add mir implementation

paddle/fluid/lite/core/executor.h

@@ -52,7 +52,7 @@ class Executor {
       ops_.emplace_back(LiteOpRegistry::Global().Create(op_type));
       // pick initial kernel
       ops_.back()->PickKernel(valid_places_);
-      ops_.back()->Attach(*op_desc, exec_scope_);
+      ops_.back()->AttachImpl(*op_desc, exec_scope_);
     }
   }
 

paddle/fluid/lite/core/kernel.cc

@@ -17,11 +17,6 @@
 namespace paddle {
 namespace lite {
 
-bool operator==(const Place &a, const Place &b) {
-  return a.target == b.target && a.precision == b.precision &&
-         a.layout == b.layout;
-}
-
 bool operator<(const Place &a, const Place &b) {
   if (a.target != b.target)
     return a.target < b.target;

paddle/fluid/lite/core/kernel.h

@@ -19,6 +19,7 @@
 #include "paddle/fluid/framework/op_desc.h"
 #include "paddle/fluid/lite/core/context.h"
 #include "paddle/fluid/lite/core/target_wrapper.h"
+#include "paddle/fluid/lite/core/type_system.h"
 #include "paddle/fluid/lite/core/types.h"
 #include "paddle/fluid/lite/operators/op_params.h"
 #include "paddle/fluid/lite/utils/all.h"
@@ -51,6 +52,7 @@ class KernelBase {
 
   virtual TargetType target() const = 0;
   virtual PrecisionType precision() const = 0;
+  virtual DataLayoutType layout() const = 0;
 
   virtual ~KernelBase() = default;
 
@@ -66,17 +68,21 @@ class KernelBase {
  * registered in the `TypeSystem`.
  */
 struct ParamType {
+  // For unsupported types.
   size_t element_type_hash{};
   Place tensor_place{};
+  const Type* type_;
 
   ParamType() = default;
   ParamType(size_t element_type_hash) : element_type_hash(element_type_hash) {}
   ParamType(size_t element_type_hash, const Place& place)
       : element_type_hash(element_type_hash), tensor_place(place) {}
+  ParamType(const Type* type) : type_(type) {}
 };
 
 /*
- * The data types of kernel parameters.
+ * The data types of kernel parameters. It is used to track the type of kernel's
+ * inputs and outputs.
  */
 struct ParamTypes {
   std::vector<std::vector<ParamType>> inputs;
@@ -115,6 +121,8 @@ struct ParamTypes {
  */
 class ParamTypeRegistry {
  public:
+  enum class IO : int { kInput = 0, kOutput };
+
   template <TargetType target, PrecisionType precision,
             DataLayoutType layout = DataLayoutType::kNCHW>
   /*
@@ -130,7 +138,12 @@ class ParamTypeRegistry {
     NewInstance(const std::string& kernel_type) : kernel_type_(kernel_type) {}
 
     NewInstance& BindInput(int offset, const ParamType& ptype) {
-      ParamTypeRegistry::Global().Register(
+      ParamTypeRegistry::Global().Register<IO::kInput>(
+          kernel_type_, Place{target, precision, layout}, offset, ptype);
+      return *this;
+    }
+    NewInstance& BindOutput(int offset, const ParamType& ptype) {
+      ParamTypeRegistry::Global().Register<IO::kOutput>(
           kernel_type_, Place{target, precision, layout}, offset, ptype);
       return *this;
     }
@@ -141,8 +154,12 @@ class ParamTypeRegistry {
     std::string kernel_type_;
   };
 
+  template <IO io>
   void Register(const std::string& kernel_type, const Place& place, int offset,
-                ParamType data_type) {}
+                ParamType data_type) {
+    KernelIdTy key{kernel_type, place, io, offset};
+    types_[key] = data_type;
+  }
 
   ParamType Retrive(const Place& place, int offset);
 
@@ -155,16 +172,15 @@ class ParamTypeRegistry {
   ParamTypeRegistry() = default;
 
  public:
-  enum class IO : int { kInput = 0, kOutput };
   // Identification for a Kernel.
-  struct KernelIdT {
+  struct KernelIdTy {
     std::string kernel_type;
     Place place;
     IO io;
     int offset;
   };
 
