invoke-ai / invoke-ai/launcher

[enhancement]: Put low-VRAM options into Settings

Open
#124 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
TypeScript
Stars
48
Forks
23
Avg merge
1m
Merged PRs (30d)
1

Description

The low-VRAM guide seems to be a bit much for artists who aren't techies. Instead of making them fiddle around with the invokeai.yaml file, the launcher could make the changes with an option under the gear menu.

There would be a checkmark for enabling low-VRAM, and if feasible, the launcher would detect the VRAM installed and advise the user.

Under that, greyed out until Low-VRAM is checked, there would be two radio buttons for the two different modes, perhaps with tooltips explaining to try both and see which works better for this particular system

It would look something like:

☐ Enable Low-VRAM Mode (12GB VRAM detected)
    ◉ Use default (native)
    ◎ Use CUDA allocator (cudaMallocAsync)

Changing the options would edit the invokeai.yaml appropriately.

Given that the changes will not take effect until Invoke is restarted, the user could be advised of that in text accompanying the option, or the entire option could be greyed out if Invoke is open.

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.