extract load input

mindspore/ccsrc/backend/session/ascend_session.cc

@@ -571,8 +571,6 @@ void AscendSession::RunGraphImpl(const GraphId &graph_id, const std::vector<tens
   std::set<KernelGraphPtr> memo;
   SyncDataToExtraParams(NOT_NULL(kernel_graph), NOT_NULL(&memo));
   memo.clear();
-  // load input data from user input
-  LoadInputData(kernel_graph, inputs);
   if (debugger_) {
     debugger_->PreExecute(kernel_graph, graph_sum_);
   }

mindspore/ccsrc/backend/session/cpu_session.cc

@@ -130,6 +130,32 @@ void CPUSession::SyncValueNodeDeviceAddr(const std::shared_ptr<KernelGraph> &ker
   runtime_.SyncValueNodeDeviceAddr(kernel_graph.get());
 }
 
+void CPUSession::LoadInputData(const std::shared_ptr<KernelGraph> &kernel_graph,
+                               const std::vector<tensor::TensorPtr> &inputs_const) const {
+  MS_EXCEPTION_IF_NULL(kernel_graph);
+  auto &input_nodes = kernel_graph->inputs();
+  if (input_nodes.size() != inputs_const.size()) {
+    MS_LOG(EXCEPTION) << "Input size not equal to input node size!";
+  }
+  for (size_t input_idx = 0; input_idx < input_nodes.size(); ++input_idx) {
+    auto &item = input_nodes[input_idx];
+    MS_EXCEPTION_IF_NULL(item);
+    if (item->isa<Parameter>() && !HasAbstractMonad(item)) {
+      auto address = AnfAlgo::GetMutableOutputAddr(item, 0);
+      auto tensor = inputs_const[input_idx];
+      auto tensor_address = tensor->device_address();
+      MS_EXCEPTION_IF_NULL(address);
+      MS_EXCEPTION_IF_NULL(tensor);
+      if (tensor_address != nullptr && tensor_address != address &&
+          (std::dynamic_pointer_cast<device::DeviceAddress>(tensor_address)->DeviceType() !=
+             device::DeviceAddressType::kCPU ||
+           AnfAlgo::IsParameterWeight(item->cast<ParameterPtr>()))) {
+        tensor->data_sync(false);
+      }
+    }
+  }
+}
+
 void CPUSession::RunGraphImpl(const GraphId &graph_id, const std::vector<tensor::TensorPtr> &inputs,
                               VectorRef *outputs) {
   auto kernel_graph = GetGraph(graph_id);

mindspore/ccsrc/backend/session/cpu_session.h

@@ -44,6 +44,8 @@ class CPUSession : public SessionBasic {
                    const std::vector<int64_t> &tensors_mask) override;
   void RunOpImpl(const GraphInfo &graph_info, OpRunInfo *op_run_info, std::vector<tensor::TensorPtr> *input_tensors,
                  VectorRef *outputs, const std::vector<int64_t> &tensors_mask) override;
+  void LoadInputData(const std::shared_ptr<KernelGraph> &kernel_graph,
+                     const std::vector<tensor::TensorPtr> &inputs_const) const override;
 
  private:
   void Reorder(std::vector<CNodePtr> *node_list);

mindspore/ccsrc/backend/session/executor.cc

@@ -161,6 +161,7 @@ void RunGraphTask::Run() {
   }
   graph->ResetGraphRunningStatus();
   try {
+    session_->LoadInputs(graph_id_, input_tensors_);
     session_->RunGraphImpl(graph_id_, input_tensors_, &outputs_);
     UpdateOutputTensors(&outputs_, tensor_to_node_);
   } catch (const std::exception &e) {