-  using key_t = KernelIdT;
+  using key_t = KernelIdTy;
   struct KeyCmp {
     bool operator()(const key_t& a, const key_t& b) const;
   };
@@ -188,6 +204,7 @@ class OpKernel : public KernelBase {
 
   TargetType target() const override { return Target; }
   PrecisionType precision() const override { return Precision; }
+  DataLayoutType layout() const override { return DataLayout; }
 
   void Touch() {}
 

paddle/fluid/lite/core/mir/CMakeLists.txt

-cc_library(mir_pass SRCS pass.cc)
 cc_library(mir_node SRCS node.cc)
-cc_library(mir_ssa_graph SRCS ssa_graph.cc)
+cc_library(mir_ssa_graph SRCS ssa_graph.cc DEPS mir_node)
\ No newline at end of file
+cc_library(mir_pass SRCS pass.cc DEPS mir_ssa_graph)
+cc_library(mir_pass_manager SRCS pass_manager.cc DEPS mir_pass mir_ssa_graph)
+cc_library(mir_pass_registry SRCS pass_registry.cc DEPS mir_pass_manager)
+cc_library(mir_demo_pass SRCS demo_pass.cc DEPS mir_pass)
+
+cc_test(test_mir_pass_manager SRCS pass_manager_test.cc DEPS mir_pass_manager mir_demo_pass)

paddle/fluid/lite/core/mir/demo_pass.cc

+// Copyright (c) 2019 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/lite/core/mir/pass.h"
+#include "paddle/fluid/lite/core/mir/pass_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class DemoPass : public mir::Pass {
+ public:
+  void Apply(std::unique_ptr<mir::SSAGraph>& graph) override {}
+};
+
+bool RegisterDemoPass() {
+  return PassManager::Global().AddNewPass("demo", new DemoPass);
+}
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/node.cc

+// Copyright (c) 2019 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/lite/core/mir/node.h"

paddle/fluid/lite/core/mir/node.h

+// Copyright (c) 2019 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.
+
+#pragma once
+
+#include <list>
+#include <memory>
+#include <string>
+#include <vector>
+#include "paddle/fluid/lite/core/kernel.h"
+
 namespace paddle {
 namespace lite {
 namespace mir {
 
+// Node in a MIR graph.
 class Node {
  public:
-  // Tell is instruction.
-  bool IsInstruct() const;
-  // Tell is an argument.
-  bool IsArgument() const;
-};
+  std::list<Node*> inlinks;
+  std::list<Node*> outlinks;
+
+  Node() = default;
+
+  enum class Role {
+    kUnk = -1,
+    kArgument,
+    kInstruct,
+    kNumRoles /*should be last*/
+  };
 
+  struct Instruct {
+    std::string op_type;
+    Place place;
+    // The kernel instances this Instruct contains.
+    std::vector<std::unique_ptr<KernelBase>> valid_kernels;
+  };
+
+  struct Argument {
+    std::string name;
+    Place place;
+  };
+
+  // Set roles.
+  Argument& AsArgument() {
+    if (role_ != Role::kUnk) {
+      CHECK(role_ == Role::kArgument);
+      return *argument_;
+    }
+    role_ = Role::kArgument;
+    argument_.reset(new Argument);
+    return *argument_;
+  }
+  Instruct& AsInstruct() {
+    if (role_ != Role::kUnk) {
+      CHECK(role_ == Role::kInstruct);
+      return *instruct_;
+    }
+    role_ = Role::kInstruct;
+    instruct_.reset(new Instruct);
+    return *instruct_;
+  }
+  // Check roles.
+  bool IsRoleSet() const { return role_ == Role::kUnk; }
+  bool IsInstruct() const { return role_ == Role::kInstruct; }
+  bool IsArgument() const { return role_ == Role::kArgument; }
+
+ private:
+  // Either instruct_ or argument_ is used.
+  std::unique_ptr<Instruct> instruct_;
+  std::unique_ptr<Argument> argument_;
+
+  Role role_{Role::kUnk};
+};
 }  // namespace mir
 }  // namespace lite
-}  // namespace paddle
+}  // namespace paddle
\ No newline at end of file

