oneDNN BatchNorm + Act fusion pass. (#27912)

paddle/fluid/framework/ir/CMakeLists.txt

@@ -110,6 +110,7 @@ if(WITH_MKLDNN)
     pass_library(cpu_quantize_squash_pass inference DIR mkldnn)
     pass_library(reshape_transpose_matmul_mkldnn_fuse_pass inference DIR mkldnn)
     pass_library(matmul_transpose_reshape_fuse_pass inference DIR mkldnn)
+    pass_library(batch_norm_act_fuse_pass inference DIR mkldnn)
 endif()
 
 cc_library(fuse_bn_act_pass SRCS fuse_bn_act_pass.cc DEPS pass graph_pattern_detector )
@@ -151,6 +152,7 @@ if (WITH_MKLDNN)
     cc_test(test_conv_activation_mkldnn_fuse_pass SRCS mkldnn/conv_activation_mkldnn_fuse_pass_tester.cc DEPS conv_activation_mkldnn_fuse_pass)
     cc_test(test_conv_concat_relu_mkldnn_fuse_pass SRCS mkldnn/conv_concat_relu_mkldnn_fuse_pass_tester.cc DEPS conv_concat_relu_mkldnn_fuse_pass)
     cc_test(test_conv_elementwise_add_mkldnn_fuse_pass SRCS mkldnn/conv_elementwise_add_mkldnn_fuse_pass_tester.cc DEPS conv_elementwise_add_mkldnn_fuse_pass)
+    cc_test(test_batch_norm_act_fuse_pass SRCS mkldnn/batch_norm_act_fuse_pass_tester.cc DEPS batch_norm_act_fuse_pass)
     set(TEST_CONV_BN_PASS_DEPS conv_bn_fuse_pass graph_to_program_pass conv_op conv_transpose_op math_function im2col vol2col batch_norm_op gelu_op activation_op elementwise_add_op concat_and_split naive_executor device_context)
 if (WITH_GPU)
     set(TEST_CONV_BN_PASS_DEPS ${TEST_CONV_BN_PASS_DEPS} depthwise_conv)

paddle/fluid/framework/ir/graph_pattern_detector.cc

@@ -1188,6 +1188,26 @@ PDNode *patterns::BatchNormActGrad::operator()(
   return bn_grad;
 }
 
+PDNode *patterns::BatchNormActOneDNN::operator()(const std::string &act_type) {
+  auto *bn_x = pattern->NewNode(bn_in_repr())
+                   ->AsInput()
+                   ->assert_is_op_input("batch_norm", "X");
+  auto *bn = pattern->NewNode(batch_norm_repr())->assert_is_op("batch_norm");
+  auto *bn_out = pattern->NewNode(bn_out_repr())
+                     ->assert_is_op_output("batch_norm", "Y")
+                     ->assert_is_op_input(act_type);
+  auto *act =
+      pattern->NewNode(act_repr())->assert_is_op(act_type)->AsIntermediate();
+  auto *act_out = pattern->NewNode(act_out_repr())
+                      ->assert_is_op_output(act_type, "Out")
+                      ->AsOutput();
+
+  bn->LinksFrom({bn_x}).LinksTo({bn_out});
+  act->LinksFrom({bn_out}).LinksTo({act_out});
+
+  return act_out;
+}
+
 PDNode *patterns::ElewiseAddAct::operator()(
     paddle::framework::ir::PDNode *ele_x_var,
     std::unordered_set<std::string> act_types) {

paddle/fluid/framework/ir/graph_pattern_detector.h

@@ -664,6 +664,27 @@ struct BatchNormActGrad : public PatternBase {
   PATTERN_DECL_NODE(d_bn_bias);
 };
 
