!10465 npu Avoid multiple checks

From: @yeyunpeng2020
Reviewed-by: @zhang_xue_tong,@zhanghaibo5
Signed-off-by: @zhang_xue_tong

mindspore/lite/src/runtime/agent/npu/npu_manager.cc

@@ -60,10 +60,10 @@ void NPUManager::Reset() {
   domi::HiaiIrBuild ir_build;
   for (const auto &model_map : models_) {
     auto model = model_map.second;
-    if (!model->is_freed) {
+    if (!model->is_freed_) {
       ir_build.ReleaseModelBuff(*model->model_buffer_data_);
       model->model_buffer_data_ = nullptr;
-      model->is_freed = true;
+      model->is_freed_ = true;
       model->desc_.reset();
       model->desc_ = nullptr;
     }
@@ -88,12 +88,19 @@ bool NPUManager::CheckDDKVersion() {
   return true;
 }
 bool NPUManager::IsSupportNPU() {
-  if (IsKirinChip() && CheckEMUIVersion() && CheckDDKVersion()) {
-    MS_LOG(INFO) << "The current device support NPU.";
-    return true;
+  // Avoid multiple checks
+  if (!is_check_version_) {
+    is_check_version_ = true;
+    if (IsKirinChip() && CheckEMUIVersion() && CheckDDKVersion()) {
+      is_support_ = true;
+      MS_LOG(INFO) << "The current device support NPU.";
+    } else {
+      is_support_ = false;
+      MS_LOG(INFO) << "The current device NOT SUPPORT NPU.";
+    }
+    return is_support_;
   } else {
-    MS_LOG(INFO) << "The current device NOT SUPPORT NPU.";
-    return false;
+    return is_support_;
   }
 }
 
@@ -173,7 +180,7 @@ int NPUManager::LoadOMModel() {
     }
     total++;
     auto model = model_map.second;
-    if (model->is_loaded && model->is_freed) {
+    if (model->is_loaded_ && model->is_freed_) {
       continue;
     }
     models_desc.push_back(model->desc_);
@@ -222,7 +229,7 @@ int NPUManager::LoadModel(const std::shared_ptr<hiai::AiModelMngerClient> &clien
 
   for (const auto &desc : desc_list) {
     auto it = models_.find(desc->GetName());
-    it->second->is_loaded = true;
+    it->second->is_loaded_ = true;
     it->second->client_ = client;
   }
 

mindspore/lite/src/runtime/agent/npu/npu_manager.h

@@ -36,8 +36,8 @@ struct SubGraphModel {
   SubGraphModel(int index, std::string model_name, domi::ModelBufferData *model_buffer_data)
       : index_(index), model_name_(std::move(model_name)), model_buffer_data_(model_buffer_data) {}
 
-  bool is_freed = false;
-  bool is_loaded = false;
+  bool is_freed_ = false;
+  bool is_loaded_ = false;
   int index_;
   std::string model_name_;
   std::shared_ptr<domi::ModelBufferData> model_buffer_data_;
@@ -84,6 +84,8 @@ class NPUManager {
 
  private:
   int index_ = 0;
+  bool is_check_version_ = false;
+  bool is_support_ = false;
   std::unordered_map<std::string, SubGraphModel *> models_;
   std::vector<std::shared_ptr<hiai::AiModelMngerClient>> clients_;
 };

mindspore/lite/src/runtime/agent/npu/optimizer/npu_insert_transform_pass.cc

@@ -112,9 +112,6 @@ int NPUInsertTransformPass::InsertPostNode(const InnerContext *context, kernel::
 }
 
 int NPUInsertTransformPass::Run() {
-  if (!context_->IsNpuEnabled()) {
-    return RET_OK;
-  }
   for (size_t i = 0; i < all_kernels_->size(); i++) {
     auto kernel = (*all_kernels_)[i];
     if (kernel->desc().arch != kNPU) {

mindspore/lite/src/runtime/agent/npu/optimizer/npu_pass_manager.cc

@@ -20,7 +20,10 @@
 namespace mindspore::lite {
 
 void NPUPassManager::AddPass(NPUBasePass *pass) { all_pass_.push_back(pass); }
-int NPUPassManager::Run() {
+int NPUPassManager::Run(const InnerContext *context) {
+  if (!context->IsNpuEnabled()) {
+    return RET_OK;
+  }
   for (auto pass : all_pass_) {
     auto ret = pass->Run();
     if (ret != RET_OK) {

mindspore/lite/src/runtime/agent/npu/optimizer/npu_pass_manager.h

@@ -18,6 +18,7 @@
 #define MINDSPORE_LITE_SRC_RUNTIME_AGENT_NPU_OPTIMIZER_NPU_PASS_MANAGER_H_
 #include <vector>
 #include "src/runtime/agent/npu/optimizer/npu_base_pass.h"
+#include "src/inner_context.h"
 namespace mindspore::lite {
 class NPUPassManager {
  public:
@@ -30,7 +31,7 @@ class NPUPassManager {
 
   void AddPass(NPUBasePass *pass);
 
-  int Run();
+  int Run(const InnerContext *context);
 
   void Clear();
 

mindspore/lite/src/runtime/agent/npu/optimizer/npu_transform_pass.cc

@@ -97,9 +97,6 @@ int NPUTransformPass::InsertPostNode(const InnerContext *context, kernel::LiteKe
 }
 
 int NPUTransformPass::Run() {
-  if (!context_->IsNpuEnabled()) {
-    return RET_OK;
-  }
   for (size_t i = 0; i < all_kernels_->size();) {
     auto kernel = (*all_kernels_)[i];
     if (kernel->desc().arch != kNPU || npu_trans_nodes.find(kernel->Type()) == npu_trans_nodes.end()) {

mindspore/lite/src/scheduler.cc

@@ -574,7 +574,7 @@ int Scheduler::RunPass(std::vector<kernel::LiteKernel *> *dst_kernels) {
   auto fusion_pass = new NPUFusionPass(dst_kernels);
   mindspore::lite::NPUPassManager::GetInstance()->AddPass(fusion_pass);
 
-  ret = mindspore::lite::NPUPassManager::GetInstance()->Run();
+  ret = mindspore::lite::NPUPassManager::GetInstance()->Run(context_);
 #endif
   return ret;
 }

mindspore/lite/test/models_npu.cfg

@@ -1,3 +1,3 @@
-mobilenet_v1_1.0_224.tflite 3
-squeezenet.tflite 3
-inception_v3.tflite 3
+mobilenet_v1_1.0_224.tflite 1.5
+squeezenet.tflite 1.5
+inception_v3.tflite 0.5