# MMClassification算例测试

## 测试前准备

### 数据集准备

使用ImageNet-pytorch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+git2040069.dtk2210-cp37-cp37m-manylinux2014_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install torchvision-0.10.0a0+e04d001.dtk2210-cp37-cp37m-manylinux2014_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install mmcv_full-1.6.1+gitdebbc80.dtk2210-cp37-cp37m-manylinux2014_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple
mmcls 安装：
cd mmclassification-0.24.1
pip3 install -e .
```

注：测试不同版本的dtk，需安装对应版本的库whl包

## ResNet18测试
### 单精度测试

### 单卡测试（单精度）

```python
./sing_test.sh configs/resnet/resnet18_b32x8_imagenet.py
```
#### 参数说明

configs/_base_/datasets/imagenet_bs32.py 中batch_size=samples_per_gpu*卡数,性能计算方法：batch_size/time

#### 性能关注：time

### 多卡测试（单精度）
#### 单机多卡训练

1.pytorch单机多卡训练

```python
./tools/dist_train.sh configs/resnet/resnet18_b32x8_imagenet.py $GPUS
```
2.mpirun单机多卡训练
mpirun --allow-run-as-root --bind-to none -np 4 single_process.sh a03r3n15
a03r3n15为master节点ip

#### 多机多卡训练

1.pytorch多机多卡训练
在第一台机器上:
NODES=2 NODE_RANK=0 PORT=12345 MASTER_ADDR=10.1.3.56 sh tools/dist_train.sh configs/resnet/resnet18_b32x8_imagenet.py $GPUS
在第二台机器上:
NODES=2 NODE_RANK=1 PORT=12345 MASTER_ADDR=10.1.3.56 sh tools/dist_train.sh configs/resnet/resnet18_b32x8_imagenet.py $GPUS

2.mpirun多机多卡训练
mpirun --allow-run-as-root --hostfile hostfile --bind-to none -np 4 single_process.sh a03r3n15
a03r3n15为master节点ip

hostfile 文件

a03r3n15 slots=4

e10r4n04 slots=4

### 半精度测试
修改configs文件，添加fp16 = dict(loss_scale=512.)，单机多卡和多机多卡测试与单精度测试方法相同。

### 其他模型测试

其他模型的测试步骤和ResNet18相同，只需修改对应的config文件即可，下面列出相关模型对应的config文件列表：

| 模型          | configs                                                      |
| ------------- | ------------------------------------------------------------ |
| ResNet34      | configs/resnet/resnet34_b32x8_imagenet.py                    |
| ResNet50      | configs/resnet/resnet50_b32x8_imagenet.py                    |
| ResNet152     | configs/resnet/resnet152_b32x8_imagenet.py                   |
| Vgg11         | configs/vgg/vgg11_b32x8_imagenet.py                          |
| Vgg16         | configs/vgg/vgg16_b32x8_imagenet.py                          |
| SeresNet50    | configs/seresnet/seresnet50_b32x8_imagenet.py                |
| ResNext50     | configs/resnext/resnext50_32x4d_b32x8_imagenet.py            |
| MobileNet-v2  | configs/mobilenet_v2/mobilenet_v2_b32x8_imagenet.py          |
| ShuffleNet-v1 | configs/shufflenet_v1/shufflenet_v1_1x_b64x16_linearlr_bn_nowd_imagenet.py |
| ShuffleNet-v2 | configs/shufflenet_v2/shufflenet_v2_1x_b64x16_linearlr_bn_nowd_imagenet.py |