sd-webui can just be used by one user at the same time?

[bigbear0609]

I deploy a sd webui on a GPU server with NVIDIA Tesla T4,and configed a remote service ,it means anyone from any place can visit the sd webui url through the GPU server's public IP address.

For example , Li visit the webui url from NewYork, and James visit webui url from Chicago

But I find if Li is generating a image, the same time, James click the generate button ,he have to wait Li to finish her generating work (the webui show "waiting, queuing"). It is so confusing, the GPU server is almost idle when Li is generating image, Why not James can start generating immediately? If i want 2 or 200 people can start generating image from different place at the same time ,what can I do?

[bigbear0609]

The sd-webui web port is default port 7860 on my GPU server, if I give someone the public IP address and port ,he can use it to generate image freely.

[Sakura-Luna]

Start multiple WebUI instances.

[bigbear0609]

Thanks , how to config to "Start multiple WebUI instances" , i am a newbie.

[Sakura-Luna]

If you start multiple WebUIs, they will be mounted on different ports by default, and they can work in parallel.
Due to the nature of Python, single instance parallelism is not possible.

[bigbear0609]

Are you mean I start web service on different web port ,and deploy load balance to response multi users generate request?

[bigbear0609]

It's done,thanks you so much!

[sololll]

It's done,thanks you so much!

Done ?
How many GPUs are needed to generating multiple images at the same time ?