README.md 2.66 KB
Newer Older
1
## 简介
facebook-github-bot's avatar
facebook-github-bot committed
2

3
**PyTorch3D** 是一个用于深度学习中处理**3D数据**的库。它提供了一系列功能,方便处理和操作三维数据,
facebook-github-bot's avatar
facebook-github-bot committed
4

sangwz's avatar
sangwz committed
5
## 安装
facebook-github-bot's avatar
facebook-github-bot committed
6

sangwz's avatar
sangwz committed
7
组件支持:
facebook-github-bot's avatar
facebook-github-bot committed
8

9
* Python 3.8,3.9 or 3.10
facebook-github-bot's avatar
facebook-github-bot committed
10

11
12
- dtk-23.10
- pytorch-2.1
facebook-github-bot's avatar
facebook-github-bot committed
13

sangwz's avatar
sangwz committed
14
### 1. pip安装
facebook-github-bot's avatar
facebook-github-bot committed
15

sangwz's avatar
sangwz committed
16
17
18
19
20
21
22
23
24
25
可以在光合[光合开发者社区](https://developer.hpccube.com/tool/#sdk) AI 生态包中获取最新的  `pytorch3d` 版本,注意应与DCU Toolkit版本以及pytorch版本对应

```shell
wget http://10.6.10.68:8000/customized/pytorch3d/23.10/pytorch3d-0.7.6-cp38-cp38-linux_x86_64.whl
python -m pip install pytorch3d-0.7.6-cp38-cp38-linux_x86_64.whl
```

### 2. 源码编译安装

#### 环境准备
facebook-github-bot's avatar
facebook-github-bot committed
26

27
*[开发者社区](https://developer.hpccube.com/tool/#sdk) DCU Toolkit 中下载 DTK-23.10 解压至 /opt/ 路径下,并建立软链接
facebook-github-bot's avatar
facebook-github-bot committed
28

29
30
31
  ```plaintext
  cd /opt && ln -s dtk-23.10 dtk
  ```
32

sangwz's avatar
sangwz committed
33
* 在光合[光合开发者社区](https://developer.hpccube.com/tool/#sdk) AI 生态包中获取对应的 pytorch-2.1版本(需对应 DCU Toolkit 版本)
facebook-github-bot's avatar
facebook-github-bot committed
34

35
  ```shell
sangwz's avatar
sangwz committed
36
37
  cd path/to/torch
  python3 -m pip install torch*
38
  ```
facebook-github-bot's avatar
facebook-github-bot committed
39

40
* 导入环境变量以及安装必要依赖库
facebook-github-bot's avatar
facebook-github-bot committed
41

42
43
44
45
  ```shell
  source /opt/dtk/env.sh
  pip3 install wheel -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn
  ```
facebook-github-bot's avatar
facebook-github-bot committed
46

sangwz's avatar
sangwz committed
47
#### 编译安装
Jeremy Reizenstein's avatar
Jeremy Reizenstein committed
48

49
* 拉取代码
Jeremy Reizenstein's avatar
Jeremy Reizenstein committed
50

51
  ```shell
sangwz's avatar
sangwz committed
52
  git clone http://developer.hpccube.com/codes/aicomponent/pytorch3d.git #注意根据需要切换分支
53
  ```
Jeremy Reizenstein's avatar
Jeremy Reizenstein committed
54

55
* 若使用 `GPU` 支持,需要设置环境变量`FORCE_CUDA`为1
Jeremy Reizenstein's avatar
Jeremy Reizenstein committed
56

57
58
59
  ```shell
  export FORCE_CUDA=1
  ```
Christoph Lassner's avatar
Christoph Lassner committed
60

61
- 执行编译命令
Nikhila Ravi's avatar
Nikhila Ravi committed
62

63
64
65
66
  ```shell
  cd path/to/pytorch3d
  pip install -e .
  ```
67

sangwz's avatar
sangwz committed
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
#### 注意事项

当出现错误时,参考以下说明:

* 报错提示缺少 `intel-mkl` 相关库,应安装 `intel-mkl` 或 添加库的路径到环境变量

  ```shell
  #安装
  yum-config-manager --add-repo https://yum.repos.intel.com/mkl/setup/intel-mkl.repo
  yum install intel-mkl-2020.0-088 -y --nogpgchec
  
  # 将库路径添加到环境变量中
  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/mkl/lib/intel64
  ```

* 缺少 `magma`时,安装对应库或添加环境变量

  ```shell
  # 默认dtk安装路径为 /opt/dtk
  cd /opt/dtk
  
  #安装
  wget http://10.6.10.68:8000/debug/pytorch/third_party/magma_v2.7.2-hip_nfs3.2_DTK23.10_intel-2020.1.217_07Oct2023.tar.gz
  tar -zxf magma_v2.7.2-hip_nfs3.2_DTK23.10_intel-2020.1.217_07Oct2023.tar.gz
  mv magma_v2.7.2-hip_nfs3.2_DTK23.10_intel-2020.1.217_07Oct2023 magma
  
  # 添加环境变量
  export LD_LIBRARY_PATH=${ROCM_PATH}/magma/lib:$LD_LIBRARY_PATH
  ```

## 验证

* python -c "import pytorch3d; pytorch3d.\_\_version__",版本号与官方版本同步,查询该软件的版本号。