+//
+// \brief   Pattern looking for batch_norm and a directly following activation
+// operator.
+//
+// \note    Currently only ReLU is supported as an activation function.
+//          Formula: act(bn(x))
+//          Op: batch_norm + act
+struct BatchNormActOneDNN : public PatternBase {
+  BatchNormActOneDNN(PDPattern* pattern, const std::string& name_scope)
+      : PatternBase(pattern, name_scope, "bn_act_onednn") {}
+
+  PDNode* operator()(const std::string& act_type);
+
+  // declare operator node's name
+  PATTERN_DECL_NODE(bn_in);
+  PATTERN_DECL_NODE(batch_norm);
+  PATTERN_DECL_NODE(act);
+  PATTERN_DECL_NODE(bn_out);
+  PATTERN_DECL_NODE(act_out);
+};
+
 // The following patterns are used to fuse elewise_add and act
 // formula: act(ele_add(x, y))
 // op: elementwise_add + act

paddle/fluid/framework/ir/mkldnn/batch_norm_act_fuse_pass.cc

+// Copyright (c) 2020 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/ir/mkldnn/batch_norm_act_fuse_pass.h"
+#include "paddle/fluid/framework/ir/graph_pattern_detector.h"
+#include "paddle/fluid/framework/op_version_registry.h"
+#include "paddle/fluid/platform/enforce.h"
+#include "paddle/fluid/string/pretty_log.h"
+
+namespace paddle {
+namespace framework {
+namespace ir {
+
+using string::PrettyLogDetail;
+
+void FuseBatchNormActOneDNNPass::ApplyImpl(Graph *graph) const {
+  std::string act_type("relu");
+  FuseBatchNormAct(graph, act_type);
+}
+
+void FuseBatchNormActOneDNNPass::FuseBatchNormAct(
+    Graph *graph, const std::string &act_type) const {
+  PADDLE_ENFORCE_NOT_NULL(
+      graph, platform::errors::InvalidArgument(
+                 "The input graph of "
+                 "FuseBatchNormActOneDNNPass should not be nullptr."));
+  FusePassBase::Init("bn_act", graph);
+
+  GraphPatternDetector gpd;
+  patterns::BatchNormActOneDNN bn_act_pattern(gpd.mutable_pattern(), "bn_act");
+  bn_act_pattern(act_type);
+
+  int found_bn_act_count = 0;
+  auto handler = [&](const GraphPatternDetector::subgraph_t &subgraph,
+                     Graph *g) {
+    VLOG(4) << "Fuse BatchNorm with ReLU activation op.";
+    // BN output
+    GET_IR_NODE_FROM_SUBGRAPH(bn_out, bn_out, bn_act_pattern);
+    // ACT output
+    GET_IR_NODE_FROM_SUBGRAPH(act_out, act_out, bn_act_pattern);
+    // ops
+    GET_IR_NODE_FROM_SUBGRAPH(batch_norm, batch_norm, bn_act_pattern);
+    GET_IR_NODE_FROM_SUBGRAPH(act, act, bn_act_pattern);
+
+    auto *bn_op = batch_norm->Op();
+
+    if (bn_op->HasAttr("use_mkldnn")) {
+      PADDLE_ENFORCE(
+          BOOST_GET_CONST(bool, bn_op->GetAttr("use_mkldnn")),
+          platform::errors::PreconditionNotMet(
+              "The BatchNorm+Act fusion may happen only when oneDNN library "
+              "is used."));
+    }
+
+    if (bn_op->HasAttr("trainable_statistics")) {
+      PADDLE_ENFORCE(
+          !BOOST_GET_CONST(bool, bn_op->GetAttr("trainable_statistics")),
+          platform::errors::PreconditionNotMet(
+              "The BatchNorm+Act fusion may happen only when mean and variance "
+              "are not calculated by current batch statistics."));
+    }
+
+    if (bn_op->HasAttr("is_test")) {
+      PADDLE_ENFORCE(
+          BOOST_GET_CONST(bool, bn_op->GetAttr("is_test")),
+          platform::errors::PreconditionNotMet(
+              "The BatchNorm+Act fusion may happen only during inference."));
+    }
+
+    bn_op->SetAttr("use_mkldnn", true);
+    bn_op->SetAttr("is_test", true);
+    bn_op->SetAttr("fuse_with_relu", true);
+    bn_op->SetAttr("trainable_statistics", false);
+    bn_op->SetOutput("Y", {act_out->Name()});
+
+    IR_OP_VAR_LINK(batch_norm, act_out);
+    GraphSafeRemoveNodes(g, {act, bn_out});
+    found_bn_act_count++;
+  };
+
+  gpd(graph, handler);
+  AddStatis(found_bn_act_count);
+  PrettyLogDetail("---    fused %d batch norm with relu activation",
+                  found_bn_act_count);
+}
+
+}  // namespace ir
+}  // namespace framework
+}  // namespace paddle
+
+REGISTER_PASS(batch_norm_act_fuse_pass,
+              paddle::framework::ir::FuseBatchNormActOneDNNPass);
+REGISTER_PASS_CAPABILITY(batch_norm_act_fuse_pass)
+    .AddCombination(
+        paddle::framework::compatible::OpVersionComparatorCombination()
+            .EQ("batch_norm", 0)
+            .EQ("relu", 0));

