Skip to content

Commit bdb6b3a

Browse files
Update 3.0.0b0 (#6692)
* Update 3.0.0b0 * Update Tables * Fix * Fix nccl * Fix en pip * Fix
1 parent a25d60f commit bdb6b3a

18 files changed

+271
-1038
lines changed

docs/install/Tables.md

Lines changed: 72 additions & 317 deletions
Large diffs are not rendered by default.

docs/install/Tables_en.md

Lines changed: 72 additions & 317 deletions
Large diffs are not rendered by default.

docs/install/conda/linux-conda.md

Lines changed: 6 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -86,39 +86,29 @@ python3 -c "import platform;print(platform.architecture()[0]);print(platform.mac
8686

8787
#### CPU 版的 PaddlePaddle
8888

89+
8990
如果您的计算机没有 NVIDIA® GPU,请安装 CPU 版的 PaddlePaddle
9091

9192
```
92-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
93+
conda install paddlepaddle==3.0.0b0 -c paddle
9394
```
9495

9596

96-
9797
#### GPU 版的 PaddlePaddle
9898

9999

100-
101-
* 对于 `CUDA 11.2`,需要搭配 cuDNN 8.2.1(多卡环境下 NCCL>=2.7),安装命令为:
100+
* 对于 `CUDA 11.8` 安装命令为:
102101

103102
```
104-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.2 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
103+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=11.8 -c paddle -c nvidia
105104
```
106105

107-
* 对于 `CUDA 11.6`,需要搭配 cuDNN 8.4.0(多卡环境下 NCCL>=2.7),安装命令为:
106+
* 对于 `CUDA 12.3` 安装命令为:
108107

109108
```
110-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.6 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
109+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=12.3 -c paddle -c nvidia
111110
```
112111

113-
* 对于 `CUDA 11.7`,需要搭配 cuDNN 8.4.1(多卡环境下 NCCL>=2.7),安装命令为:
114-
115-
```
116-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.7 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
117-
```
118-
119-
您可参考 NVIDIA 官方文档了解 CUDA 和 CUDNN 的安装流程和配置方法,请见[CUDA](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)[cuDNN](https://docs.nvidia.com/deeplearning/sdk/cudnn-install/)
120-
121-
122112

123113
## **三、验证安装**
124114

docs/install/conda/linux-conda_en.md

Lines changed: 6 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -91,37 +91,29 @@ You can choose the following version of PaddlePaddle to start installation:
9191

9292
#### CPU Version of PaddlePaddle
9393

94+
9495
If your computer doesn't have NVIDIA® GPU, please install `the CPU Version of PaddlePaddle`
9596

9697
```
97-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
98+
conda install paddlepaddle==3.0.0b0 -c paddle
9899
```
99100

100101

101-
102102
#### GPU Version of PaddlePaddle
103103

104104

105-
* If you are using CUDA 11.2,cuDNN 8.2.1(for multi card support, NCCL>=2.7):
105+
* If you are using CUDA 11.8:
106106

107107
```
108-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.2 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
108+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=11.8 -c paddle -c nvidia
109109
```
110110

111-
* If you are using CUDA 11.6,cuDNN 8.4.0(for multi card support, NCCL>=2.7):
111+
* If you are using CUDA 12.3:
112112

113113
```
114-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.6 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
114+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=12.3 -c paddle -c nvidia
115115
```
116116

117-
* If you are using CUDA 11.7,cuDNN 8.4.1(for multi card support, NCCL>=2.7):
118-
119-
```
120-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.7 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
121-
```
122-
123-
You can refer to NVIDIA official documents for installation process and configuration method of CUDA and cudnn. Please refer to [CUDA](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)[cuDNN](https://docs.nvidia.com/deeplearning/sdk/cudnn-install/)
124-
125117

126118
## Verify installation
127119

docs/install/conda/macos-conda.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -83,7 +83,7 @@ python3 -c "import platform;print(platform.architecture()[0]);print(platform.mac
8383
* 目前在 macOS 环境仅支持 CPU 版 PaddlePaddle,请参考如下命令安装 Paddle:
8484

8585
```
86-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
86+
conda install paddlepaddle==3.0.0b0 -c paddle
8787
```
8888

