diff --git a/tutorials/mobilenetv3_prod/Step6/test_tipc/configs/mobilenet_v3_small/serving_infer_cpp.txt b/tutorials/mobilenetv3_prod/Step6/test_tipc/configs/mobilenet_v3_small/serving_infer_cpp.txt
new file mode 100644
index 0000000000000000000000000000000000000000..a0227e09f7e713a3fc27065a436d0cc398cede7f
--- /dev/null
+++ b/tutorials/mobilenetv3_prod/Step6/test_tipc/configs/mobilenet_v3_small/serving_infer_cpp.txt
@@ -0,0 +1,13 @@
+===========================serving_params===========================
+model_name:mobilenet_v3_small
+python:python3.7
+trans_model:-m paddle_serving_client.convert
+--dirname:./inference/mobilenet_v3_small_infer/
+--model_filename:inference.pdmodel
+--params_filename:inference.pdiparams
+--serving_server:./deploy/serving_python/serving_server/
+--serving_client:./deploy/serving_python/serving_client/
+serving_dir:./deploy/serving_python
+--model:./deploy/serving_python/serving_server/
+--port:9993
+cpp_client:serving_client.py
\ No newline at end of file
diff --git a/tutorials/mobilenetv3_prod/Step6/test_tipc/test_serving_infer_cpp.sh b/tutorials/mobilenetv3_prod/Step6/test_tipc/test_serving_infer_cpp.sh
new file mode 100644
index 0000000000000000000000000000000000000000..b449fcf65d4cbc3ef4cfb062b3d817f78eec8de5
--- /dev/null
+++ b/tutorials/mobilenetv3_prod/Step6/test_tipc/test_serving_infer_cpp.sh
@@ -0,0 +1,90 @@
+#!/bin/bash
+source test_tipc/common_func.sh
+
+FILENAME=$1
+dataline=$(awk 'NR==1, NR==18{print}' $FILENAME)
+MODE=$2
+
+# parser params
+IFS=$'\n'
+lines=(${dataline})
+
+# parser serving
+model_name=$(func_parser_value "${lines[1]}")
+python_list=$(func_parser_value "${lines[2]}")
+trans_model_py=$(func_parser_value "${lines[3]}")
+infer_model_dir_key=$(func_parser_key "${lines[4]}")
+infer_model_dir_value=$(func_parser_value "${lines[4]}")
+model_filename_key=$(func_parser_key "${lines[5]}")
+model_filename_value=$(func_parser_value "${lines[5]}")
+params_filename_key=$(func_parser_key "${lines[6]}")
+params_filename_value=$(func_parser_value "${lines[6]}")
+serving_server_key=$(func_parser_key "${lines[7]}")
+serving_server_value=$(func_parser_value "${lines[7]}")
+serving_client_key=$(func_parser_key "${lines[8]}")
+serving_client_value=$(func_parser_value "${lines[8]}")
+serving_dir_value=$(func_parser_value "${lines[9]}")
+run_model_path_key=$(func_parser_value "${lines[10]}")
+run_model_path_value=$(func_parser_value "${lines[11]}")
+port_key=$(func_parser_value "${lines[12]}")
+port_value=$(func_parser_key "${lines[13]}")
+cpp_client_value=$(func_parser_value "${lines[14]}")
+
+
+LOG_PATH="./log/${model_name}/${MODE}"
+mkdir -p ${LOG_PATH}
+status_log="${LOG_PATH}/results_serving.log"
+
+function func_serving(){
+ IFS='|'
+ _python=$1
+ _script=$2
+ _model_dir=$3
+
+ # phrase 1: save model
+ set_dirname=$(func_set_params "${infer_model_dir_key}" "${infer_model_dir_value}")
+ set_model_filename=$(func_set_params "${model_filename_key}" "${model_filename_value}")
+ set_params_filename=$(func_set_params "${params_filename_key}" "${params_filename_value}")
+ set_serving_server=$(func_set_params "${serving_server_key}" "${serving_server_value}")
+ set_serving_client=$(func_set_params "${serving_client_key}" "${serving_client_value}")
+ python_list=(${python_list})