paddle/fluid/framework/ir/mkldnn/batch_norm_act_fuse_pass.h

+// Copyright (c) 2020 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/framework/ir/fuse_pass_base.h"
+#include "paddle/fluid/framework/ir/graph.h"
+
+namespace paddle {
+namespace framework {
+namespace ir {
+
+/*
+ * \brief   Fuse the BatchNorm and activation operators into single OneDNN's
+ *          BatchNorm with post-op.
+ *
+ * \note    Currently only ReLU is supported as an activation function.
+ */
+class FuseBatchNormActOneDNNPass : public FusePassBase {
+ public:
+  virtual ~FuseBatchNormActOneDNNPass() {}
+
+ protected:
+  void ApplyImpl(ir::Graph *graph) const override;
+
+  void FuseBatchNormAct(ir::Graph *graph, const std::string &act_types) const;
+};
+
+}  // namespace ir
+}  // namespace framework
+}  // namespace paddle

paddle/fluid/framework/ir/mkldnn/batch_norm_act_fuse_pass_tester.cc

+// Copyright (c) 2020 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 <gtest/gtest.h>
+#include <algorithm>
+#include <exception>
+#include <functional>
+#include <iterator>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include "paddle/fluid/framework/ir/graph_traits.h"
+#include "paddle/fluid/framework/ir/mkldnn/batch_norm_act_fuse_pass.h"
+#include "paddle/fluid/framework/op_desc.h"
+#include "paddle/fluid/framework/program_desc.h"
+#include "paddle/fluid/platform/errors.h"
+
+namespace paddle {
+namespace framework {
+namespace ir {
+
+// -------------------------- helper functions --------------------------------
+namespace {
+
+using InOutVarNamePair = std::pair<std::string, std::string>;
+using OpTypeCountPair = std::pair<std::string, int>;
+
+///
+/// @brief      Creates the specified operator and sets up its inputs/outputs.
+///
+/// @param      prog          The program descriptor to which we add new op.
+/// @param[in]  op_type_name  The operator type name.
+/// @param[in]  inputs        The vector of input pairs: {input_name, variable
+///                           name}
+/// @param[in]  outputs       The vector of output pairs {output_name, variable}
+/// @param[in]  use_mkldnn    The flag deciding whether or not to set
+///                           'use_mkldnn' attribute.
+///
+/// @return     Returns pointer to the created operator descriptor.
+///
+OpDesc* CreateOp(ProgramDesc* prog, const std::string& op_type_name,
+                 const std::vector<InOutVarNamePair>& inputs,
+                 const std::vector<InOutVarNamePair>& outputs,
+                 bool use_mkldnn = true) {
+  auto op = prog->MutableBlock(0)->AppendOp();
+  op->SetType(op_type_name);
+  op->SetAttr("use_mkldnn", use_mkldnn);
+
+  for (const auto& input : inputs) {
+    op->SetInput(input.first, {input.second});
+  }
+  for (const auto& output : outputs) {
+    op->SetOutput(output.first, {output.second});
+  }
+
+  return op;
+}
+
+///
+/// @brief      Check whether node 'to' is reachable from node 'from' in graph.
+///
+/// @param[in]  graph  The graph we're checking for reachability.
+/// @param[in]  from   The 'from' node name.
+/// @param[in]  to     The 'to' node name.
+///
+/// @return     True if there is connection between nodes 'from' and 'to'.
+///
+bool TestIsReachable(const Graph& graph, std::string from, std::string to) {
+  auto hash = [](const Node* node) -> std::string {
+    return node->Name() + std::to_string(node->id());
+  };
+
+  auto find_node = [&](const Graph& graph, const std::string& name) -> Node* {
+    for (auto& node : GraphTraits::DFS(graph)) {
+      if (name == hash(&node)) {
+        return &node;
+      }
+    }
+
+    return nullptr;
+  };
+
+  if (from == to) return true;
+
