added check for availability of GNA HW in QueueInference (#18549)

src/plugins/intel_gna/src/gna_device.cpp

@@ -583,5 +583,10 @@ uint32_t GNADeviceHelper::retrieveMaxLayersCount() {
         return Limitations::kMaxLayersCountGNA3_X;
     }
 }
+
+bool GNADeviceHelper::isHwAvailable() {
+    return target->get_detected_device_version() != DeviceVersion::SoftwareEmulation &&
+           target->get_detected_device_version() != DeviceVersion::NotSet;
+}
 }  // namespace intel_gna
 }  // namespace ov

src/plugins/intel_gna/src/gna_device.hpp

@@ -123,6 +123,7 @@ public:
     void getGnaPerfCounters(std::map<std::string, InferenceEngine::InferenceEngineProfileInfo>& retPerfCounters);
     static std::string GetGnaLibraryVersion();
 
+    bool isHwAvailable();
     const GnaAllocations& getAllAllocations() const {
         return allAllocations;
     }

src/plugins/intel_gna/src/gna_plugin.cpp

@@ -796,6 +796,10 @@ void GNAPlugin::DumpXNNToFile() const {
 }
 
 uint32_t GNAPlugin::QueueInference(const InferenceEngine::BlobMap& inputs, InferenceEngine::BlobMap& result) {
+    if (config.GetParameter(ov::intel_gna::execution_mode.name()).as<std::string>() == "GNA_HW" &&
+        !gnadevice->isHwAvailable()) {
+        THROW_GNA_EXCEPTION << "Execution mode GNA_HW is set, but hardware acceleration is unavailable";
+    }
     auto freeWorker = requestWorkerPool_->findFreeModelWorker();
     if (freeWorker == nullptr) {
         if (!m_graph_compiler->memory_connection.empty()) {