Merge 30b463fe52 into f0742ad340

Sync forward type schema

3rd_party/flatbuffers/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 2.8)
+cmake_minimum_required(VERSION 3.6)
 # generate compile_commands.json
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 include(CheckCXXSymbolExists)

apps/Android/MnnLlmChat/app/src/main/java/com/alibaba/mls/api/download/ModelDownloadManager.kt

@@ -181,7 +181,7 @@ class ModelDownloadManager private constructor(private val context: Context) {
             if (getDownloadedFile(modelId) != null) {
                 downloadInfo.downlodaState = DownloadInfo.DownloadSate.COMPLETED
                 downloadInfo.progress = 1.0
-            } else if (getDownloadSizeTotal(ApplicationProvider.get(), modelId) > 0) {
+            } else if (getDownloadSizeSaved(ApplicationProvider.get(), modelId) > 0) {
                 val totalSize = getDownloadSizeTotal(ApplicationProvider.get(), modelId)
                 val savedSize = getDownloadSizeSaved(ApplicationProvider.get(), modelId)
                 downloadInfo.totalSize = totalSize

apps/frameworks/sherpa-mnn/sherpa-mnn/csrc/offline-tts-kokoro-model.cc

@@ -32,9 +32,9 @@ class OfflineTtsKokoroModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto model_buf = ReadFile(config.kokoro.model);
+    auto model_path = config.kokoro.model.c_str();
     auto voices_buf = ReadFile(config.kokoro.voices);
-    Init(model_buf.data(), model_buf.size(), voices_buf.data(),
+    Init(model_path, voices_buf.data(),
          voices_buf.size());
   }
 
@@ -43,9 +43,9 @@ class OfflineTtsKokoroModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto model_buf = ReadFile(mgr, config.kokoro.model);
+    auto model_path = config.kokoro.model.c_str();
     auto voices_buf = ReadFile(mgr, config.kokoro.voices);
-    Init(model_buf.data(), model_buf.size(), voices_buf.data(),
+    Init(model_path, voices_buf.data(),
          voices_buf.size());
   }
 
@@ -96,9 +96,9 @@ class OfflineTtsKokoroModel::Impl {
   }
 
  private:
-  void Init(void *model_data, size_t model_data_length, const char *voices_data,
+  void Init(const char *model_path, const char *voices_data,
             size_t voices_data_length) {
-    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, (const uint8_t*)model_data, model_data_length,
+    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, model_path,
                                            sess_opts_.pManager, &sess_opts_.pConfig));
 
     GetInputNames(sess_.get(), &input_names_, &input_names_ptr_);

apps/frameworks/sherpa-mnn/sherpa-mnn/csrc/offline-tts-matcha-model.cc

@@ -31,8 +31,8 @@ class OfflineTtsMatchaModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto buf = ReadFile(config.matcha.acoustic_model);
-    Init(buf.data(), buf.size());
+    auto model_path = config.matcha.acoustic_model.c_str();
+    Init(model_path);
   }
 
   template <typename Manager>
@@ -40,8 +40,8 @@ class OfflineTtsMatchaModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto buf = ReadFile(mgr, config.matcha.acoustic_model);
-    Init(buf.data(), buf.size());
+    auto model_path = config.matcha.acoustic_model.c_str();
+    Init(model_path);
   }
 
   const OfflineTtsMatchaModelMetaData &GetMetaData() const {
@@ -117,8 +117,8 @@ class OfflineTtsMatchaModel::Impl {
   }
 
  private:
-  void Init(void *model_data, size_t model_data_length) {
-    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, (const uint8_t*)model_data, model_data_length,
+  void Init(const char *model_path) {
+    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, model_path,
                                            sess_opts_.pManager, &sess_opts_.pConfig));
 
     GetInputNames(sess_.get(), &input_names_, &input_names_ptr_);

apps/frameworks/sherpa-mnn/sherpa-mnn/csrc/offline-tts-vits-model.cc

@@ -31,8 +31,8 @@ class OfflineTtsVitsModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto buf = ReadFile(config.vits.model);
-    Init(buf.data(), buf.size());
+    auto model_path = config.vits.model.c_str();
+    Init(model_path);
   }
 
   template <typename Manager>
@@ -40,8 +40,8 @@ class OfflineTtsVitsModel::Impl {
       : config_(config),
         sess_opts_(GetSessionOptions(config)),
         allocator_{} {
-    auto buf = ReadFile(mgr, config.vits.model);
-    Init(buf.data(), buf.size());
+    auto model_path = config.vits.model.c_str();
+    Init(model_path);
   }
 
