Skip to content

Commit 5be8629

Browse files
TeslaZhaoTeslaZhao
authored
Merge pull request #1796 from TeslaZhao/v0.9.0
Update doc
2 parents 080bc85 + 300ec10 commit 5be8629

File tree

2 files changed

+58
-39
lines changed

2 files changed

+58
-39
lines changed

‎doc/Install_CN.md‎

Lines changed: 32 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -15,20 +15,24 @@
1515

1616
**提示-1**：本项目仅支持<mark>**Python3.6/3.7/3.8/3.9**</mark>，接下来所有的与Python/Pip相关的操作都需要选择正确的Python版本。
1717

18-
**提示-2**：以下示例中GPU环境均为cuda11.2-cudnn8，如果您使用Python Pipeline来部署，并需要Nvidia TensorRT来优化预测性能，请参考[支持的镜像环境和说明](#4支持的镜像环境和说明)来选择其他版本。
18+
**提示-2**：以下示例中GPU环境均为cuda11.2-cudnn8，如果您使用Python Pipeline来部署，并需要Nvidia TensorRT来优化预测性能，请参考以下说明来选择其他版本。
1919

2020
<a name="1"></a>
2121

2222
## 1.使用开发镜像
2323

24-
<mark>**同时支持使用Serving镜像和Paddle镜像，1.1和1.2章节中的操作2选1即可。**</mark> 在Paddle docker镜像上部署Serving服务需要安装额外依赖库，因此，我们直接使用Serving开发镜像。
24+
- Serving 镜像: registry.baidubce.com/paddlepaddle/serving:{Tag}
25+
- Paddle 镜像: registry.baidubce.com/paddlepaddle/paddle:{Tag}
2526

26-
| 环境 | Serving开发镜像Tag | 操作系统 | Paddle开发镜像Tag | 操作系统 |
27+
<mark>**同时支持使用 Serving 镜像和 Paddle 镜像，`1.1` 和 `1.2` 章节中的操作2选1即可。**</mark> 在Paddle docker镜像上部署 Servin g服务需要安装额外依赖库，因此，我们直接使用 Serving 开发镜像。
28+
29+
| 环境 | Serving镜像 Tag | 操作系统 | Paddle镜像 Tag | 操作系统 |
2730
| :--------------------------: | :-------------------------------: | :-------------: | :-------------------: | :----------------: |
28-
| CPU | 0.9.0-devel | Ubuntu 16.04 | 2.3.0 | Ubuntu 18.04. |
29-
| CUDA10.1 + CUDNN7 | 0.9.0-cuda10.1-cudnn7-devel | Ubuntu 16.04 | 无 | 无 |
30-
| CUDA10.2 + CUDNN8 | 0.9.0-cuda10.2-cudnn8-devel | Ubuntu 16.04 | 无 | Ubuntu 18.04 |
31-
| CUDA11.2 + CUDNN8 | 0.9.0-cuda11.2-cudnn8-devel | Ubuntu 16.04 | 2.3.0-gpu-cuda11.2-cudnn8 | Ubuntu 18.04 |
31+
| CPU | 0.9.0-devel | Ubuntu 16 | 2.3.0 | Ubuntu 18 |
32+
| CUDA10.1 + cuDNN 7 | 0.9.0-cuda10.1-cudnn7-devel | Ubuntu 16 | 无 | 无 |
33+
| CUDA10.2 + cuDNN 7 | 0.9.0-cuda10.2-cudnn7-devel | Ubuntu 16 | 2.3.0-gpu-cuda10.2-cudnn7 | Ubuntu 18
34+
| CUDA10.2 + cuDNN 8 | 0.9.0-cuda10.2-cudnn8-devel | Ubuntu 16 | 无 | Ubuntu 18 |
35+
| CUDA11.2 + cuDNN 8 | 0.9.0-cuda11.2-cudnn8-devel | Ubuntu 16 | 2.3.0-gpu-cuda11.2-cudnn8 | Ubuntu 18 |
3236

3337
对于**Windows 10 用户**，请参考文档[Windows平台使用Paddle Serving指导](Windows_Tutorial_CN.md)。
3438

@@ -157,16 +161,23 @@ pip3 install https://paddle-inference-lib.bj.bcebos.com/2.3.0/python/Linux/GPU/x
157161