+  std::map<std::string, bool> visited;
+  // update the from and to strings to hashed equivs in loop from graph traits
+  for (auto& node : GraphTraits::DFS(graph)) {
+    auto hashed = hash(&node);
+    if (node.Name() == from) {
+      from = hashed;
+    }
+    if (node.Name() == to) {
+      to = hashed;
+    }
+    visited[hashed] = false;
+  }
+
+  visited[from] = true;
+
+  std::list<std::string> queue;
+  queue.push_back(from);
+
+  while (!queue.empty()) {
+    auto cur = find_node(graph, queue.front());
+    queue.pop_front();
+    if (cur == nullptr) {
+      return false;
+    }
+
+    for (auto n : cur->outputs) {
+      auto hashed_name = hash(n);
+      if (hashed_name == to) {
+        return true;
+      }
+
+      if (!visited[hashed_name]) {
+        visited[hashed_name] = true;
+        queue.push_back(hashed_name);
+      }
+    }
+  }
+  return false;
+}
+
+///
+/// @brief      Search through graph and counts provided operator occurences.
+///
+/// @param[in]  graph          The graph we search through.
+/// @param[in]  op_type_count  The vector of pairs {op_type_name, op count}
+///
+/// @note       After going through all graph nodes this function asserts
+///             whether counted number for each requested op is as expected.
+///
+void AssertOpsCount(const Graph& graph,
+                    std::vector<OpTypeCountPair> op_type_count) {
+  for (auto* node : graph.Nodes()) {
+    if (!node->IsOp()) {
+      continue;
+    }
+
+    const std::string op_type_name = node->Op()->Type();
+    auto op_it =
+        std::find_if(std::begin(op_type_count), std::end(op_type_count),
+                     [op_type_name](const OpTypeCountPair& p) {
+                       return op_type_name == p.first;
+                     });
+    if (op_it != std::end(op_type_count)) {
+      op_it->second--;
+    }
+  }
+
+  for (const OpTypeCountPair& p : op_type_count) {
+    EXPECT_EQ(p.second, 0);
+  }
+}
+
+///
+/// @brief      Builds a program descriptor.
+///
+/// @param[in]  transient_vars   The vector of transient variables names.
+/// @param[in]  persistent_vars  The vector of persistent variables names. Those
+///                              will have persistable attribute set to true.
+///
+/// @return     The program descriptor object.
+///
+ProgramDesc BuildProgramDesc(const std::vector<std::string>& transient_vars,
+                             const std::vector<std::string>& persistent_vars) {
+  ProgramDesc prog;
+
+  auto add_var_to_prog = [&prog](const std::string& var_name) -> VarDesc* {
+    auto var = prog.MutableBlock(0)->Var(var_name);
+    var->SetType(proto::VarType::LOD_TENSOR);
+    return var;
+  };
+
+  for (const auto& v : transient_vars) {
+    add_var_to_prog(v);
+  }
+
+  for (const auto& v : persistent_vars) {
+    auto* var = add_var_to_prog(v);
+    var->SetPersistable(true);
+  }
+
+  return prog;
+}
+
+///
+/// @brief      Execute pass on provided graph and perform checks.
+///
+/// @param      graph                The graph we run pass on.
+/// @param[in]  from                 The name of a 'starting' node sequence in a
+///                                  graph. This would be used to test for
+///                                  correct node connections.
+/// @param[in]  to                   The name of a 'ending' node sequence in a
+///                                  graph. This would be used to test for
+///                                  correct node connections.
+/// @param[in]  removed_nodes_count  The number of nodes we expect will be
+///                                  removed/fused after pass execution.