mindspore/ccsrc/backend/session/gpu_session.cc

@@ -425,8 +425,6 @@ void GPUSession::RunGraphImpl(const GraphId &graph_id, const std::vector<tensor:
   MS_LOG(INFO) << "RunGraph graph_id: " << graph_id;
   // In pynative mode, device addresses of tensors in value nodes change.
   SyncValueNodeDeviceAddr(kernel_graph);
-  // Load input data from user input
-  LoadInputData(kernel_graph, inputs);
   if (debugger_) {
     debugger_->PreExecute(kernel_graph, graph_sum_);
   }

mindspore/ccsrc/backend/session/gpu_session.h

@@ -47,6 +47,8 @@ class GPUSession : public SessionBasic {
                  VectorRef *outputs, const std::vector<int64_t> &tensors_mask) override;
   std::shared_ptr<device::Bucket> CreateBucket(uint32_t bucket_id, uint32_t bucket_size) override;
   std::string GetCommWorldGroup() override { return kNcclWorldGroup; }
+  void LoadInputData(const std::shared_ptr<KernelGraph> &kernel_graph,
+                     const std::vector<tensor::TensorPtr> &inputs_const) const override;
 
  private:
   void SelectKernel(const std::shared_ptr<KernelGraph> &kernel_graph) const;
@@ -71,9 +73,6 @@ class GPUSession : public SessionBasic {
 
   void RunOpClearMemory(KernelGraph *kernel_graph) const;
 
-  void LoadInputData(const std::shared_ptr<KernelGraph> &kernel_graph,
-                     const std::vector<tensor::TensorPtr> &inputs_const) const override;
-
   void Execute(const std::shared_ptr<KernelGraph> &kernel_graph) const;
 
   void Dump(const std::shared_ptr<KernelGraph> &kernel_graph) const;

mindspore/ccsrc/backend/session/kernel_graph.cc

@@ -180,8 +180,8 @@ std::vector<AnfNodePtr> KernelGraph::outputs() const {
   return std::vector<AnfNodePtr>(1, graph_output);
 }
 
-void KernelGraph::VisitNodeDescendants(const AnfNodePtr &node, std::queue<AnfNodePtr> *visit_queue,
-                                       std::unordered_set<AnfNodePtr> *visited_nodes, bool comm_first) {
+void KernelGraph::EnqueueActiveNodes(const AnfNodePtr &node, std::queue<AnfNodePtr> *visit_queue,
+                                     std::unordered_set<AnfNodePtr> *visited_nodes, bool comm_first) {
   MS_EXCEPTION_IF_NULL(visit_queue);
   MS_EXCEPTION_IF_NULL(visited_nodes);
   auto it = node_output_edges_.find(node);
@@ -241,7 +241,7 @@ void KernelGraph::SetExecOrderByDefault() {
   while (!seed_nodes.empty() || !delay_comm_stack.empty()) {
     // seed nodes first, then delay comm nodes
     if (seed_nodes.empty()) {
-      VisitNodeDescendants(delay_comm_stack.top(), &communication_descendants, &visited_nodes, false);
+      EnqueueActiveNodes(delay_comm_stack.top(), &communication_descendants, &visited_nodes, false);
       delay_comm_stack.pop();
     } else {
       zero_input_nodes.push(seed_nodes.front());
@@ -272,16 +272,16 @@ void KernelGraph::SetExecOrderByDefault() {
       }
       if (optimize_comm) {
         while (!delay_comm_stack.empty()) {
-          VisitNodeDescendants(delay_comm_stack.top(), &communication_descendants, &visited_nodes, false);
+          EnqueueActiveNodes(delay_comm_stack.top(), &communication_descendants, &visited_nodes, false);
           delay_comm_stack.pop();
         }
         delay_comm_stack.push(node);
       } else if (is_fused_comm) {
         delay_comm_stack.push(node);
       } else if (is_communication_descendant) {
-        VisitNodeDescendants(node, &communication_descendants, &visited_nodes);
+        EnqueueActiveNodes(node, &communication_descendants, &visited_nodes);
       } else {
-        VisitNodeDescendants(node, &zero_input_nodes, &visited_nodes);
+        EnqueueActiveNodes(node, &zero_input_nodes, &visited_nodes);
       }
     }
   }