158162
**1.安装离线 Wheel 包**
159163

160-
Serving 和 Paddle Wheel包的离线依赖包下载在 `py3x_offline_whls` 目录下 `serving_dependent_wheels/` 和 `paddle_dependent_wheels/`。
164+
Serving 和 Paddle Wheel包的离线依赖包下载有4个链接。
165+
166+
```
167+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py36_offline_whl_packages.tar
168+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py37_offline_whl_packages.tar
169+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py38_offline_whl_packages.tar
170+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py39_offline_whl_packages.tar
171+
```
161172

162173
通过运行 `install.py` 脚本可本地安装 Serving 和 Paddle Wheel 包。`install.py` 脚本的参数列表如下：
163174
```
164175
python3 install.py
165176
--python_version : Python version for installing wheels, one of [py36, py37, py38, py39], py37 default.
166177
--device : Type of devices, one of [cpu, gpu], cpu default.
167178
--cuda_version : CUDA version for GPU, one of [101, 102, 112, empty], empty default.
168-
--serving_version : Verson of Serving, one of [0.8.3, no_install], 0.8.3 default.
169-
--paddle_version Verson of Paddle, one of [2.2.2, no_install], 2.2.2 default.
179+
--serving_version : Verson of Serving, one of [0.9.0, no_install], 0.9.0 default.
180+
--paddle_version Verson of Paddle, one of [2.3.0, no_install], 2.3.0 default.
170181
```
171182

172183
**2.在环境变量中指定 `SERVING_BIN` 路径**
@@ -182,34 +193,34 @@ python3 -m paddle_serving_server.serve --model serving_model --thread 10 --port
182193

183194
由于所有版本的二进制程序包有 20 GB，非常大。因此提供多个版本的下载链接，通过手动 `wget` 下载指定版本到 `serving_bin` 目录下，解压后导出到环境变量中。
184195

185-
- cpu-avx-mkl: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-mkl-0.8.3.tar.gz
186-
- cpu-avx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-openblas-0.8.3.tar.gz
187-
- cpu-noavx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-noavx-openblas-0.8.3.tar.gz
188-
- cuda10.1-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-101-0.8.3.tar.gz
189-
- cuda10.2-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-102-0.8.3.tar.gz
190-
- cuda10.2-cudnn8-TensorRT7: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-1028-0.8.3.tar.gz
191-
- cuda11.2-cudnn8-TensorRT8: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-112-0.8.3.tar.gz
196+
- cpu-avx-mkl: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-mkl-0.9.0.tar.gz
197+
- cpu-avx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-openblas-0.9.0.tar.gz
198+
- cpu-noavx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-noavx-openblas-0.9.0.tar.gz
199+
- cuda10.1-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-101-0.9.0.tar.gz
200+
- cuda10.2-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-102-0.9.0.tar.gz
201+
- cuda10.2-cudnn8-TensorRT7: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-1028-0.9.0.tar.gz
202+
- cuda11.2-cudnn8-TensorRT8: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-112-0.9.0.tar.gz
192203

193204
以 GPU CUDA 10.2 为例，在命令行或启动程序中设置环境变量如下：
194205
```
195-
export SERVING_BIN = $PWD/serving_bin/serving-gpu-102-0.8.3/serving
206+
export SERVING_BIN = $PWD/serving_bin/serving-gpu-102-0.9.0/serving
196207
```
197208

198209
**3.运行 `install.py` 安装 Wheel 包**
199210

200211
1.同时安装 Serving 和 Paddle 的 py38 版本 GPU wheel 包：
201212
```
202-
python3 install.py --cuda_version="102" --python_version="py38" --device="GPU" --serving_version="0.8.3" --paddle_version="2.2.2"
213+
python3 install.py --cuda_version="102" --python_version="py38" --device="GPU" --serving_version="0.9.0" --paddle_version="2.3.0"
203214
```
204215

205216
2.仅安装 Serving 的 py39 版本 CPU wheel 包，设置 `--paddle_version="no_install"` 表示不安装 Paddle 预测库，设置 `--device="cpu"` 表示 cpu 版本
206217
```
207-
python3 install.py --cuda_version="" --python_version="py39" --device="cpu" --serving_version="0.8.3" --paddle_version="no_install"
218+
python3 install.py --cuda_version="" --python_version="py39" --device="cpu" --serving_version="0.9.0" --paddle_version="no_install"
208219
```
209220

