Maximize GPU Memory Usage on Headless System (GTX 750 Ti)

[megvadulthangya]

Hello lllyasviel & Devs!

I’m using Stable Diffusion WebUI Forge on a headless system with a GTX 750 Ti (4GB VRAM), and I’ve run into a challenge regarding GPU memory usage. In the GUI, the maximum GPU weight I can set is 4032 MB, but it seems like there is still some unused VRAM, and I would like to make sure that the program utilizes all of the available 4 GB of GPU memory, especially on my headless setup where I have 32 GB of system RAM and no need to reserve unused memory.

After a normal startup, the system only reserves 8 MB of GPU memory, which leaves a significant amount of unused VRAM. However, I’m still getting the following low VRAM warning during diffusion iterations:

[Low GPU VRAM Warning] Your current GPU free memory is 885.42 MB for this diffusion iteration.
[Low GPU VRAM Warning] This number is lower than the safe value of 1536.00 MB.
[Low GPU VRAM Warning] If you continue, you may cause NVIDIA GPU performance degradation for this diffusion process, and the speed may be extremely slow (about 10x slower).
[Low GPU VRAM Warning] To solve the problem, you can set the 'GPU Weights' (on the top of page) to a lower value.
[Low GPU VRAM Warning] If you cannot find 'GPU Weights', you can click the 'all' option in the 'UI' area on the left-top corner of the webpage.
[Low GPU VRAM Warning] If you want to take the risk of NVIDIA GPU fallback and test the 10x slower speed, you can (but are highly not recommended to) add '--disable-gpu-warning' to CMD flags to remove this warning.
It appears that the program might not be fully utilizing the available 4 GB VRAM, and I’m not sure whether this is properly accounted for in the memory management, especially since the warning seems to be optimized for non-headless systems. I’d like to force the program to use the full GPU memory on my headless system.

Is there a way to adjust the memory management or a specific flag/configuration that would allow full VRAM usage, or force to use all somehow...?

Thanks in advance for your help!