+/// @param[in]  added_nodes_count    The number of nodes we expect will be
+///                                  added after pass execution.
+///
+void RunPassAndAssert(Graph* graph, const std::string& from,
+                      const std::string& to, int removed_nodes_count,
+                      int added_nodes_count = 0) {
+  EXPECT_TRUE(TestIsReachable(*graph, from, to));
+  int original_nodes_num = graph->Nodes().size();
+  auto pass = PassRegistry::Instance().Get("batch_norm_act_fuse_pass");
+  pass->Apply(graph);
+  int current_nodes_num = graph->Nodes().size();
+
+  EXPECT_TRUE(TestIsReachable(*graph, from, to));
+  EXPECT_EQ(original_nodes_num - removed_nodes_count + added_nodes_count,
+            current_nodes_num);
+}
+
+void SetBatchNormAttrs(OpDesc* bn_op, bool is_test = true,
+                       bool trainable_stats = true) {
+  bn_op->SetAttr("is_test", is_test);
+  bn_op->SetAttr("trainable_statistics", trainable_stats);
+  bn_op->SetAttr("fuse_with_relu", false);
+}
+
+}  // namespace
+
+// ------------------------------ Test cases -----------------------------------
+
+// The below test cases are distinguished by whether following attributes have
+// true or false value:
+// - is_test
+// - trainable_statistics
+// The test case name would have only attributes with true value in its name.
+
+TEST(FuseBatchNormActOneDNNPass, ThrowIsTestTrainableStats) {
+  auto prog = BuildProgramDesc(
+      {"x", "m", "v", "bn_y", "act_y", "m_out", "var_out", "sm", "sv"},
+      {"scale", "bias"});
+  auto* bn_op = CreateOp(&prog, "batch_norm", {{"X", "x"},
+                                               {"Scale", "scale"},
+                                               {"Bias", "bias"},
+                                               {"Mean", "m"},
+                                               {"Variance", "v"}},
+                         {{"Y", "bn_y"},
+                          {"MeanOut", "m_out"},
+                          {"VarianceOut", "var_out"},
+                          {"SavedMean", "sm"},
+                          {"SavedVariance", "sv"}});
+  SetBatchNormAttrs(bn_op, true, true);
+  CreateOp(&prog, "relu", {{"X", "bn_y"}}, {{"Out", "act_y"}}, false);
+
+  Graph graph(prog);
+  // No fusion in this attribute configuration
+  constexpr int removed_nodes_count = 0;
+
+  EXPECT_THROW(RunPassAndAssert(&graph, "x", "act_y", removed_nodes_count),
+               paddle::platform::EnforceNotMet);
+}
+
+TEST(FuseBatchNormActOneDNNPass, FuseIsTest) {
+  auto prog =
+      BuildProgramDesc({"x", "m", "v", "bn_y", "act_y"}, {"scale", "bias"});
+  auto* bn_op = CreateOp(&prog, "batch_norm", {{"X", "x"},
+                                               {"Scale", "scale"},
+                                               {"Bias", "bias"},
+                                               {"Mean", "m"},
+                                               {"Variance", "v"}},
+                         {{"Y", "bn_y"}});
+  SetBatchNormAttrs(bn_op, true, false);
+  CreateOp(&prog, "relu", {{"X", "bn_y"}}, {{"Out", "act_y"}}, false);
+
+  Graph graph(prog);
+  constexpr int removed_nodes_count = 2;
+
+  RunPassAndAssert(&graph, "x", "act_y", removed_nodes_count);
+  AssertOpsCount(graph, {{"batch_norm", 1}, {"relu", 0}});
+
+  for (const auto* node : graph.Nodes()) {
+    if (node->IsOp() && node->Op()->Type() == "batch_norm") {
+      const auto* op = node->Op();
+      ASSERT_TRUE(op->HasAttr("use_mkldnn"));
+      EXPECT_TRUE(BOOST_GET_CONST(bool, op->GetAttr("use_mkldnn")));
+      ASSERT_TRUE(op->HasAttr("fuse_with_relu"));
+      EXPECT_TRUE(BOOST_GET_CONST(bool, op->GetAttr("fuse_with_relu")));
