stable-diffusion.cpp
stable-diffusion.cpp copied to clipboard
metal-flash-attention support
Can this project help for you? https://github.com/philipturner/metal-flash-attention
So far, metal-flash-attention can indeed provide the fastest generation speed for stable diffusion on MacOS.
Thank you for the feedback. I'm currently focusing on making it run faster, and I'll make time to take a look at this project and see if I can offer any assistance.
I came here to suggest the same thing.
linking this here for reference https://github.com/ggerganov/ggml/issues/293
What needs to be done to make this happen? I'm not very good with cpp, but I want to help.
https://github.com/leejet/stable-diffusion.cpp/pull/386 I am in the dark on metal flash attention support, or metal support in general. So would be nice if someone with the hardware could test the pr. :)