210221
3.仅安装 Paddle 的 py36 版本`cuda=11.2` 的 GPU wheel 包，
211222
```
212-
python3 install.py --cuda_version="112" --python_version="py36" --device="GPU" --serving_version="no_install" --paddle_version="2.2.2"
223+
python3 install.py --cuda_version="112" --python_version="py36" --device="GPU" --serving_version="no_install" --paddle_version="2.3.0"
213224
```
214225

215226
<a name="3"></a>

‎doc/Install_EN.md‎

Lines changed: 26 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -23,10 +23,11 @@
2323

2424
| Environment | Serving Development Image Tag | Operating System | Paddle Development Image Tag | Operating System |
2525
| :--------------------------: | :-------------------------------: | :-------------: | :-------------------: | :----------------: |
26-
| CPU | 0.9.0-devel | Ubuntu 16.04 | 2.3.0 | Ubuntu 18.04. |
27-
| CUDA10.1 + CUDNN7 | 0.9.0-cuda10.1-cudnn7-devel | Ubuntu 16.04 | 无 | 无 |
28-
| CUDA10.2 + CUDNN8 | 0.9.0-cuda10.2-cudnn8-devel | Ubuntu 16.04 | 无 | Ubuntu 18.04 |
29-
| CUDA11.2 + CUDNN8 | 0.9.0-cuda11.2-cudnn8-devel | Ubuntu 16.04 | 2.3.0-gpu-cuda11.2-cudnn8 | Ubuntu 18.04 |
26+
| CPU | 0.9.0-devel | Ubuntu 16 | 2.3.0 | Ubuntu 18 |
27+
| CUDA10.1 + cuDNN 7 | 0.9.0-cuda10.1-cudnn7-devel | Ubuntu 16 | None | None |
28+
| CUDA10.2 + cuDNN 7 | 0.9.0-cuda10.2-cudnn7-devel | Ubuntu 16 | 2.3.0-gpu-cuda10.2-cudnn7 | Ubuntu 18
29+
| CUDA10.2 + cuDNN 8 | 0.9.0-cuda10.2-cudnn8-devel | Ubuntu 16 | None | None |
30+
| CUDA11.2 + cuDNN 8 | 0.9.0-cuda11.2-cudnn8-devel | Ubuntu 16 | 2.3.0-gpu-cuda11.2-cudnn8 | Ubuntu 18 |
3031

3132
For **Windows 10 users**, please refer to the document [Paddle Serving Guide for Windows Platform](Windows_Tutorial_CN.md).
3233

@@ -91,7 +92,7 @@ pip3 install -r python/requirements.txt
9192
Install the service whl package. There are three types of client, app and server. The server is divided into CPU and GPU. Choose one installation according to the environment.
9293
- post112 = CUDA11.2 + cuDNN8 + TensorRT8（Recommanded）
9394
- post101 = CUDA10.1 + cuDNN7 + TensorRT6
94-
- post102 = CUDA10.2 + cuDNN7 + TensorRT6 (与Paddle 镜像一致
95+
- post102 = CUDA10.2 + cuDNN7 + TensorRT6 (The same with paddle docker images)
9596
- post1028 = CUDA10.2 + cuDNN8 + TensorRT7
9697

9798
<a name="2.1"></a>
@@ -154,16 +155,23 @@ pip3 install https://paddle-inference-lib.bj.bcebos.com/2.3.0/python/Linux/GPU/x
154155

155156
**1.Install offline wheel packages**
156157

157-
The independent dependencies of the Serving and Paddle Wheel packages are downloaded in `serving_dependent_wheels/` and `paddle_dependent_wheels/` under the `py3x_offline_whls` directory.
158+
The independent dependencies of the Serving and Paddle Wheel packages are downloaded by four links.
159+
160+
```
161+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py36_offline_whl_packages.tar
162+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py37_offline_whl_packages.tar
163+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py38_offline_whl_packages.tar
164+
wget https://paddle-serving.bj.bcebos.com/offline_wheels/0.9.0/py39_offline_whl_packages.tar
165+
```
158166