8989
## **三、验证安装**

docs/install/conda/macos-conda_en.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -87,7 +87,7 @@ conda config --set show_channel_urls yes
8787
* Currently, only the CPU version of PaddlePaddle is supported in the macOS environment. Please use the following command to install PaddlePaddle:
8888

8989
```
90-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
90+
conda install paddlepaddle==3.0.0b0 -c paddle
9191
```
9292

9393

docs/install/conda/windows-conda.md

Lines changed: 6 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -90,35 +90,27 @@ python -c "import platform;print(platform.architecture()[0]);print(platform.mach
9090

9191
如果您的计算机没有 NVIDIA® GPU,请安装 CPU 版的 PaddlePaddle
9292

93+
9394
```
94-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
95+
conda install paddlepaddle==3.0.0b0 -c paddle
9596
```
9697

9798

98-
9999
#### GPU 版的 PaddlePaddle
100100

101101

102-
* 对于 `CUDA 11.2`,需要搭配 cuDNN 8.2.1,安装命令为:
102+
* 对于 `CUDA 11.8` 安装命令为:
103103

104104
```
105-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.2 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
105+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=11.8 -c paddle -c nvidia
106106
```
107107

108-
* 对于 `CUDA 11.6`,需要搭配 cuDNN 8.4.0,安装命令为:
108+
* 对于 `CUDA 12.3` 安装命令为:
109109

110110
```
111-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.6 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
111+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=12.3 -c paddle -c nvidia
112112
```
113113

114-
* 对于 `CUDA 11.7`,需要搭配 cuDNN 8.4.1,安装命令为:
115-
116-
```
117-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.7 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
118-
```
119-
120-
您可参考 NVIDIA 官方文档了解 CUDA 和 CUDNN 的安装流程和配置方法,请见[CUDA](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)[cuDNN](https://docs.nvidia.com/deeplearning/sdk/cudnn-install/)
121-
122114

123115
## **三、验证安装**
124116

docs/install/conda/windows-conda_en.md

Lines changed: 6 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -93,10 +93,11 @@ You can choose the following version of PaddlePaddle to start installation:
9393

9494
#### CPU Version of PaddlePaddle
9595

96+
9697
If your computer doesn't have NVIDIA® GPU, please install `the CPU Version of PaddlePaddle`
9798

9899
```
99-
conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/
100+
conda install paddlepaddle==3.0.0b0 -c paddle
100101
```
101102

102103

@@ -105,26 +106,18 @@ conda install paddlepaddle==2.6.1 --channel https://mirrors.tuna.tsinghua.edu.cn
105106
#### GPU Version of PaddlePaddle
106107

107108

108-
* If you are using CUDA 11.2,cuDNN 8.2.1:
109-
110-
```
111-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.2 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
112-
```
113-
114-
* If you are using CUDA 11.6,cuDNN 8.4.0:
109+
* If you are using CUDA 11.8:
115110

116111
```
117-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.6 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
112+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=11.8 -c paddle -c nvidia
118113
```
119114

120-
* If you are using CUDA 11.7,cuDNN 8.4.1:
115+
* If you are using CUDA 12.3:
121116

122117
```
123-
conda install paddlepaddle-gpu==2.6.1 cudatoolkit=11.7 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge
118+
conda install paddlepaddle-gpu==3.0.0b0 paddlepaddle-cuda=12.3 -c paddle -c nvidia
124119
```
125120

126-
You can refer to NVIDIA official documents for installation process and configuration method of CUDA and cudnn. Please refer to [CUDA](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)[cuDNN](https://docs.nvidia.com/deeplearning/sdk/cudnn-install/)
127-
128121

129122
## Verify installation
130123

docs/install/docker/linux-docker.md

Lines changed: 22 additions & 32 deletions
Original file line numberDiff line numberDiff line change
@@ -21,46 +21,40 @@
2121

