Merge pull request #33 from PaddlePaddle/develop

Sync from PaddlePaddle/Serving repo

paddle_inference/paddle/include/paddle_engine.h

@@ -104,7 +104,22 @@ class PaddleInferenceEngine : public PaddleEngineBase {
 
     Config config;
     // todo, auto config(zhangjun)
-    if (engine_conf.has_combined_model()) {
+    if (engine_conf.has_encrypted_model() && engine_conf.encrypted_model()) {
+      // decrypt model
+      std::string model_buffer, params_buffer, key_buffer;
+      predictor::ReadBinaryFile(model_path + "encrypt_model", &model_buffer);
+      predictor::ReadBinaryFile(model_path + "encrypt_params", &params_buffer);
+      predictor::ReadBinaryFile(model_path + "key", &key_buffer);
+
+      auto cipher = paddle::MakeCipher("");
+      std::string real_model_buffer = cipher->Decrypt(model_buffer, key_buffer);
+      std::string real_params_buffer =
+          cipher->Decrypt(params_buffer, key_buffer);
+      config.SetModelBuffer(&real_model_buffer[0],
+                            real_model_buffer.size(),
+                            &real_params_buffer[0],
+                            real_params_buffer.size());
+    } else if (engine_conf.has_combined_model()) {
       if (!engine_conf.combined_model()) {
         config.SetModel(model_path);
       } else {
@@ -156,22 +171,6 @@ class PaddleInferenceEngine : public PaddleEngineBase {
       config.EnableMemoryOptim();
     }
 
-    if (engine_conf.has_encrypted_model() && engine_conf.encrypted_model()) {
-      // decrypt model
-      std::string model_buffer, params_buffer, key_buffer;
-      predictor::ReadBinaryFile(model_path + "encrypt_model", &model_buffer);
-      predictor::ReadBinaryFile(model_path + "encrypt_params", &params_buffer);
-      predictor::ReadBinaryFile(model_path + "key", &key_buffer);
-
-      auto cipher = paddle::MakeCipher("");
-      std::string real_model_buffer = cipher->Decrypt(model_buffer, key_buffer);
-      std::string real_params_buffer =
-          cipher->Decrypt(params_buffer, key_buffer);
-      config.SetModelBuffer(&real_model_buffer[0],
-                            real_model_buffer.size(),
-                            &real_params_buffer[0],
-                            real_params_buffer.size());
-    }
 
     predictor::AutoLock lock(predictor::GlobalCreateMutex::instance());
     _predictor = CreatePredictor(config);

python/examples/ocr/ocr_debugger_server.py

@@ -107,7 +107,7 @@ ocr_service.prepare_server(workdir="workdir", port=9292)
 ocr_service.init_det_debugger(det_model_config="ocr_det_model")
 if sys.argv[1] == 'gpu':
     ocr_service.set_gpus("2")
-    ocr_service.run_debugger_service()
+    ocr_service.run_debugger_service(gpu = True)
 elif sys.argv[1] == 'cpu':
     ocr_service.run_debugger_service()
 ocr_service.run_web_service()

python/examples/xpu/fit_a_line_xpu/README.md

@@ -13,7 +13,10 @@ sh get_data.sh
 ## RPC service
 
 ### Start server
-
+``` shell
+python test_server.py uci_housing_model/
+```
+You can alse use the following code to start the RPC service 
 ```shell
 python -m paddle_serving_server.serve --model uci_housing_model --thread 10 --port 9393 --use_lite --use_xpu --ir_optim
 ```

python/examples/xpu/resnet_v2_50_xpu/README.md

@@ -18,5 +18,5 @@ python -m paddle_serving_server.serve --model resnet_v2_50_imagenet_model --port
 ### Client Prediction
 
 ```
-python resnet50_v2_client.py
+python resnet50_client.py
 ```

python/examples/xpu/resnet_v2_50_xpu/README_CN.md

@@ -18,5 +18,5 @@ python -m paddle_serving_server.serve --model resnet_v2_50_imagenet_model --port
 ### 客户端预测
 
 ```
-python resnet50_v2_client.py
+python resnet50_client.py
 ```

python/paddle_serving_server/web_service.py

@@ -266,8 +266,12 @@ class WebService(object):
     def _launch_local_predictor(self, gpu):
         from paddle_serving_app.local_predict import LocalPredictor
         self.client = LocalPredictor()
-        self.client.load_model_config(
-            "{}".format(self.model_config), use_gpu=True, gpu_id=self.gpus[0])
+        if gpu:
+            self.client.load_model_config(
+                "{}".format(self.model_config), use_gpu=True, gpu_id=self.gpus[0])
+        else:
+            self.client.load_model_config(
+                "{}".format(self.model_config), use_gpu=False)
 
     def run_web_service(self):
         print("This API will be deprecated later. Please do not use it")