mindspore/ccsrc/backend/session/kernel_graph.h

@@ -283,8 +283,8 @@ class KernelGraph : public FuncGraph {
   void SetKernelInfoForNode(const AnfNodePtr &node) const;
   void ResetInFormat(const AnfNodePtr &node, const std::string &format) const;
   AnfNodePtr MakeValueNode(const AnfNodePtr &node);
-  void VisitNodeDescendants(const AnfNodePtr &node, std::queue<AnfNodePtr> *visit_queue,
-                            std::unordered_set<AnfNodePtr> *visited_nodes, bool comm_first = true);
+  void EnqueueActiveNodes(const AnfNodePtr &node, std::queue<AnfNodePtr> *visit_queue,
+                          std::unordered_set<AnfNodePtr> *visited_nodes, bool comm_first = true);
   // update node edge list
   void UpdateNodeEdgeList(std::queue<AnfNodePtr> *seed_nodes);
   // add node depend edge by data edge or control depend

mindspore/ccsrc/backend/session/session_basic.h

@@ -181,6 +181,13 @@ class SessionBasic : public std::enable_shared_from_this<SessionBasic> {
                                const std::map<KernelWithIndex, size_t> &cnode_refcount) {}
   virtual void SetSummaryNodes(KernelGraph *graph);
 
+  void LoadInputs(const GraphId &graph_id, const std::vector<tensor::TensorPtr> &inputs_const) {
+    auto kernel_graph = GetGraph(graph_id);
+    MS_EXCEPTION_IF_NULL(kernel_graph);
+    MS_LOG(INFO) << "Load inputs";
+    LoadInputData(kernel_graph, inputs_const);
+  }
+
   virtual void LoadInputData(const std::shared_ptr<KernelGraph> &kernel_graph,
                              const std::vector<tensor::TensorPtr> &inputs_const) const;
   void EraseValueNodeTensor(const std::vector<int64_t> &tensors_mask, std::vector<tensor::TensorPtr> *input_tensors);

mindspore/ccsrc/runtime/device/cpu/cpu_kernel_runtime.cc

@@ -283,20 +283,14 @@ void CPUKernelRuntime::BindInputTensorAddressPtr(const session::KernelGraph &ker
   if (input_nodes.size() != inputs.size()) {
     MS_LOG(EXCEPTION) << "Input size not equal to input node size!";
   }
-  size_t input_idx = 0;
-  for (auto &item : input_nodes) {
+  for (size_t input_idx = 0; input_idx < input_nodes.size(); ++input_idx) {
+    auto &item = input_nodes[input_idx];
     MS_EXCEPTION_IF_NULL(item);
     if (item->isa<Parameter>() && !HasAbstractMonad(item)) {
       auto address = AnfAlgo::GetMutableOutputAddr(item, 0);
       auto tensor = inputs[input_idx];
-      auto tensor_address = tensor->device_address();
       MS_EXCEPTION_IF_NULL(address);
       MS_EXCEPTION_IF_NULL(tensor);
-      if (tensor_address != nullptr && tensor_address != address &&
-          (std::dynamic_pointer_cast<device::DeviceAddress>(tensor_address)->DeviceType() != DeviceAddressType::kCPU ||
-           AnfAlgo::IsParameterWeight(item->cast<ParameterPtr>()))) {
-        tensor->data_sync(false);
-      }
       if (GetTypeByte(TypeIdToType(tensor->data_type())) == GetTypeByte(TypeIdToType(address->type_id_))) {
         address->ptr_ = tensor->data_c();
       } else {
@@ -318,7 +312,6 @@ void CPUKernelRuntime::BindInputTensorAddressPtr(const session::KernelGraph &ker
       address->ref_count_ = INIT_NODE_REF;
       tensor->set_device_address(address);
     }
-    input_idx++;
   }
 }