paddle/fluid/lite/core/mir/pass.cc

+// Copyright (c) 2019 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/lite/core/mir/pass.h"

paddle/fluid/lite/core/mir/pass.h

+// Copyright (c) 2019 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.
+
+#pragma once
+#include "paddle/fluid/lite/core/mir/node.h"
+#include "paddle/fluid/lite/core/mir/ssa_graph.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class Pass {
+ public:
+  virtual void Apply(std::unique_ptr<mir::SSAGraph>& graph) = 0;
+
+  const std::string& name() const { return name_; }
+
+  virtual ~Pass() = default;
+
+ private:
+  std::string name_;
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/pass_manager.cc

+// Copyright (c) 2019 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/lite/core/mir/pass_manager.h"
+#include "paddle/fluid/lite/core/mir/pass_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+PassManager::PassManager() {}
+
+// Manually register here.
+extern bool RegisterDemoPass();
+static bool xx __attribute__((unused)) = RegisterDemoPass();
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/pass_manager.h

+// Copyright (c) 2019 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.
+
+#pragma once
+#include <list>
+#include <map>
+#include <string>
+
+#include "paddle/fluid/lite/core/mir/pass.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class PassManager {
+ public:
+  static PassManager& Global() {
+    static PassManager x;
+    return x;
+  }
+
+  PassManager();
+
+  void Run() {
+    for (auto& pass : passes_) {
+      LOG(INFO) << "Running MIR pass " << pass->name();
+      pass->Apply(graph_);
+    }
+  }
+
+  bool AddNewPass(const std::string& name, Pass* pass) {
+    passes_.emplace_back(pass);
+    pass_map_.emplace(name, passes_.back().get());
+    return true;
+  }
+
+  // Clear all the passes.
+  void Clear() { passes_.clear(); }
+
+  std::list<std::unique_ptr<mir::Pass>>::iterator passes_begin() {
+    return passes_.begin();
+  }
+  std::list<std::unique_ptr<mir::Pass>>::iterator passes_end() {
+    return passes_.end();
+  }
+  std::list<std::unique_ptr<mir::Pass>>::const_iterator passes_const_begin()
+      const {
+    return passes_.begin();
+  }
+  std::list<std::unique_ptr<mir::Pass>>::const_iterator passes_const_end()
+      const {
+    return passes_.end();
+  }
+
+  Pass* LookUp(const std::string& key) {
+    auto it = pass_map_.find(key);
+    CHECK(it != pass_map_.end());
+    return it->second;
+  }
+
+ private:
+  std::unique_ptr<mir::SSAGraph> graph_;
+  std::list<std::unique_ptr<mir::Pass>> passes_;
+  std::map<std::string, mir::Pass*> pass_map_;
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/pass_manager_test.cc

+// Copyright (c) 2019 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/lite/core/mir/pass_manager.h"
+#include <gtest/gtest.h>
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+TEST(PassManager, test) {
+  auto* pass = PassManager::Global().LookUp("demo");
+  LOG(INFO) << "pass: " << pass;
+  ASSERT_TRUE(pass != nullptr);
+}
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/pass_registry.cc

+// Copyright (c) 2019 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/lite/core/mir/pass_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/pass_registry.h

+// Copyright (c) 2019 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.
+
+#pragma once
+
+#include <string>
+#include "paddle/fluid/lite/core/mir/pass.h"
+#include "paddle/fluid/lite/core/mir/pass_manager.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class PassRegistry {
+ public:
+  PassRegistry(const std::string& name, mir::Pass* pass) {
+    LOG(INFO) << "Registry add MIR pass " << name;
+    PassManager::Global().AddNewPass(name, pass);
+  }
+
+  bool Touch() const { return true; }
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/mir/ssa_graph.cc

