[pass][OpenCL]add fuse flatten_contiguous_range and fc pass (#6040) (#6057)

daming5432 · web-flow · commit 7da838bae0ad · 2021-05-12T15:39:38.000+08:00
* add fuse flatten_contiguous_range and fc pass

* modify Copyright 2021 test=develop
diff --git a/lite/api/paddle_use_passes.h b/lite/api/paddle_use_passes.h
@@ -78,6 +78,7 @@ USE_MIR_PASS(control_flow_op_shared_inputs_and_outputs_place_sync_pass);
 USE_MIR_PASS(lite_scale_activation_fuse_pass);
 USE_MIR_PASS(lite_instance_norm_activation_fuse_pass);
 USE_MIR_PASS(ssd_boxes_calc_offline_pass);
+USE_MIR_PASS(lite_flatten_fc_fuse_pass);
 USE_MIR_PASS(lite_fc_prelu_fuse_pass);
 USE_MIR_PASS(__xpu__graph_dedup_pass);
 USE_MIR_PASS(__xpu__resnet_fuse_pass);
diff --git a/lite/core/mir/CMakeLists.txt b/lite/core/mir/CMakeLists.txt
@@ -58,6 +58,7 @@ lite_cc_library(mir_passes
       fusion/sequence_reverse_embedding_fuse_pass.cc
       fusion/instance_norm_activation_fuse_pass.cc
       fusion/elementwise_add_scale_fuse_pass.cc
+      fusion/flatten_fc_fuse_pass.cc
       fusion/fc_prelu_fuse_pass.cc
       elimination/identity_scale_eliminate_pass.cc
       elimination/identity_dropout_eliminate_pass.cc
diff --git a/lite/core/mir/fusion/CMakeLists.txt b/lite/core/mir/fusion/CMakeLists.txt
@@ -70,6 +70,9 @@ lite_cc_library(fuse_instance_norm_activation
 lite_cc_library(fuse_elementwise_add_scale
         SRCS elementwise_add_scale_fuser.cc
         DEPS pattern_matcher_high_api)
+lite_cc_library(fuse_flatten_fc
+        SRCS flatten_fc_fuser.cc
+        DEPS pattern_matcher_high_api)
 lite_cc_library(fuse_fc_prelu
         SRCS fc_prelu_fuser.cc
         DEPS pattern_matcher_high_api)
@@ -98,6 +101,7 @@ set(mir_fusers
     fuse_sequence_reverse_embedding
     fuse_instance_norm_activation
     fuse_elementwise_add_scale
+    fuse_flatten_fc
     fuse_fc_prelu
     fuse_conv_scale
     CACHE INTERNAL "fusers")
diff --git a/lite/core/mir/fusion/flatten_fc_fuse_pass.cc b/lite/core/mir/fusion/flatten_fc_fuse_pass.cc
@@ -0,0 +1,37 @@
+// Copyright (c) 2021 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 "lite/core/mir/fusion/flatten_fc_fuse_pass.h"
+#include <memory>
+#include <vector>
+#include "lite/core/mir/fusion/flatten_fc_fuser.h"
+#include "lite/core/mir/pass_registry.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+void FlattenFcFusePass::Apply(const std::unique_ptr<SSAGraph>& graph) {
+  fusion::FlattenFcFuser flatten_fuser(" ");
+  flatten_fuser(graph.get());
+}
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle
+
+REGISTER_MIR_PASS(lite_flatten_fc_fuse_pass,
+                  paddle::lite::mir::FlattenFcFusePass)
+    .BindTargets({TARGET(kOpenCL)})
+    .BindKernel("fc");
diff --git a/lite/core/mir/fusion/flatten_fc_fuse_pass.h b/lite/core/mir/fusion/flatten_fc_fuse_pass.h
@@ -0,0 +1,32 @@
+// Copyright (c) 2021 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 <memory>
+#include <string>
+#include "lite/core/mir/pass.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+
+class FlattenFcFusePass : public ProgramPass {
+ public:
+  void Apply(const std::unique_ptr<SSAGraph>& graph) override;
+};
+
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle
diff --git a/lite/core/mir/fusion/flatten_fc_fuser.cc b/lite/core/mir/fusion/flatten_fc_fuser.cc
@@ -0,0 +1,86 @@
+// Copyright (c) 2021 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 "lite/core/mir/fusion/flatten_fc_fuser.h"
+#include <memory>
+#include <vector>
+
+namespace paddle {
+namespace lite {
+namespace mir {
+namespace fusion {
+
+void FlattenFcFuser::BuildPattern() {