+      ASSERT_TRUE(op->HasAttr("trainable_statistics"));
+      EXPECT_FALSE(BOOST_GET_CONST(bool, op->GetAttr("trainable_statistics")));
+    }
+  }
+}
+
+TEST(FuseBatchNormActOneDNNPass, ThrowTrainableStats) {
+  auto prog = BuildProgramDesc(
+      {"x", "m", "v", "bn_y", "act_y", "m_out", "var_out", "sm", "sv"},
+      {"scale", "bias"});
+  auto* bn_op = CreateOp(&prog, "batch_norm", {{"X", "x"},
+                                               {"Scale", "scale"},
+                                               {"Bias", "bias"},
+                                               {"Mean", "m"},
+                                               {"Variance", "v"}},
+                         {{"Y", "bn_y"},
+                          {"MeanOut", "m_out"},
+                          {"VarianceOut", "var_out"},
+                          {"SavedMean", "sm"},
+                          {"SavedVariance", "sv"}});
+  SetBatchNormAttrs(bn_op, false, true);
+  CreateOp(&prog, "relu", {{"X", "bn_y"}}, {{"Out", "act_y"}}, false);
+
+  Graph graph(prog);
+  // No fusion in this attribute configuration
+  constexpr int removed_nodes_count = 0;
+
+  EXPECT_THROW(RunPassAndAssert(&graph, "x", "act_y", removed_nodes_count),
+               paddle::platform::EnforceNotMet);
+}
+
+TEST(FuseBatchNormActOneDNNPass, AllAttrsFalse) {
+  auto prog = BuildProgramDesc(
+      {"x", "m", "v", "bn_y", "act_y", "m_out", "var_out", "sm", "sv"},
+      {"scale", "bias"});
+  auto* bn_op = CreateOp(&prog, "batch_norm", {{"X", "x"},
+                                               {"Scale", "scale"},
+                                               {"Bias", "bias"},
+                                               {"Mean", "m"},
+                                               {"Variance", "v"}},
+                         {{"Y", "bn_y"},
+                          {"MeanOut", "m_out"},
+                          {"VarianceOut", "var_out"},
+                          {"SavedMean", "sm"},
+                          {"SavedVariance", "sv"}});
+  SetBatchNormAttrs(bn_op, false, false);
+  CreateOp(&prog, "relu", {{"X", "bn_y"}}, {{"Out", "act_y"}}, false);
+
+  Graph graph(prog);
+  // No fusion in this attribute configuration
+  constexpr int removed_nodes_count = 0;
+
+  EXPECT_THROW(RunPassAndAssert(&graph, "x", "act_y", removed_nodes_count),
+               paddle::platform::EnforceNotMet);
+}
+
+TEST(FuseBatchNormActOneDNNPass, ThrowUseMkldnn) {
+  auto prog = BuildProgramDesc(
+      {"x", "m", "v", "bn_y", "act_y", "m_out", "var_out", "sm", "sv"},
+      {"scale", "bias"});
+  auto* bn_op = CreateOp(&prog, "batch_norm", {{"X", "x"},
+                                               {"Scale", "scale"},
+                                               {"Bias", "bias"},
+                                               {"Mean", "m"},
+                                               {"Variance", "v"}},
+                         {{"Y", "bn_y"},
+                          {"MeanOut", "m_out"},
+                          {"VarianceOut", "var_out"},
+                          {"SavedMean", "sm"},
+                          {"SavedVariance", "sv"}},
+                         false);
+  SetBatchNormAttrs(bn_op, false, false);
+  CreateOp(&prog, "relu", {{"X", "bn_y"}}, {{"Out", "act_y"}}, false);
+
+  Graph graph(prog);
+  // No fusion in this attribute configuration
+  constexpr int removed_nodes_count = 0;
+
+  EXPECT_THROW(RunPassAndAssert(&graph, "x", "act_y", removed_nodes_count),
+               paddle::platform::EnforceNotMet);
+}
+
+}  // namespace ir
+}  // namespace framework
+}  // namespace paddle
+
+USE_PASS(batch_norm_act_fuse_pass);