+// Copyright (c) 2019 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/lite/core/mir/ssa_graph.h"

paddle/fluid/lite/core/mir/ssa_graph.h

+// Copyright (c) 2019 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.
+
+#pragma once
+
+#include <list>
+#include <map>
+#include <string>
+#include <vector>
+#include "paddle/fluid/lite/core/mir/node.h"
+#include "paddle/fluid/lite/core/op_lite.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+// A program is used to represent a code program, in Paddle, a code program
+// contains:
+// - main block, which is a list of OpLite
+// - scope: which contains all the weights
+struct Program {
+  std::list<std::unique_ptr<OpLite>> ops;
+  lite::Scope *scope;
+};
+
+// An Graph for MIR. It is built from a list of Op and a scope.
+class GraphBase {};
+
+class SSAGraph : GraphBase {
+ public:
+  // @param program: the op program
+  // @param valid_places: the valid places user set for the system.
+  void Build(const Program &program, const std::vector<Place> &valid_places) {
+    for (auto &op : program.ops) {
+      node_storage_.emplace_back();
+      // TODO(Superjomn) remove one valid_places here.
+      op->SetValidPlaces(valid_places);
+      auto &new_node = node_storage_.back();
+      auto &new_kernel = node_storage_.back().AsInstruct();
+      new_kernel.valid_kernels = op->CreateKernels(valid_places);
+
+      CHECK(new_node.inlinks.empty()) << "duplicate Build found";
+      CHECK(new_node.outlinks.empty()) << "duplicate Build found";
+      // collect inputs and outputs
+      for (const std::string &name : op->input_names()) {
+        new_node.inlinks.push_back(arguments_.at(name));
+      }
+      for (const std::string &name : op->output_names()) {
+        new_node.outlinks.push_back(arguments_.at(name));
+      }
+    }
+  }
+
+  std::vector<mir::Node *> TopoloticalOrder() const;
+
+ private:
+  std::list<mir::Node> node_storage_;
+  std::map<std::string, mir::Node *> arguments_;
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/op_lite.cc

@@ -44,5 +44,15 @@ void OpLite::PickKernel(const std::vector<Place> &valid_places,
   }
 }
 
+bool OpLite::Run() {
+  CHECK(kernel_);
+  SyncInputEvents();
+
+  kernel_->Run();
+
+  RecordOutputEvents();
+  return true;
+}
+
 }  // namespace lite
 }  // namespace paddle

paddle/fluid/lite/core/op_lite.h

@@ -36,6 +36,11 @@ struct Registry {
   void Touch() {}
 };
 
+namespace mir {
+class Node;
+class SSAGraph;
+}
+
 /**
  * The base class of an light-weight operators, currently just used in inference
  * to eliminate overhead of some operations in current framework.
@@ -71,19 +76,13 @@ class OpLite : public Registry {
   // Inference the outputs' shape.
   virtual bool InferShape() const { return true; }
   // Run this operator.
-  virtual bool Run() {
-    CHECK(kernel_);
-    SyncInputEvents();
-
-    kernel_->Run();
+  virtual bool Run();
 
-    RecordOutputEvents();
-    return true;
+  bool Attach(const framework::OpDesc &opdesc, lite::Scope *scope) {
+    ExtractInputsAndOutputs(opdesc);
+    return AttachImpl(opdesc, scope);
   }
 
-  // Attach it with the runtime environment.
-  virtual bool Attach(const framework::OpDesc &opdesc, lite::Scope *scope) = 0;
-
   // Human-readable information.
   virtual std::string DebugString() const = 0;
 
@@ -92,9 +91,29 @@ class OpLite : public Registry {
   void PickKernel(const std::vector<Place> &valid_places,
                   KernelStrategy kernel_strategy = KernelStrategy::kStatic);
 