2222
* CPU 版的 PaddlePaddle:
2323
```
24-
docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1
24+
docker pull registry.baidubce.com/paddlepaddle/paddle:3.0.0b0
2525
```
2626
2727
* CPU 版的 PaddlePaddle,且镜像中预装好了 jupyter:
2828
```
29-
docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1-jupyter
29+
docker pull registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-jupyter
3030
```
3131
3232
* GPU 版的 PaddlePaddle(**建议拉取最新版本镜像,并确保已经成功安装 NVIDIA Container Toolkit**):
3333
```
34-
docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda11.2-cudnn8.2-trt8.0
34+
docker pull registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda11.8-cudnn8.6-trt8.5
3535
```
3636
```
37-
docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda11.7-cudnn8.4-trt8.4
38-
```
39-
```
40-
docker pull registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda12.0-cudnn8.9-trt8.6
37+
docker pull registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6
4138
```
4239
4340
如果您的机器不在中国大陆地区,可以直接从 DockerHub 拉取镜像:
4441
4542
* CPU 版的 PaddlePaddle:
4643
```
47-
docker pull paddlepaddle/paddle:2.6.1
44+
docker pull paddlepaddle/paddle:3.0.0b0
4845
```
4946
5047
* CPU 版的 PaddlePaddle,且镜像中预装好了 jupyter:
5148
```
52-
docker pull paddlepaddle/paddle:2.6.1-jupyter
49+
docker pull paddlepaddle/paddle:3.0.0b0-jupyter
5350
```
5451
5552
* GPU 版的 PaddlePaddle(**建议拉取最新版本镜像,并确保已经成功安装 NVIDIA Container Toolkit**):
5653
```
57-
docker pull paddlepaddle/paddle:2.6.1-gpu-cuda11.2-cudnn8.2-trt8.0
58-
```
54+
registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda11.8-cudnn8.6-trt8.5
5955
```
60-
docker pull paddlepaddle/paddle:2.6.1-gpu-cuda11.7-cudnn8.4-trt8.4
6156
```
62-
```
63-
docker pull paddlepaddle/paddle:2.6.1-gpu-cuda12.0-cudnn8.9-trt8.6
57+
registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6
6458
```
6559
6660
您还可以访问[DockerHub](https://hub.docker.com/r/paddlepaddle/paddle/tags/)获取更多镜像。
@@ -72,7 +66,7 @@
7266
7367
7468
```
75-
docker run --name paddle_docker -it -v $PWD:/paddle registry.baidubce.com/paddlepaddle/paddle:2.6.1 /bin/bash
69+
docker run --name paddle_docker -it -v $PWD:/paddle registry.baidubce.com/paddlepaddle/paddle:3.0.0b0 /bin/bash
7670
```
7771
7872
- `--name paddle_docker`:设定 Docker 的名称,`paddle_docker` 是自己设置的名称;
@@ -83,7 +77,7 @@
8377
8478
- `-v $PWD:/paddle`:指定将当前路径(PWD 变量会展开为当前路径的绝对路径)挂载到容器内部的 /paddle 目录;
8579
86-
- `registry.baidubce.com/paddlepaddle/paddle:2.6.1`:指定需要使用的 image 名称,您可以通过`docker images`命令查看;/bin/bash 是在 Docker 中要执行的命令
80+
- `registry.baidubce.com/paddlepaddle/paddle:3.0.0b0`:指定需要使用的 image 名称,您可以通过`docker images`命令查看;/bin/bash 是在 Docker 中要执行的命令
8781
8882
8983
* 使用 CPU 版本的 PaddlePaddle,且镜像中预装好了 jupyter:
@@ -98,7 +92,7 @@
9892
cd ./jupyter_docker
9993
```
10094
```
101-
docker run -p 80:80 --rm --env USER_PASSWD="password you set" -v $PWD:/home/paddle registry.baidubce.com/paddlepaddle/paddle:2.6.1-jupyter
95+
docker run -p 80:80 --rm --env USER_PASSWD="password you set" -v $PWD:/home/paddle registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-jupyter
10296
```