   MNN::Express::VARP Run(MNN::Express::VARP x, int sid, float speed) {
@@ -114,8 +114,8 @@ class OfflineTtsVitsModel::Impl {
   const OfflineTtsVitsModelMetaData &GetMetaData() const { return meta_data_; }
 
  private:
-  void Init(void *model_data, size_t model_data_length) {
-    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, (const uint8_t*)model_data, model_data_length,
+  void Init(const char *model_path) {
+    sess_ = std::unique_ptr<MNN::Express::Module>(MNN::Express::Module::load({}, {}, model_path,
                                            sess_opts_.pManager, &sess_opts_.pConfig));
 
     GetInputNames(sess_.get(), &input_names_, &input_names_ptr_);

apps/frameworks/sherpa-mnn/sherpa-mnn/jni/offline-tts.cc

@@ -16,13 +16,13 @@ static OfflineTtsConfig GetOfflineTtsConfig(JNIEnv *env, jobject config) {
   jfieldID fid;
 
   fid = env->GetFieldID(cls, "model",
-                        "Lcom/k2fsa/sherpa/onnx/OfflineTtsModelConfig;");
+                        "Lcom/k2fsa/sherpa/mnn/OfflineTtsModelConfig;");
   jobject model = env->GetObjectField(config, fid);
   jclass model_config_cls = env->GetObjectClass(model);
 
   // vits
   fid = env->GetFieldID(model_config_cls, "vits",
-                        "Lcom/k2fsa/sherpa/onnx/OfflineTtsVitsModelConfig;");
+                        "Lcom/k2fsa/sherpa/mnn/OfflineTtsVitsModelConfig;");
   jobject vits = env->GetObjectField(model, fid);
   jclass vits_cls = env->GetObjectClass(vits);
 
@@ -67,7 +67,7 @@ static OfflineTtsConfig GetOfflineTtsConfig(JNIEnv *env, jobject config) {
 
   // matcha
   fid = env->GetFieldID(model_config_cls, "matcha",
-                        "Lcom/k2fsa/sherpa/onnx/OfflineTtsMatchaModelConfig;");
+                        "Lcom/k2fsa/sherpa/mnn/OfflineTtsMatchaModelConfig;");
   jobject matcha = env->GetObjectField(model, fid);
   jclass matcha_cls = env->GetObjectClass(matcha);
 
@@ -115,7 +115,7 @@ static OfflineTtsConfig GetOfflineTtsConfig(JNIEnv *env, jobject config) {
 
   // kokoro
   fid = env->GetFieldID(model_config_cls, "kokoro",
-                        "Lcom/k2fsa/sherpa/onnx/OfflineTtsKokoroModelConfig;");
+                        "Lcom/k2fsa/sherpa/mnn/OfflineTtsKokoroModelConfig;");
   jobject kokoro = env->GetObjectField(model, fid);
   jclass kokoro_cls = env->GetObjectClass(kokoro);
 

cmake/oneDNN.cmake

@@ -5,7 +5,7 @@ set(ROOT ${CMAKE_CURRENT_LIST_DIR}/../3rd_party/)
 set(ONEDNN_DIR ${ROOT}/oneDNN/)
 set(MNN_BUILD_DIR ${CMAKE_CURRENT_LIST_DIR}/../build/)
 
-set(CONFIGURE_CMD cd ${ONEDNN_DIR} && cmake -DCMAKE_INSTALL_PREFIX=${MNN_BUILD_DIR} -DDNNL_BUILD_EXAMPLES=OFF -DDNNL_BUILD_TESTS=OFF -DDNNL_CPU_RUNTIME=SEQ)
+set(CONFIGURE_CMD cd ${ONEDNN_DIR} && cmake -DCMAKE_INSTALL_PREFIX=${MNN_BUILD_DIR} -DDNNL_BUILD_EXAMPLES=OFF -DDNNL_BUILD_TESTS=OFF -DDNNL_CPU_RUNTIME=SEQ -DCMAKE_POLICY_VERSION_MINIMUM=3.10)
 set(BUILD_CMD cd ${ONEDNN_DIR} && make -j8)
 set(INSTALL_CMD cd ${ONEDNN_DIR} && make install)
 

schema/current/MNN_generated.h

@@ -2655,18 +2655,24 @@ bool VerifyOpParameterVector(flatbuffers::Verifier &verifier, const flatbuffers:
 enum ForwardType {
   ForwardType_CPU = 0,
   ForwardType_METAL = 1,
-  ForwardType_OPENCL = 2,
-  ForwardType_OPENGLES = 3,
-  ForwardType_VULKAN = 4,
+  ForwardType_CUDA = 2,
+  ForwardType_OPENCL = 3,
+  ForwardType_AUTO = 4,
+  ForwardType_NNAPI = 5,
+  ForwardType_OPENGLES = 6,
+  ForwardType_VULKAN = 7,
   ForwardType_MIN = ForwardType_CPU,
   ForwardType_MAX = ForwardType_VULKAN
 };
 