+  const std::list<std::string> &input_names() const { return input_names_; }
+  const std::list<std::string> &output_names() const { return output_names_; }
+
   virtual ~OpLite() = default;
 
  protected:
+  // Attach it with the runtime environment.
+  virtual bool AttachImpl(const framework::OpDesc &opdesc,
+                          lite::Scope *scope) = 0;
+
+  void ExtractInputsAndOutputs(const framework::OpDesc &opdesc) {
+    for (const auto &item : opdesc.Inputs()) {
+      for (const auto &x : item.second) {
+        input_names_.push_back(x);
+      }
+    }
+    for (const auto &item : opdesc.Outputs()) {
+      for (const auto &x : item.second) {
+        output_names_.push_back(x);
+      }
+    }
+  }
+
   // Specify the kernel to run by default. This will specify the value of
   // `kernel_place_`.
   virtual void StaticPickKernel(const std::vector<Place> &valid_targets) {
@@ -113,12 +132,17 @@ class OpLite : public Registry {
   std::vector<std::unique_ptr<KernelBase>> CreateKernels(
       const std::vector<Place> &places, const std::string &kernel_type = "");
 
+  friend class mir::Node;
+  friend class mir::SSAGraph;
+
  protected:
   std::unique_ptr<OpContext> op_context_;
   std::unique_ptr<KernelBase> kernel_;
   std::string op_type_;
   std::vector<Place> valid_places_;
   Place kernel_place_{TARGET(kHost), PRECISION(kFloat)};
+  std::list<std::string> input_names_;
+  std::list<std::string> output_names_;
 };
 
 }  // namespace lite

paddle/fluid/lite/core/type_system.cc

@@ -13,3 +13,46 @@
 // limitations under the License.
 
 #include "paddle/fluid/lite/core/type_system.h"
+
+namespace paddle {
+namespace lite {
+
+// ------------------------- GetType specification ----------------------------
+template <>
+const Type*
+Type::Get<false /*is_unsupported*/, false /*is_tensor*/, TargetType::kHost,
+          PrecisionType::kFloat, DataLayoutType::kNCHW>() {
+  static UnsupportedTy x;
+  return &x;
+}
+
+template <>
+const Type*
+Type::Get<false /*is_unsupported*/, true /*is_tensor*/, TargetType::kX86,
+          PrecisionType::kFloat, DataLayoutType::kNCHW>() {
+  static TensorFp32NCHWTy x(TargetType::kX86);
+  return &x;
+}
+
+template <>
+const Type* Type::Get<UnsupportedTy>(TargetType target, int device) {
+  return Get<false, false, TargetType::kHost, PrecisionType::kFloat,
+             DataLayoutType::kNCHW>();
+}
+
+template <>
+const Type* Type::Get<TensorFp32NCHWTy>(TargetType target) {
+  switch (target) {
+    case TargetType::kX86:
+      return Get<false, true, TargetType::kX86, PrecisionType::kFloat,
+                 DataLayoutType::kNCHW>();
+    default:
+      LOG(FATAL) << "unsupported target " << TargetToStr(target);
+      return nullptr;
+  }
+}
+
+// ------------------------- end GetType specification ------------------------
+
+}  // namespace lite
+}  // namespace paddle

paddle/fluid/lite/core/type_system.h

@@ -82,7 +82,7 @@ class DataTypeBase {
  * Datatype with device info considered.
  * NOTE A Type with different device is treated as different DeviceDataType.
  */
-class DeviceDataType : public DataTypeBase {
+class Type : public DataTypeBase {
  public:
   TargetType target() const { return place_.target; }
   PrecisionType precision() const { return place_.precision; }
@@ -90,23 +90,31 @@ class DeviceDataType : public DataTypeBase {
   const Place& place() const { return place_; }
   const std::string& name() const { return name_; }
 
-  bool operator==(const DeviceDataType& other) {
+  bool operator==(const Type& other) {
     return id_ == other.id() && place_ == other.place();
   }
 
   // Can cast to another type. This is heavily used in MIR, by determine whether
   // is is possible to add a instruction to transform a type to another.
-  virtual bool TypeCastable(const DeviceDataType& type) const {
-    return id_ == type.id();
-  }
+  virtual bool TypeCastable(const Type& type) const { return id_ == type.id(); }
+
+  template <bool is_unknown, bool is_tensor = true,
+            TargetType target = TargetType::kHost,
+            PrecisionType precision = PrecisionType::kFloat,
+            DataLayoutType layout = DataLayoutType::kNCHW>
+  // Get a type.
+  static const Type* Get();
+
+  template <typename TypeTy>
+  static const Type* Get(TargetType target = TargetType::kHost);
 
