ENS:Best practices for deploying and using WAN2.1-T2V models in the edge cloud

Create an instance in the ENS console

1. Log on to the ENS console.

2. In the left-side navigation pane, choose Resources and Images > Instances.

3. On the Instances page, click Create Instance. For information about how to configure the ENS instance parameters, see Create an instance.

   1. You can configure the parameters based on your business requirements. The following table describes the recommended configurations.

      Step	Parameter	Recommended value
      Basic Configurations	Billing Method	Subscription
      	Instance type	x86 Computing
      	Instance Specification	NVIDIA 48GB * 2 For detailed specifications, contact the customer manager.
      	Image	Ubuntu ubuntu_22_04_x64_20G_alibase_20240926
      Network and Storage	Network	Self-built Network
      	System Disk	Ultra Disk, 80 GB or more
      	Data Disk	Ultra Disk, 1 TB or more
      System Settings	Set Password	Custom Key or Key Pair

   2. Confirm the order.

      After you complete the system settings, click Confirm Order in the lower-right corner. The system configures the instance based on your configuration and displays the price. After the payment is made, the page jumps to the ENS console.

      You can view the created instance in the ENS the console. If the instance is in the Running state, the instance is available.

Create an instance by calling an operation

You can also call an operation to create an instance in OpenAPI Portal.

The following section describes the reference code of the operation parameters.

{
  "InstanceType": "ens.gnxxxx",         <The instance type>
  "InstanceChargeType": "PrePaid",
  "ImageId": "ubuntu_22_04_x64_20G_alibase_20240926",
  "ScheduleAreaLevel": "Region",
  "EnsRegionId": "cn-your-ens-region",     <The edge node>
  "Password": <YOURPASSWORD>,           <The password>
  "InternetChargeType": "95BandwidthByMonth", 
  "SystemDisk": {
    "Size": 80,
    "Category": "cloud_efficiency"
  },
  "DataDisk": [
    {
      "Category": "cloud_efficiency",
      "Size": 1024
    }
  ],
  "InternetMaxBandwidthOut": 5000,
  "Amount": 1,
  "NetWorkId": "n-xxxxxxxxxxxxxxx",
  "VSwitchId": "vsw-xxxxxxxxxxxxxxx",
  "InstanceName": "test",        
  "HostName": "test",
  "PublicIpIdentification": true,
  "InstanceChargeStrategy": "instance",      <Billing based on instance>
}

Log on to the instance

For more information about how to log on to an instance, see Connect to an instance.

Initialize the disk

1. Expand the root directory.

   After you create or resize an instance, you need to expand the root partition online without restarting the instance.

   # Install the cloud environment toolkit.
   sudo apt-get update
   sudo apt-get install -y cloud-guest-utils
   
   # Ensure that the GPT partitioning tool sgdisk exists.
   type sgdisk || sudo apt-get install -y gdisk
   
   # Expand the physical partition.
   sudo LC_ALL=en_US.UTF-8 growpart /dev/vda 3
   
   # Resize the file system.
   sudo resize2fs /dev/vda3
   
   # Verify the resizing result.
   df -h

   

2. Mount the data disk.

   You need to format and mount the data disk. The following section provides the sample commands. Run the commands based on your business requirements.

   # Identify the new disk.
   lsblk
   
   # Format the disk without partitioning it.
   sudo mkfs -t ext4 /dev/vdb
   
   # Configure the mount.
   sudo mkdir /data
   echo "UUID=$(sudo blkid -s UUID -o value /dev/vdb) /data ext4 defaults,nofail 0 0" | sudo tee -a /etc/fstab
   
   # Verify the mount.
   sudo mount -a
   df -hT /data
   
   # Modify permissions.
   sudo chown $USER:$USER $MOUNT_DIR

   

   Note

   If you want to create an image based on the instance, you must delete the ext4 defaults 0 0 row from the /etc/fstab file. Otherwise, instances created based on the image cannot be started.

(Optional) Optimize the test script

You can achieve the following effect by modifying some code and parameters in the test script examples/wanvideo/wan_14b_text_to_video_tensor_parallel.py:

1. Avoid downloading the model again every time you run it.

2. Modify the compression quality of the video. A larger video file indicates better clarity.

3. Change the generated video from 480P (default) to 720P.

   # Modify the test code file.
   vim /root/DiffSynth-Studio/examples/wanvideo/wan_14b_text_to_video_tensor_parallel.py
   
   # 1. Comment out the code around line 125 to avoid downloading the model again every time you run it, thus using the previously downloaded model.
   
   # snapshot_download("Wan-AI/Wan2.1-T2V-14B", local_dir="models/Wan-AI/Wan2.1-T2V-14B")
   
   # 2. Modify the code around line 121, modify the save_video input parameter in the test_step function, video frame rate, and compression quality. 1 indicates the lowest quality. 10 indicates the highest quality.
   save_video(video, output_path, fps=24, quality=10)
   
   # 3. Add the following input parameter around line 135 to modify the resolution. The default value is 480p.
   ....
   "output_path": "video1.mp4",   
   # The following two lines are newly added.
   "height": 720,
   "width": 1280,
   },
   
   # 4. Save the file and exit the editor.
   :wq
   

Run the test script

# Run the test script.
torchrun --standalone --nproc_per_node=2 /root/DiffSynth-Studio/examples/wanvideo/wan_14b_text_to_video_tensor_parallel.py

Generate a video

You can refer to the remaining time prompt in the progress bar for the test time. After the running is complete, the video file video1.mp4 is generated. For information about how to use ossutil to import the generated video to Aliababa Coud OSS, see Copy objects.