-inline const ForwardType (&EnumValuesForwardType())[5] {
+inline const ForwardType (&EnumValuesForwardType())[8] {
   static const ForwardType values[] = {
     ForwardType_CPU,
     ForwardType_METAL,
+    ForwardType_CUDA,
     ForwardType_OPENCL,
+    ForwardType_AUTO,
+    ForwardType_NNAPI,
     ForwardType_OPENGLES,
     ForwardType_VULKAN
   };
@@ -2677,7 +2683,10 @@ inline const char * const *EnumNamesForwardType() {
   static const char * const names[] = {
     "CPU",
     "METAL",
+    "CUDA",
     "OPENCL",
+    "AUTO",
+    "NNAPI",
     "OPENGLES",
     "VULKAN",
     nullptr
@@ -8577,6 +8586,9 @@ inline const flatbuffers::TypeTable *OpParameterTypeTable() {
 
 inline const flatbuffers::TypeTable *ForwardTypeTypeTable() {
   static const flatbuffers::TypeCode type_codes[] = {
+    { flatbuffers::ET_CHAR, 0, 0 },
+    { flatbuffers::ET_CHAR, 0, 0 },
+    { flatbuffers::ET_CHAR, 0, 0 },
     { flatbuffers::ET_CHAR, 0, 0 },
     { flatbuffers::ET_CHAR, 0, 0 },
     { flatbuffers::ET_CHAR, 0, 0 },
@@ -8589,12 +8601,15 @@ inline const flatbuffers::TypeTable *ForwardTypeTypeTable() {
   static const char * const names[] = {
     "CPU",
     "METAL",
+    "CUDA",
     "OPENCL",
+    "AUTO",
+    "NNAPI",
     "OPENGLES",
     "VULKAN"
   };
   static const flatbuffers::TypeTable tt = {
-    flatbuffers::ST_ENUM, 5, type_codes, type_refs, nullptr, names
+    flatbuffers::ST_ENUM, 8, type_codes, type_refs, nullptr, names
   };
   return &tt;
 }

schema/default/MNN.fbs

@@ -458,7 +458,10 @@ table TensorDescribe {
 enum ForwardType : byte {
     CPU = 0,
     METAL,
+    CUDA,
     OPENCL,
+    AUTO,
+    NNAPI,
     OPENGLES,
     VULKAN,
 }

source/backend/cpu/CPUConvolution.cpp

@@ -278,7 +278,7 @@ public:
                                 const MNN::Op* op, Backend* backend) const override {
         auto convOp = op->main_as_Convolution2D();
 #ifdef MNN_USE_ONEDNN
-        return OneDNNConvInt8::create(backend, convOp, inputs, outputs);
+        return OneDNNConvInt8::create(backend, op, inputs, outputs);
 #endif
         auto core = static_cast<CPUBackend*>(backend)->functions();
         

source/backend/cpu/OneDNNConvInt8.cpp

@@ -14,9 +14,10 @@ OneDNNConvInt8::~OneDNNConvInt8() {
     // Do nothing
 }
 
-Execution* OneDNNConvInt8::create(Backend* backend, const MNN::Convolution2D* convParam, const std::vector<Tensor*>& inputs, const std::vector<Tensor *> &outputs) {
+Execution* OneDNNConvInt8::create(Backend* backend, const MNN::Op* op, const std::vector<Tensor*>& inputs, const std::vector<Tensor *> &outputs) {
     std::shared_ptr<OneDNNConvInt8::Resource> resource(new OneDNNConvInt8::Resource);
     resource->backend = backend;
+    const auto convParam              = op->main_as_Convolution2D();
     const auto convCommon             = convParam->common();
     const auto kw                     = convCommon->kernelX();
     const auto kh                     = convCommon->kernelY();
@@ -68,7 +69,7 @@ Execution* OneDNNConvInt8::create(Backend* backend, const MNN::Convolution2D* co
     }
     std::shared_ptr<ConvolutionCommon::Int8Common> quanCommon;
     if (convParam->quanParameter() != nullptr) {
-        quanCommon = ConvolutionCommon::load(convParam, backend, false);
+        quanCommon = ConvolutionCommon::load(op, backend, false);
         weightSrc = quanCommon->weight.get();
     }
     auto user_weights = memory(user_weights_md, eng, (int8_t*)weightSrc);

source/backend/cpu/OneDNNConvInt8.hpp

@@ -20,7 +20,7 @@ public:
         primitive_attr conv_attr;
         engine eng;
     };
-    static Execution* create(Backend *backend, const MNN::Convolution2D *convOp, const std::vector<Tensor *> &inputs, const std::vector<Tensor *> &outputs);
+    static Execution* create(Backend *backend, const MNN::Op* op, const std::vector<Tensor *> &inputs, const std::vector<Tensor *> &outputs);
     OneDNNConvInt8(std::shared_ptr<OneDNNConvInt8::Resource> resource, const MNN::Convolution2DCommon* common, Backend* bn);
     virtual ~OneDNNConvInt8();
     virtual ErrorCode onResize(const std::vector<Tensor *> &inputs, const std::vector<Tensor *> &outputs) override;