-  virtual ~DeviceDataType() = default;
+  virtual ~Type() = default;
 
  protected:
-  DeviceDataType(ID id, const std::string& name, bool is_tensor,
-                 TargetType target = TargetType::kHost,
-                 PrecisionType precision = PrecisionType::kFloat,
-                 DataLayoutType layout = DataLayoutType::kNCHW)
+  Type(ID id, const std::string& name, bool is_tensor,
+       TargetType target = TargetType::kHost,
+       PrecisionType precision = PrecisionType::kFloat,
+       DataLayoutType layout = DataLayoutType::kNCHW)
       : DataTypeBase(id, is_tensor),
         place_{target, precision, layout},
         name_(name) {}
@@ -117,30 +125,33 @@ class DeviceDataType : public DataTypeBase {
 };
 
 // -------------------------------- predefined types ---------------------------
-class Void : public DeviceDataType {
+// TODO(Superjomn) make all the Types' constructs protected to make sure there
+// is only one instance across the system.
+class VoidTy : public Type {
+ public:
+  VoidTy() : Type(ID::Void, "Void", false /*is_tensor*/) {}
+};
+class UnsupportedTy : public Type {
  public:
-  Void() : DeviceDataType(ID::Void, "Void", false /*is_tensor*/) {}
+  UnsupportedTy() : Type(ID::Unsupported, "Unsupported", false /*is_tensor*/) {}
 };
-class TensorFp32NCHW : public DeviceDataType {
+class TensorFp32NCHWTy : public Type {
  public:
-  TensorFp32NCHW(TargetType target)
-      : DeviceDataType(ID::Tensor_Fp32_NCHW, "TensorFp32NCHW",
-                       true /*is_tensor*/, target, PrecisionType::kFloat,
-                       DataLayoutType::kNCHW) {}
+  TensorFp32NCHWTy(TargetType target)
+      : Type(ID::Tensor_Fp32_NCHW, "TensorFp32NCHW", true /*is_tensor*/, target,
+             PrecisionType::kFloat, DataLayoutType::kNCHW) {}
 };
-class TensorInt8NCHW : public DeviceDataType {
+class TensorInt8NCHWTy : public Type {
  public:
-  TensorInt8NCHW(TargetType target)
-      : DeviceDataType(ID::Tensor_Int8_NCHW, "TensorInt8NCHW",
-                       true /*is_tensor*/, target, PrecisionType::kInt8,
-                       DataLayoutType::kNCHW) {}
+  TensorInt8NCHWTy(TargetType target)
+      : Type(ID::Tensor_Int8_NCHW, "TensorInt8NCHW", true /*is_tensor*/, target,
+             PrecisionType::kInt8, DataLayoutType::kNCHW) {}
 };
-class TensorInt64NCHW : public DeviceDataType {
+class TensorInt64NCHWTy : public Type {
  public:
-  TensorInt64NCHW(TargetType target)
-      : DeviceDataType(ID::Tensor_Int64_NCHW, "TensorInt64NCHW",
-                       true /*is_tensor*/, target, PrecisionType::kInt8,
-                       DataLayoutType::kNCHW) {}
+  TensorInt64NCHWTy(TargetType target)
+      : Type(ID::Tensor_Int64_NCHW, "TensorInt64NCHW", true /*is_tensor*/,
+             target, PrecisionType::kInt8, DataLayoutType::kNCHW) {}
 };
 // ------------------------- end predefined types ---------------------------
 

paddle/fluid/lite/kernels/host/fc_compute.cc

@@ -15,6 +15,7 @@
 #include "paddle/fluid/lite/kernels/host/fc_compute.h"
 #include <Eigen/Core>
 #include "paddle/fluid/lite/core/op_registry.h"
+#include "paddle/fluid/lite/core/type_system.h"
 
 namespace paddle {
 namespace lite {
@@ -51,6 +52,8 @@ PrecisionType FcCompute::precision() const { return PRECISION(kFloat); }
 }  // namespace paddle
 