+ python=${python_list[0]}
+ trans_model_cmd="${python} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
+ eval $trans_model_cmd}
+ cd ${serving_dir_value}
+ echo $PWD
+ unset https_proxy
+ unset http_proxy
+
+ web_service_cmd="${python} ${web_service_py} &"
+ eval $web_service_cmd
+ sleep 2s
+ _save_log_path="../../log/${model_name}/${MODE}/server_infer_gpu_batchsize_1.log"
+
+ # phrase 2: run server
+ cpp_server_cmd="${python} -m paddle_serving_server.serve ${run_model_path_key} ${run_model_path_value} ${port_key} ${port_value} > ${_save_log_path} 2>&1 "
+ eval $cpp_server_cmd
+ last_status=${PIPESTATUS[0]}
+ eval "cat ${_save_log_path}"
+ cd ../../
+ status_check $last_status "${cpp_server_cmd}" "${status_log}"
+ ps ux | grep -i 'paddle_serving_server' | awk '{print $2}' | xargs kill -s 9
+}
+
+
+# set cuda device
+GPUID=$3
+if [ ${#GPUID} -le 0 ];then
+ env=" "
+else
+ env="export CUDA_VISIBLE_DEVICES=${GPUID}"
+fi
+set CUDA_VISIBLE_DEVICES
+eval $env
+
+
+echo "################### run test ###################"
+
+export Count=0
+IFS="|"
+func_serving "${web_service_cmd}"
diff --git a/tutorials/tipc/serving_cpp/README.md b/tutorials/tipc/serving_cpp/README.md
index e1ea698c26f8ee1703175d8181f1f884b9596a1c..81640e76970cc5e9aa45c127ef96632024bed075 100644
--- a/tutorials/tipc/serving_cpp/README.md
+++ b/tutorials/tipc/serving_cpp/README.md
@@ -9,3 +9,81 @@
- [3. 服务化部署测试开发与规范](#3)
- [3.1 开发流程](#3.1)
- [3.2 核验点](#3.2)
+
+
+
+## 1.简介
+该系列文档主要介绍 Linux GPU/CPU C++ 服务化部署开发过程，主要包含2个步骤。
+- 步骤一：参考[《Linux GPU/CPU C++ 服务化部署功能开发文档》](./serving_cpp.md)，完成Linux GPU/CPU C++ 服务化部署功能开发。
+- 步骤二：参考[《Linux GPU/CPU C++ 服务化部署功能测试开发文档》](./test_serving_cpp.md)，完成Linux GPU/CPU C++ 服务化部署功能测试开发。
+
+
+
+## 2.服务化部署流程开发规范
+
+
+
+### 2.1 开发流程
+
+Linux GPU/CPU C++ 服务化部署功能开发过程以分为下面8个步骤。
+
+
+
+
+
+
+模型转换配置参数（点击以展开详细内容或者折叠）
+
+| 行号 | 参考内容 | 含义 | key是否需要修改 | value是否需要修改 | 修改内容 |
+|----|-------------------------------------|---------------|-----------|-------------|----------------------------------|
+| 2 | model_name:mobilenet_v3_small | 模型名字 | 否 | 是 | value修改为自己的模型名字 |
+| 3 | python:python3.7 | python环境 | 否 | 是 | value修改为自己的python环境 |
+| 5 | --dirname:./inference/mobilenet_v3_small_infer/ | Paddle inference 模型保存路径 | 否 | 是 | value修改为自己 Inference 模型的路径 |
+| 6 | --model_filename:inference.pdmodel | pdmodel 文件名 | 否 | 是 | value修改为 pdmodel 文件名 |
+| 7 | --params_filename:inference.pdiparams | pdiparams 文件名 | 否 | 是 | value修改为 pdiparams 文件名 |
+| 8 | --serving_server:./deploy/serving_cpp/serving_server/ | 转换出的部署模型目录 | 否 | 是 | value修改为部署模型模型保存路径 |
+| 9 | --serving_client:./deploy/serving_cpp/serving_client/ | 转换出的服务模型目录 | 否 | 是 | value修改为服务模型保存路径 |
+
+
+
+以模型转换命令为例，总共包含5个超参数。
+```python
+python3.7 -m paddle_serving_client.convert --dirname=./inference/resnet50_infer/ --model_filename=inference.pdmodel --params_filename=inference.pdiparams --serving_server=./deploy/serving_cpp/serving_server/ --serving_client=./deploy/serving_cpp/serving_client/
+```
+
+- 推理模型路径：`--dirname=./inference/resnet50_infer/`，则需要修改第5行
+- pdmodel文件名：`--model_filename=inference.pdmodel`，则需要修改第6行
+- 其他参数以此类推
+
+### 2.2.2 C++ 服务部署配置参数
+
+C++ 服务的客户端采用 PYTHON 语言编写。
+```python
+python3.7 serving_client.py
+```
+
+
+
+# 3. C++ 服务化部署功能测试开发
+
+服务化部署功能测试开发主要分为以下6个步骤。
+
+
+
+