paddle/fluid/inference/api/paddle_pass_builder.cc

@@ -207,6 +207,7 @@ void CpuPassStrategy::EnableMKLDNN() {
              "matmul_transpose_reshape_fuse_pass",         //
              // Disabled due to topology-dependent speed-up
              // "fc_mkldnn_pass",
+             "batch_norm_act_fuse_pass",
              "mkldnn_inplace_pass",  // This pass should be activated after
                                      // fuses
          })) {

python/paddle/fluid/tests/unittests/ir/inference/test_mkldnn_batch_norm_act_fuse_pass.py

+# Copyright (c) 2020 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.
+"""Test for fusion of batch norm and activation."""
+from __future__ import print_function
+
+import unittest
+import numpy as np
+
+import paddle.fluid as fluid
+from inference_pass_test import InferencePassTest
+from paddle import enable_static
+from paddle.fluid.core import PassVersionChecker
+
+enable_static()
+
+
+class BnReluOneDnnFusePassTest(InferencePassTest):
+    def setUp(self):
+        self.set_params()
+        with fluid.program_guard(self.main_program, self.startup_program):
+            data = fluid.data(
+                name="data", shape=[-1, 3, 100, 100], dtype="float32")
+            bn_out = fluid.layers.batch_norm(
+                input=data, is_test=True, use_global_stats=self.global_stats)
+            relu_out = fluid.layers.relu(bn_out)
+
+        self.feeds = {
+            "data": np.random.random((1, 3, 100, 100)).astype("float32")
+        }
+        self.fetch_list = [relu_out]
+        self.enable_mkldnn = True
+
+    def set_params(self):
+        self.global_stats = False
+        self.pass_name = "batch_norm_act_fuse_pass"
+
+    def test_check_output(self):
+        self.check_output()
+        self.assertTrue(PassVersionChecker.IsCompatible(self.pass_name))
+
+
+class BnReluGlobalStatsOneDnnFusePassTest(InferencePassTest):
+    def setUp(self):
+        self.set_params()
+        with fluid.program_guard(self.main_program, self.startup_program):
+            data = fluid.data(
+                name="data", shape=[-1, 3, 100, 100], dtype="float32")
+            bn_out = fluid.layers.batch_norm(
+                input=data, is_test=True, use_global_stats=self.global_stats)
+            relu_out = fluid.layers.relu(bn_out)
+
+        self.feeds = {
+            "data": np.random.random((1, 3, 100, 100)).astype("float32")
+        }
+        self.fetch_list = [relu_out]
+        self.enable_mkldnn = True
+
+    def set_params(self):
+        self.global_stats = True
+        self.pass_name = "batch_norm_act_fuse_pass"
+
+    def test_check_output(self):
+        self.check_output()
+        self.assertTrue(PassVersionChecker.IsCompatible(self.pass_name))
+
+
+if __name__ == "__main__":
+    unittest.main()