
-    } else if (platform::is_cpu_place(place)) {
+    } else {
 #endif
-      gc = new CPUGarbageCollector<Tensor>(
-          boost::get<platform::CPUPlace>(place), max_memory_size);
+      if (platform::is_cpu_place(place)) {
+        gc.reset(new CPUGarbageCollector(boost::get<platform::CPUPlace>(place),
+                                         max_memory_size));
         VLOG(10) << "Created GarbageCollector at " << place;
+      } else {
+        PADDLE_THROW("Unsupported place for garbage collection");
+      }
 #ifdef PADDLE_WITH_CUDA
     }
 #endif
 
-    if (gc) {
-      gcs_[place] = std::unique_ptr<GarbageCollector<Tensor>>(gc);
-    }
+    gcs_.emplace(place, std::move(gc));
   }
 
   if (!gcs_.empty()) {
@@ -144,8 +146,6 @@ std::unique_ptr<ir::Graph> ParallelExecutorPrivate::PrepareGCAndRefCnts(
     eager_deletion_pass->SetNotOwned(details::kAllPlaces, &places_);
     graph = eager_deletion_pass->Apply(std::move(graph));
     VLOG(10) << "EagerDeletionPass Applied";
-
-    graph->SetNotOwned(details::kGarbageCollector, &gcs_);
   }
 
   return graph;

paddle/fluid/framework/scope.cc

@@ -38,7 +38,7 @@ DEFINE_double(
     "Memory size threshold (GB) when the garbage collector clear tensors."
     "Disabled when this value is less than 0");
 
-DEFINE_bool(fast_eager_deletion_mode, true,
+DEFINE_bool(fast_eager_deletion_mode, false,
             "Fast eager deletion mode. If enabled, memory would release "
             "immediately without waiting GPU kernel ends.");
 

paddle/fluid/framework/tensor.h

@@ -158,6 +158,10 @@ class Tensor {
   const std::shared_ptr<memory::Allocation>& Holder() const { return holder_; }
   size_t offset() const { return offset_; }
 
+  std::shared_ptr<memory::Allocation> MoveMemory() {
+    return std::move(holder_);
+  }
+
  private:
   /*! holds the memory block if allocated. */
   std::shared_ptr<memory::Allocation> holder_;

python/paddle/fluid/tests/unittests/test_eager_deletion_gru_net.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.
+
+import unittest
+from test_eager_deletion_lstm_net import TestBase
+import paddle.fluid as fluid
+
+
+def gru_net(data,
+            label,
+            dict_dim,
+            emb_dim=128,
+            hid_dim=128,
+            hid_dim2=96,
+            class_dim=2,
+            emb_lr=400.0):
+    emb = fluid.layers.embedding(
+        input=data,
+        size=[dict_dim, emb_dim],
+        param_attr=fluid.ParamAttr(learning_rate=emb_lr))
+    fc0 = fluid.layers.fc(input=emb, size=hid_dim * 3)
+    gru_h = fluid.layers.dynamic_gru(input=fc0, size=hid_dim, is_reverse=False)
+    gru_max = fluid.layers.sequence_pool(input=gru_h, pool_type='max')
+    gru_max_tanh = fluid.layers.tanh(gru_max)
+    fc1 = fluid.layers.fc(input=gru_max_tanh, size=hid_dim2, act='tanh')
+    prediction = fluid.layers.fc(input=fc1, size=class_dim, act='softmax')
+    cost = fluid.layers.cross_entropy(input=prediction, label=label)
+    avg_cost = fluid.layers.mean(x=cost)
+    return avg_cost
+
+
+class GRUTest(TestBase):
+    def setUp(self):
+        self.net = gru_net
+
+
+if __name__ == "__main__":
+    unittest.main()

python/paddle/fluid/tests/unittests/test_eager_deletion_lstm_net.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.
+
+import os
+os.environ['FLAGS_eager_delete_tensor_gb'] = '0.0'
+os.environ['CPU_NUM'] = '2'
+
+import six
+import unittest
+
+import paddle
+import paddle.fluid.core as core
+import paddle.fluid as fluid
+
+
+def train(network, use_cuda, use_parallel_executor, batch_size=32, pass_num=2):
+    if use_cuda and not core.is_compiled_with_cuda():
+        print('Skip use_cuda=True because Paddle is not compiled with cuda')
+        return
+
+    word_dict = paddle.dataset.imdb.word_dict()
+    train_reader = paddle.batch(
+        paddle.dataset.imdb.train(word_dict), batch_size=batch_size)
+
+    data = fluid.layers.data(
+        name="words", shape=[1], dtype="int64", lod_level=1)
+
+    label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+
+    cost = network(data, label, len(word_dict))
+    optimizer = fluid.optimizer.Adagrad(learning_rate=0.2)
+    optimizer.minimize(cost)
+
+    place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
+    feeder = fluid.DataFeeder(feed_list=[data, label], place=place)
+    reader = feeder.decorate_reader(
+        train_reader, multi_devices=use_parallel_executor)
+
+    exe = fluid.Executor(place)
+    exe.run(fluid.default_startup_program())
+
+    if use_parallel_executor:
+        train_exe = fluid.ParallelExecutor(
+            use_cuda=use_cuda, loss_name=cost.name)
+        fetch_list = [cost.name]
+    else:
+        train_exe = exe
+        fetch_list = [cost]
+
+    for pass_id in six.moves.xrange(pass_num):
+        batch_id = 0
+        for data in reader():
+            train_exe.run(feed=data,
+                          fetch_list=fetch_list if batch_id % 4 == 0 else [])
+            batch_id += 1
+            if batch_id > 16:
+                break
+
+
+def lstm_net(data,
+             label,
+             dict_dim,
+             emb_dim=128,
+             hid_dim=128,
+             hid_dim2=96,
+             class_dim=2,
+             emb_lr=30.0):
+    emb = fluid.layers.embedding(
+        input=data,
+        size=[dict_dim, emb_dim],
+        param_attr=fluid.ParamAttr(learning_rate=emb_lr))
+    fc0 = fluid.layers.fc(input=emb, size=hid_dim * 4)
+    lstm_h, c = fluid.layers.dynamic_lstm(
+        input=fc0, size=hid_dim * 4, is_reverse=False)
+    lstm_max = fluid.layers.sequence_pool(input=lstm_h, pool_type='max')
+    lstm_max_tanh = fluid.layers.tanh(lstm_max)
+    fc1 = fluid.layers.fc(input=lstm_max_tanh, size=hid_dim2, act='tanh')
+    prediction = fluid.layers.fc(input=fc1, size=class_dim, act='softmax')
+    cost = fluid.layers.cross_entropy(input=prediction, label=label)
+    avg_cost = fluid.layers.mean(x=cost)
+    return avg_cost
+
+
+class TestBase(unittest.TestCase):
+    def setUp(self):
+        self.net = lstm_net
+
+    def test_network(self):
+        for use_cuda in [True, False]:
+            for use_parallel_executor in [False, True]:
+                print('network: {}, use_cuda: {}, use_parallel_executor: {}'.
+                      format(self.net.__name__, use_cuda,
+                             use_parallel_executor))
+                with fluid.program_guard(fluid.Program(), fluid.Program()):
+                    with fluid.scope_guard(core.Scope()):
+                        train(self.net, use_cuda, use_parallel_executor)
+
+
+if __name__ == "__main__":
+    unittest.main()