 REGISTER_LITE_KERNEL(fc, kHost, kFloat, paddle::lite::kernels::host::FcCompute)
-    .BindInput(0, {typeid(paddle::lite::Tensor).hash_code(),
-                   paddle::lite::Place{TARGET(kHost), PRECISION(kFloat)}})
+    .BindInput(0, {paddle::lite::Type::Get<paddle::lite::TensorFp32NCHWTy>(
+                      TARGET(kX86))})
+    .BindOutput(0, {paddle::lite::Type::Get<paddle::lite::TensorFp32NCHWTy>(
+                       TARGET(kX86))})
     .Finalize();

paddle/fluid/lite/operators/fc_op.h

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#pragma once
+
 #include <string>
 #include <vector>
 #include "paddle/fluid/lite/core/kernel.h"
@@ -44,7 +46,8 @@ class FcOpLite : public OpLite {
    */
 
   // TODO(Superjomn) replace framework::OpDesc with a lite one.
-  bool Attach(const framework::OpDesc &op_desc, lite::Scope *scope) override {
+  bool AttachImpl(const framework::OpDesc &op_desc,
+                  lite::Scope *scope) override {
     auto input = op_desc.Input("Input").front();
     auto W = op_desc.Input("W").front();
     auto bias = op_desc.Input("Bias").front();

paddle/fluid/lite/operators/fc_op_test.cc

@@ -61,7 +61,7 @@ TEST(fc_op_lite, test) {
   fc.SetValidPlaces({Place{TARGET(kHost), PRECISION(kFloat)}});
   fc.PickKernel({Place{TARGET(kHost), PRECISION(kFloat)}});
 
-  fc.Attach(desc, &scope);
+  fc.AttachImpl(desc, &scope);
   fc.Run();
 
   for (int i = 0; i < 10 * 20; i++) {

paddle/fluid/lite/operators/mul_op.h

@@ -37,7 +37,8 @@ class MulOpLite : public OpLite {
   bool InferShape() const override;
 
   // TODO(Superjomn) replace framework::OpDesc with a lite one.
-  bool Attach(const framework::OpDesc &op_desc, lite::Scope *scope) override {
+  bool AttachImpl(const framework::OpDesc &op_desc,
+                  lite::Scope *scope) override {
     auto input = op_desc.Input("X").front();
     auto W = op_desc.Input("Y").front();
     auto out = op_desc.Output("Out").front();

paddle/fluid/lite/operators/relu_op.cc

@@ -31,7 +31,7 @@ bool ReluOp::InferShape() const {
   return true;
 }
 
-bool ReluOp::Attach(const framework::OpDesc &opdesc, lite::Scope *scope) {
+bool ReluOp::AttachImpl(const framework::OpDesc &opdesc, lite::Scope *scope) {
   param_.input = const_cast<Tensor *>(
       &scope->FindVar(opdesc.Input("Input").front())->Get<Tensor>());
   param_.output =

paddle/fluid/lite/operators/relu_op.h

@@ -32,7 +32,7 @@ class ReluOp : public OpLite {
 
   bool InferShape() const override;
 
-  bool Attach(const framework::OpDesc &opdesc, lite::Scope *scope) override;
+  bool AttachImpl(const framework::OpDesc &opdesc, lite::Scope *scope) override;
 
   std::string DebugString() const override { return "tanh"; }
 

paddle/fluid/lite/operators/scale_op.cc

@@ -44,7 +44,8 @@ class ScaleOp : public OpLite {
   }
 
   // TODO(Superjomn) replace framework::OpDesc with a lite one.
-  bool Attach(const framework::OpDesc &op_desc, lite::Scope *scope) override {
+  bool AttachImpl(const framework::OpDesc &op_desc,
+                  lite::Scope *scope) override {
     auto x = op_desc.Input("X").front();
     auto out = op_desc.Output("Out").front();