it is work on AMD GPU?

[AdokinBodoke]

RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx

some idea?

thanks!

[harold1110]

No it does not work on amd gpus but i think they are working on it

[alli1325]

hope they released it soon

[yorai1212]

Really looking for an AMD support.

[sarthaq]

Got something? New here

[mashb1t]

AMD GPUs are supported and can be used with DirectML (--directml in args).
See

• https://github.com/lllyasviel/Fooocus/blob/main/ldm_patched/modules/args_parser.py#L76
• https://github.com/lllyasviel/Fooocus?tab=readme-ov-file#minimal-requirement

[mashb1t]

In general, AMD VRAM usage has recently been cut in half by an update in the backend code, so i'm positive that with some driver updates by AMD and maybe a few more adjustments image generation might be competitive again.

[yorai1212]

Okay, so, in my end, there's only:

1. Updating Fooocus whenever there's an update.
2. Updating AMD drivers whenever there's an update as well.
   ?

[mashb1t]

I mean, you can also buy new hardware, but for now that's the gist, yes.

[mashb1t]

@yorai1212 i've just read about https://github.com/vosen/ZLUDA?tab=readme-ov-file#faq, but ROCm should be better as it's officially supported by AMD. Give it a try!

[yorai1212]

No idea what that ZLUDA is, I'll just wait for the future then I guess, thank you

[lgwjames]

if i use --directml my mac on amd card goes mad

[hqnicolas]

Working here:
#2552