!12802 Add gpu inference options

From: @wilfchen
Reviewed-by: @cristoval,@limingqi107
Signed-off-by: @limingqi107

include/api/cell.h

@@ -22,6 +22,7 @@
 #include "include/api/status.h"
 #include "include/api/types.h"
 #include "include/api/graph.h"
+#include "include/api/context.h"
 
 namespace mindspore {
 class InputAndOutput;
@@ -97,6 +98,7 @@ class MS_API GraphCell final : public Cell<GraphCell> {
   explicit GraphCell(Graph &&);
   explicit GraphCell(const std::shared_ptr<Graph> &);
 
+  void SetContext(const std::shared_ptr<Context> &context);
   const std::shared_ptr<Graph> &GetGraph() const { return graph_; }
   Status Run(const std::vector<MSTensor> &inputs, std::vector<MSTensor> *outputs) override;
   std::vector<MSTensor> GetInputs();

include/api/context.h

@@ -81,6 +81,9 @@ struct MS_API ModelContext : public Context {
   static inline void SetFusionSwitchConfigPath(const std::shared_ptr<Context> &context, const std::string &cfg_path);
   static inline std::string GetFusionSwitchConfigPath(const std::shared_ptr<Context> &context);
 
+  static inline void SetGpuTrtInferMode(const std::shared_ptr<Context> &context, const std::string &gpu_trt_infer_mode);
+  static inline std::string GetGpuTrtInferMode(const std::shared_ptr<Context> &context);
+
  private:
   // api without std::string
   static void SetInsertOpConfigPath(const std::shared_ptr<Context> &context, const std::vector<char> &cfg_path);
@@ -101,6 +104,9 @@ struct MS_API ModelContext : public Context {
 
   static void SetFusionSwitchConfigPath(const std::shared_ptr<Context> &context, const std::vector<char> &cfg_path);
   static std::vector<char> GetFusionSwitchConfigPathChar(const std::shared_ptr<Context> &context);
+
+  static void SetGpuTrtInferMode(const std::shared_ptr<Context> &context, const std::vector<char> &gpu_trt_infer_mode);
+  static std::vector<char> GetGpuTrtInferModeChar(const std::shared_ptr<Context> &context);
 };
 
 void GlobalContext::SetGlobalDeviceTarget(const std::string &device_target) {
@@ -155,5 +161,12 @@ void ModelContext::SetFusionSwitchConfigPath(const std::shared_ptr<Context> &con
 std::string ModelContext::GetFusionSwitchConfigPath(const std::shared_ptr<Context> &context) {
   return CharToString(GetFusionSwitchConfigPathChar(context));
 }
+
+void ModelContext::SetGpuTrtInferMode(const std::shared_ptr<Context> &context, const std::string &gpu_trt_infer_mode) {
+  SetGpuTrtInferMode(context, StringToChar(gpu_trt_infer_mode));
+}
+std::string ModelContext::GetGpuTrtInferMode(const std::shared_ptr<Context> &context) {
+  return CharToString(GetGpuTrtInferModeChar(context));
+}
 }  // namespace mindspore
 #endif  // MINDSPORE_INCLUDE_API_CONTEXT_H

mindspore/ccsrc/cxx_api/cell.cc

@@ -78,6 +78,8 @@ GraphCell::GraphCell(Graph &&graph)
   executor_->SetGraph(graph_);
 }
 
+void GraphCell::SetContext(const std::shared_ptr<Context> &context) { return executor_->SetContext(context); }
+
 Status GraphCell::Run(const std::vector<MSTensor> &inputs, std::vector<MSTensor> *outputs) {
   MS_EXCEPTION_IF_NULL(executor_);
   return executor_->Run(inputs, outputs);

mindspore/ccsrc/cxx_api/context.cc

@@ -31,6 +31,8 @@ constexpr auto kModelOptionPrecisionMode = "mindspore.option.precision_mode";
 // "force_fp16", "allow_fp32_to_fp16", "must_keep_origin_dtype" or "allow_mix_precision", default as "force_fp16"
 constexpr auto kModelOptionOpSelectImplMode = "mindspore.option.op_select_impl_mode";
 constexpr auto KModelOptionFusionSwitchCfgPath = "mindspore.option.fusion_switch_config_file_path";
+// "False": Inference with native backend, "True": Inference with Tensor-RT engine, default as "False"
+constexpr auto kModelOptionGpuTrtInferMode = "mindspore.option.gpu_trt_infer_mode";
 