159167
The Serving and Paddle Wheel packages can be installed locally by running the `install.py` script. The parameter list for the `install.py` script is as follows:
160168
```
161169
python3 install.py
162170
--python_version : Python version for installing wheels, one of [py36, py37, py38, py39], py37 default.
163171
--device : Type of devices, one of [cpu, gpu], cpu default.
164172
--cuda_version : CUDA version for GPU, one of [101, 102, 112, empty], empty default.
165-
--serving_version : Verson of Serving, one of [0.8.3, no_install], 0.8.3 default.
166-
--paddle_version Verson of Paddle, one of [2.2.2, no_install], 2.2.2 default.
173+
--serving_version : Verson of Serving, one of [0.9.0, no_install], 0.9.0 default.
174+
--paddle_version Verson of Paddle, one of [2.3.0, no_install], 2.3.0 default.
167175
```
168176

169177
**2.Specify the `SERVING_BIN` path in the environment variable**
@@ -176,30 +184,30 @@ python3 -m paddle_serving_server.serve --model serving_model --thread 10 --port
176184
```
177185
Since the binary package for all versions has 20 GB, it is very large. Therefore, multiple versions of download links are provided. Manually `wget` downloads the specified version to the `serving_bin` directory, decompresses it and exports it to the environment variable.
178186

179-
- cpu-avx-mkl: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-mkl-0.8.3.tar.gz
180-
- cpu-avx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-openblas-0.8.3.tar.gz
181-
- cpu-noavx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-noavx-openblas-0.8.3.tar.gz
182-
- cuda10.1-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-101-0.8.3.tar.gz
183-
- cuda10.2-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-102-0.8.3.tar.gz
184-
- cuda10.2-cudnn8-TensorRT7: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-1028-0.8.3.tar.gz
185-
- cuda11.2-cudnn8-TensorRT8: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-112-0.8.3.tar.gz
187+
- cpu-avx-mkl: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-mkl-0.9.0.tar.gz
188+
- cpu-avx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-avx-openblas-0.9.0.tar.gz
189+
- cpu-noavx-openblas: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-cpu-noavx-openblas-0.9.0.tar.gz
190+
- cuda10.1-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-101-0.9.0.tar.gz
191+
- cuda10.2-cudnn7-TensorRT6: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-102-0.9.0.tar.gz
192+
- cuda10.2-cudnn8-TensorRT7: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-1028-0.9.0.tar.gz
193+
- cuda11.2-cudnn8-TensorRT8: https://paddle-serving.bj.bcebos.com/test-dev/bin/serving-gpu-112-0.9.0.tar.gz
186194

187195
Taking GPU CUDA 10.2 as an example, set the environment variables on the command line or in the launcher as follows:
188196

189197
**3.Run `install.py` to install wheel packages**
190198

191199
1. Install the py38 version GPU wheel package of Serving and Paddle at the same time:
192200
```
193-
python3 install.py --cuda_version="102" --python_version="py38" --device="GPU" --serving_version="0.8.3" --paddle_version="2.2.2"
201+
python3 install.py --cuda_version="102" --python_version="py38" --device="GPU" --serving_version="0.9.0" --paddle_version="2.3.0"
194202
```
195203

196204
2.Only install the py39 version of the Serving CPU wheel package, set `--paddle_version="no_install"` to not install the Paddle prediction library, set `--device="cpu"` to indicate the cpu version
197205
```
198-
python3 install.py --cuda_version="" --python_version="py39" --device="cpu" --serving_version="0.8.3" --paddle_version="no_install"
206+
python3 install.py --cuda_version="" --python_version="py39" --device="cpu" --serving_version="0.9.0" --paddle_version="no_install"
199207
```
200208
3. Install only the GPU wheel package of Paddle's py36 version `cuda=11.2`
201209
```
202-
python3 install.py --cuda_version="112" --python_version="py36" --device="GPU" --serving_version="no_install" --paddle_version="2.2.2"
210+
python3 install.py --cuda_version="112" --python_version="py36" --device="GPU" --serving_version="no_install" --paddle_version="2.3.0"
203211
```
204212

205213
<a name="3"></a>

0 commit comments

Comments
 (0)