starred/ace-step-ui
1
0
Fork 0
mirror of https://github.com/fspecii/ace-step-ui.git synced 2026-04-25 14:15:49 +03:00
Code Issues 45 Projects Releases Packages Wiki Activity

[GH-ISSUE #25] VLLM / PT Backend #23

New issue
Closed
opened 2026-02-26 21:30:50 +03:00 by kerem · 1 comment
kerem commented 2026-02-26 21:30:50 +03:00
Owner
Copy link

Originally created by @GKartist75 on GitHub (Feb 5, 2026).
Original GitHub issue: https://github.com/fspecii/ace-step-ui/issues/25

Hi,

dont know if it also in your backend, but i was trying out different versions and in the one of Pinokkio Ace-Step 1.5 i found something that maybe also improves your program and reduce vram if you not already have it improved.

When using VLLM Backend then it uses 9,2 GB in VRAM.
But when using PT Backend then it is only 1.6 GB

Image
Originally created by @GKartist75 on GitHub (Feb 5, 2026). Original GitHub issue: https://github.com/fspecii/ace-step-ui/issues/25 Hi, dont know if it also in your backend, but i was trying out different versions and in the one of Pinokkio Ace-Step 1.5 i found something that maybe also improves your program and reduce vram if you not already have it improved. When using VLLM Backend then it uses 9,2 GB in VRAM. But when using PT Backend then it is only 1.6 GB <img width="1474" height="606" alt="Image" src="https://github.com/user-attachments/assets/7c4f79f4-b80f-4701-9a5e-31f4a4d2fef3" />
kerem closed this issue 2026-02-26 21:30:50 +03:00
kerem commented 2026-02-26 21:30:51 +03:00
Author
Owner
Copy link

@fspecii commented on GitHub (Feb 5, 2026):

Hi @GKartist75, thanks for reporting this! Great catch on the VRAM difference between the two backends.

We've just merged a change that adds an LM Backend selector to the UI. You can now switch between PT and VLLM directly from the Advanced Settings in the Create panel — no need to restart the server.

PT is now the default, so users with lower VRAM GPUs (4-8 GB) should have a much better experience out of the box. Users with powerful GPUs can still switch to VLLM if they prefer.

Closing this as resolved — feel free to reopen if you run into any issues!

<!-- gh-comment-id:3856127389 --> @fspecii commented on GitHub (Feb 5, 2026): Hi @GKartist75, thanks for reporting this! Great catch on the VRAM difference between the two backends. We've just merged a change that adds an **LM Backend selector** to the UI. You can now switch between PT and VLLM directly from the Advanced Settings in the Create panel — no need to restart the server. **PT is now the default**, so users with lower VRAM GPUs (4-8 GB) should have a much better experience out of the box. Users with powerful GPUs can still switch to VLLM if they prefer. Closing this as resolved — feel free to reopen if you run into any issues!
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
gradio-experimental
No results found.
Labels
Clear labels   
pull-request

Mirrored from GitHub Pull Request

No labels
pull-request
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
starred/ace-step-ui#23
No description provided.