train.md 3.14 KB
Newer Older
unknown's avatar
unknown committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
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
# MMpose算例测试

## 测试前准备
### 数据集准备
使用dummy数据集.
### 环境部署
```python
yum install python3
yum install libquadmath
yum install numactl
yum install openmpi3
yum install glog
yum install lmdb-libs
yum install opencv-core
yum install opencv
yum install openblas-serial
pip3 install --upgrade pip
pip3 install opencv-python
```
### 安装python依赖包
```python
pip3 install torch-1.10.0a0+gitcc7c9c7-cp36-cp36m-linux_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install torchvision-0.10.0a0+300a8a4-cp36-cp36m-linux_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install mmcv_full-1.3.16-cp36-cp36m-linux_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
poseval安装
cd poseval
pip install -e .  
mmpose安装
cd mmpose 
pip3 install -e .
```
注:测试不同版本的dtk,需安装对应版本的库whl包,dtk22.04.1使用python3.7,dtk21.10.1使用python3.6
## ResNet50-Bottom-Up测试
### 单卡测试(单精度)
```python
./sing_test.sh configs/speed_test/bottomup_hrnet_w32_coco_512x512_dummy.py
```
若使用dtk22.04.1测试,需设置`export MIOPEN_FIND_MODE=1,export MIOPEN_USE_APPROXIMATE_PERFORMANCE=0`以下测试均相同.

#### 参数说明
configs/speed_test/bottomup_hrnet_w32_coco_512x512_dummy.py 中batch_size=samples_per_gpu*卡数,性能计算方法:batch_size/time
![1659168500714](image/train/1659168500714.png)
#### 性能关注:time
### 多卡测试(单精度)
```python
./multi_test.sh configs/speed_test/bottomup_hrnet_w32_coco_512x512_dummy.py
```
### 单卡测试(半精度)
```python
./sing_test.sh configs/speed_test/bottomup_hrnet_w32_coco_512x512_fp16_dummy.py
```
### 多卡测试(半精度)
```python
./multi_test.sh configs/speed_test/bottomup_hrnet_w32_coco_512x512_fp16_dummy.py
```
## ResNet50-Top-Down测试
### 单卡测试(单精度)
```python
./sing_test.sh configs/speed_test/res50_coco_256x192_dummy.py
```
#### 参数说明
configs/speed_test/res50_coco_256x192_dummy.py 中batch_size=samples_per_gpu*卡数,性能计算方法:batch_size/time
![1659169127556](image/train/1659169127556.png)
#### 性能关注:time
### 多卡测试(单精度)
```python
./multi_test.sh configs/speed_test/res50_coco_256x192_dummy.py
```
### 单卡测试(半精度)
```python
./sing_test.sh configs/speed_test/res50_coco_256x192_fp16_dummy.py
```
### 多卡测试(半精度)
```python
./multi_test.sh configs/speed_test/res50_coco_256x192_fp16_dummy.py
```
## HrNet-Top-Down测试
### 单卡测试(单精度)
```python
./sing_test.sh configs/speed_test/hrnet_w32_coco_256x192_dummy.py
```
#### 参数说明
configs/speed_test/hrnet_w32_coco_256x192_dummy.py 中batch_size=samples_per_gpu*卡数,性能计算方法:batch_size/time
![1659169253165](image/train/1659169253165.png)
#### 性能关注:time
### 多卡测试(单精度)
```python
./multi_test.sh configs/speed_test/hrnet_w32_coco_256x192_dummy.py
```
### 单卡测试(半精度)
```python
./sing_test.sh configs/speed_test/hrnet_w32_coco_256x192_fp16_dummy.py
```
### 多卡测试(半精度)
```python
./multi_test.sh configs/speed_test/hrnet_w32_coco_256x192_fp16_dummy.py
```