Quickly set up ComfyUI on a WUYING workspace-Elastic Desktop Service-阿里云帮助中心

You can use WUYING Workspace Enterprise Edition to quickly deploy the ComfyUI application. This helps you avoid local performance bottlenecks and complex environment configurations. This topic describes how to quickly set up a ComfyUI environment in a WUYING workspace and access the ComfyUI console from a client.

Limits

Supported region: China (Hangzhou).
The compute specifications for cloud desktops must include GPU memory.
The PC client must be version 8.2.0 or later.

Procedure

The following steps describe how to create a cloud computer with the ComfyUI service. For more information about configurations and parameters, see Create a cloud computer.

Log on to the WUYING Workspace Enterprise Edition console.
Click Create Workspace and then select Custom Buy.
For Computing Specification, select a specification type that includes a GPU. For Operating System, select Community Image and then the ComfyUI-* image.
Note
The asterisk (*) represents the version suffix.
For the best experience, select an instance with an NVIDIA RTX 5880 GPU with 48 GB of GPU memory or a higher configuration.
Configure the remaining parameters. For more information about the configuration items, see Create a WUYING workspace.
After you create the WUYING workspace and assign a user, log on to the WUYING workspace from the client.
Access the ComfyUI console at 127.0.0.1:8188 to verify the service deployment.
Note
ComfyUI starts automatically after the WUYING workspace starts. If access to 127.0.0.1:8188 is rejected, it may be because ComfyUI has not finished starting.
After the startup is complete, the ComfyUI console automatically opens.

Additional capabilities

WUYING Workspace Enterprise Edition provides additional AI accelerator capabilities for ComfyUI using workspace policies. These capabilities include the following:

Port service management: If you enable this feature, the port service management option appears in the client for direct access.
Academic acceleration: Optimizes the download speed for models and datasets from sources such as GitHub, HF-Mirror, and the ModelScope community.
Model library: If you enable this feature, a local disk named wuying-pub is added to the WUYING workspace. This disk provides access to the platform's model library.
Note
If the WUYING workspace is running, you must restart it after you enable the model library to load the required dependencies.