+  // flatten_contiguous_range
+  PMNode* x = VarNode("x")
+                  ->assert_is_op_input("flatten_contiguous_range", "X")
+                  ->AsInput();
+  PMNode* flatten_contiguous_range =
+      OpNode("flatten_contiguous_range", "flatten_contiguous_range")
+          ->AsIntermediate();
+  PMNode* out = VarNode("output")
+                    ->assert_is_op_output("flatten_contiguous_range", "Out")
+                    ->AsIntermediate();
+  PMNode* xshape =
+      VarNode("xshape")
+          ->assert_is_op_output("flatten_contiguous_range", "XShape")
+          ->AsIntermediate();
+
+  // fc
+  // PMNode* input   = VarNode("input")->assert_is_op_input("fc",
+  // "Input")->AsIntermediate();
+  PMNode* weights =
+      VarNode("weights")->assert_is_op_input("fc", "W")->AsInput();
+  PMNode* bias = VarNode("bias")->assert_is_op_input("fc", "Bias")->AsInput();
+  PMNode* fc = OpNode("fc", "fc")->AsIntermediate();
+  PMNode* fc_out =
+      VarNode("fc_out")->assert_is_op_output("fc", "Out")->AsOutput();
+
+  // create topology.
+  std::vector<PMNode*> fc_inputs{bias, weights, out};
+  *x >> *flatten_contiguous_range >> *out;
+  *flatten_contiguous_range >> *xshape;
+  fc_inputs >> *fc >> *fc_out;
+}
+
+void FlattenFcFuser::InsertNewNode(SSAGraph* graph,
+                                   const key2nodes_t& matched) {
+  auto op_desc = GenOpDesc(matched);
+  auto fc_op = LiteOpRegistry::Global().Create("fc");
+  auto fc_old = matched.at("fc")->stmt()->op();
+  auto* scope = fc_old->scope();
+  auto& valid_places = fc_old->valid_places();
+  fc_op->Attach(op_desc, scope);
+
+  auto* new_op_node = graph->GraphCreateInstructNode(fc_op, valid_places);
+
+  IR_NODE_LINK_TO(matched.at("x"), new_op_node);
+  IR_NODE_LINK_TO(matched.at("weights"), new_op_node);
+  IR_NODE_LINK_TO(matched.at("bias"), new_op_node);
+  IR_NODE_LINK_TO(new_op_node, matched.at("fc_out"));
+}
+
+cpp::OpDesc FlattenFcFuser::GenOpDesc(const key2nodes_t& matched) {
+  cpp::OpDesc op_desc = *matched.at("fc")->stmt()->op_info();
+  op_desc.SetInput("Input", {matched.at("x")->arg()->name});
+  op_desc.SetOutput("Out", {matched.at("fc_out")->arg()->name});
+  int in_num_col_dim = 1;
+  op_desc.SetAttr("in_num_col_dims", in_num_col_dim);
+  return op_desc;
+}
+
+}  // namespace fusion
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle
diff --git a/lite/core/mir/fusion/flatten_fc_fuser.h b/lite/core/mir/fusion/flatten_fc_fuser.h
@@ -0,0 +1,39 @@
+// Copyright (c) 2021 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 <memory>
+#include <string>
+#include "lite/core/mir/pattern_matcher_high_api.h"
+
+namespace paddle {
+namespace lite {
+namespace mir {
+namespace fusion {
+
+class FlattenFcFuser : public FuseBase {
+ public:
+  explicit FlattenFcFuser(const std::string& type) {}
+  void BuildPattern() override;
+  void InsertNewNode(SSAGraph* graph, const key2nodes_t& matched) override;
+
+ private:
+  cpp::OpDesc GenOpDesc(const key2nodes_t& matched) override;
+};
+
+}  // namespace fusion
+}  // namespace mir
+}  // namespace lite
+}  // namespace paddle
diff --git a/lite/core/optimizer.h b/lite/core/optimizer.h
@@ -113,6 +113,7 @@ class Optimizer {
          "lite_scale_activation_fuse_pass",             //
          "lite_elementwise_scale_fuse_pass",            //
          "lite_instance_norm_activation_fuse_pass",     //
+         "lite_flatten_fc_fuse_pass",                   //
          "lite_fc_prelu_fuse_pass",                     //
          "lite_elementwise_activation_fuse_pass",
          "lite_conv_scale_fuse_pass",