 namespace mindspore {
 struct Context::Data {
@@ -217,4 +219,20 @@ std::vector<char> ModelContext::GetFusionSwitchConfigPathChar(const std::shared_
   const std::string &ref = GetValue<std::string>(context, KModelOptionFusionSwitchCfgPath);
   return StringToChar(ref);
 }
+
+void ModelContext::SetGpuTrtInferMode(const std::shared_ptr<Context> &context,
+                                      const std::vector<char> &gpu_trt_infer_mode) {
+  MS_EXCEPTION_IF_NULL(context);
+  if (context->data == nullptr) {
+    context->data = std::make_shared<Data>();
+    MS_EXCEPTION_IF_NULL(context->data);
+  }
+  context->data->params[kModelOptionGpuTrtInferMode] = CharToString(gpu_trt_infer_mode);
+}
+
+std::vector<char> ModelContext::GetGpuTrtInferModeChar(const std::shared_ptr<Context> &context) {
+  MS_EXCEPTION_IF_NULL(context);
+  const std::string &ref = GetValue<std::string>(context, kModelOptionGpuTrtInferMode);
+  return StringToChar(ref);
+}
 }  // namespace mindspore

mindspore/ccsrc/cxx_api/graph/gpu/gpu_graph_impl.cc

@@ -51,7 +51,10 @@ Status GPUGraphImpl::InitEnv() {
   ms_context->set_param<int>(MS_CTX_EXECUTION_MODE, kGraphMode);
   ms_context->set_param<uint32_t>(MS_CTX_DEVICE_ID, device_id_);
   ms_context->set_param<std::string>(MS_CTX_DEVICE_TARGET, kGPUDevice);
-  ms_context->set_param<bool>(MS_CTX_ENABLE_INFER_OPT, true);
+  auto enable_trt = ModelContext::GetGpuTrtInferMode(graph_context_);
+  if (enable_trt == "True") {
+    ms_context->set_param<bool>(MS_CTX_ENABLE_INFER_OPT, true);
+  }
 
   session_impl_ = session::SessionFactory::Get().Create(kGpuInferenceDevice);
   if (session_impl_ == nullptr) {

mindspore/ccsrc/cxx_api/graph/graph_impl.h

@@ -29,11 +29,12 @@
 namespace mindspore {
 class GraphCell::GraphImpl {
  public:
-  GraphImpl() = default;
+  GraphImpl() : graph_(nullptr), graph_context_(nullptr) {}
   virtual ~GraphImpl() = default;
 
   std::shared_ptr<Graph::GraphData> &MutableGraphData() const { return graph_->graph_data_; }
   void SetGraph(const std::shared_ptr<Graph> &graph) { graph_ = graph; }
+  void SetContext(const std::shared_ptr<Context> &context) { graph_context_ = context; }
 
   virtual Status Run(const std::vector<MSTensor> &inputs, std::vector<MSTensor> *outputs) = 0;
   virtual Status Load() = 0;
@@ -43,6 +44,7 @@ class GraphCell::GraphImpl {
 
  protected:
   std::shared_ptr<Graph> graph_;
+  std::shared_ptr<Context> graph_context_;
 };
 }  // namespace mindspore
 #endif  // MINDSPORE_CCSRC_CXX_API_GRAPH_GRAPH_IMPL_H

mindspore/ccsrc/cxx_api/model/ms/ms_model.cc

@@ -70,6 +70,7 @@ std::shared_ptr<GraphCell> MsModel::GenerateGraphCell(const std::vector<std::vec
   MS_EXCEPTION_IF_NULL(graph);
   auto graph_cell = std::make_shared<GraphCell>(graph);
   MS_EXCEPTION_IF_NULL(graph_cell);
+  graph_cell->SetContext(model_context_);
   auto ret = ModelImpl::Load(graph_cell);
   if (ret != kSuccess) {
     MS_LOG(ERROR) << "Load failed.";
@@ -95,6 +96,7 @@ Status MsModel::Build() {
   MS_EXCEPTION_IF_NULL(graph);
   auto graph_cell = std::make_shared<GraphCell>(graph);
   MS_EXCEPTION_IF_NULL(graph_cell);
+  graph_cell->SetContext(model_context_);
   auto ret = ModelImpl::Load(graph_cell);
   if (ret != kSuccess) {
     MS_LOG(ERROR) << "Load failed.";