10397
10498
- `--rm`:关闭容器后删除容器;
@@ -109,13 +103,13 @@
109103
110104
- `-v $PWD:/home/paddle`:指定将当前路径(PWD 变量会展开为当前路径的绝对路径)挂载到容器内部的 /home/paddle 目录;
111105
112-
- `registry.baidubce.com/paddlepaddle/paddle:2.6.1-jupyter`:指定需要使用的 image 名称,您可以通过`docker images`命令查看
106+
- `registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-jupyter`:指定需要使用的 image 名称,您可以通过`docker images`命令查看
113107
114108
115109
* 使用 GPU 版本的 PaddlePaddle:
116110
117111
```
118-
docker run --gpus all --name paddle_docker -v $PWD:/paddle --network=host -it registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2 /bin/bash
112+
docker run --gpus all --name paddle_docker -v $PWD:/paddle --network=host -it registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6 /bin/bash
119113
```
120114
121115
- `--gpus all`: 在 Docker 容器中允许使用 gpu;
@@ -127,7 +121,7 @@
127121
128122
- `-it`: 与宿主机保持交互状态;
129123
130-
- `registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2`:使用名为`registry.baidubce.com/paddlepaddle/paddle`, tag 为`latest-dev-cuda12.0-cudnn8.9-trt8.6-gcc12.2`的镜像创建 Docker 容器,/bin/bash 进入容器后启动/bin/bash 命令。
124+
- `registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6`:使用名为`registry.baidubce.com/paddlepaddle/paddle`, tag 为`3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6`的镜像创建 Docker 容器,/bin/bash 进入容器后启动/bin/bash 命令。
131125
132126
133127
@@ -146,24 +140,20 @@
146140
</thead>
147141
<tbody>
148142
<tr>
149-
<td> registry.baidubce.com/paddlepaddle/paddle:2.6.1 </td>
150-
<td> 安装了 2.6.1 版本 paddle 的 CPU 镜像 </td>
151-
</tr>
152-
<tr>
153-
<td> registry.baidubce.com/paddlepaddle/paddle:2.6.1-jupyter </td>
154-
<td> 安装了 2.6.1 版本 paddle 的 CPU 镜像,且镜像中预装好了 jupyter,启动 docker 即运行 jupyter 服务 </td>
143+
<td> registry.baidubce.com/paddlepaddle/paddle:3.0.0b0 </td>
144+
<td> 安装了 3.0.0b0 版本 paddle 的 CPU 镜像 </td>
155145
</tr>
156146
<tr>
157-
<td> registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda12.0-cudnn8.9-trt8.6 </td>
158-
<td> 安装了 2.6.1 版本 paddle 的 GPU 镜像,cuda 版本为 12.0,cudnn 版本为 8.9,trt 版本为 8.6 </td>
147+
<td> registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-jupyter </td>
148+
<td> 安装了 3.0.0b0 版本 paddle 的 CPU 镜像,且镜像中预装好了 jupyter,启动 docker 即运行 jupyter 服务 </td>
159149
</tr>
160150
<tr>
161-
<td> registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda11.7-cudnn8.4-trt8.4 </td>
162-
<td> 安装了 2.6.1 版本 paddle 的 GPU 镜像,cuda 版本为 11.7,cudnn 版本为 8.4,trt 版本为 8.4 </td>
151+
<td> registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda11.8-cudnn8.6-trt8.5 </td>
152+
<td> 安装了 3.0.0b0 版本 paddle 的 GPU 镜像,cuda 版本为 11.8,cudnn 版本为 8.6,trt 版本为 8.5 </td>
163153
</tr>
164154
<tr>
165-
<td> registry.baidubce.com/paddlepaddle/paddle:2.6.1-gpu-cuda11.2-cudnn8.2-trt8.0 </td>
166-
<td> 安装了 2.6.1 版本 paddle 的 GPU 镜像,cuda 版本为 11.2,cudnn 版本为 8.2,trt 版本为 8.0 </td>
155+
<td> registry.baidubce.com/paddlepaddle/paddle:3.0.0b0-gpu-cuda12.3-cudnn9.0-trt8.6 </td>
156+
<td> 安装了 3.0.0b0 版本 paddle 的 GPU 镜像,cuda 版本为 12.3,cudnn 版本为 9.0,trt 版本为 8.6 </td>
167157
</tr>
168158
</tbody>
169159
</table>

0 commit comments

Comments
 (0)