stable-diffusion-webui icon indicating copy to clipboard operation
stable-diffusion-webui copied to clipboard

[Bug]: I can't install xFormers

Open Franches13 opened this issue 10 months ago • 1 comments

Checklist

  • [ ] The issue exists after disabling all extensions
  • [x] The issue exists on a clean installation of webui
  • [ ] The issue is caused by an extension, but I believe it is caused by a bug in the webui
  • [ ] The issue exists in the current version of the webui
  • [ ] The issue has not been reported before recently
  • [ ] The issue has been reported before but has not been fixed yet

What happened?

i just want to install xformers for make faster images on StableDiffusion but i cant, no matter what i do i cant generate anything

Steps to reproduce the problem

  1. --xformers in webui-user.sh
  2. generate a image

What should have happened?

Xformers should be build for Cuda support

What browsers do you use to access the UI ?

Other

Sysinfo

RTX 3060TI 64gb ram

Console logs

raise NotImplementedError(msg)
    NotImplementedError: No operator found for `memory_efficient_attention_forward` with inputs:
         query       : shape=(2, 1024, 10, 64) (torch.float32)
         key         : shape=(2, 1024, 10, 64) (torch.float32)
         value       : shape=(2, 1024, 10, 64) (torch.float32)
         attn_bias   : <class 'NoneType'>
         p           : 0.0
    `decoderF` is not supported because:
        xFormers wasn't build with CUDA support
        attn_bias type is <class 'NoneType'>
        operator wasn't built - see `python -m xformers.info` for more info
    `[email protected]` is not supported because:
        xFormers wasn't build with CUDA support
        dtype=torch.float32 (supported: {torch.float16, torch.bfloat16})
        operator wasn't built - see `python -m xformers.info` for more info
    `tritonflashattF` is not supported because:
        xFormers wasn't build with CUDA support
        dtype=torch.float32 (supported: {torch.float16, torch.bfloat16})
        operator wasn't built - see `python -m xformers.info` for more info
        triton is not available
    `cutlassF` is not supported because:
        xFormers wasn't build with CUDA support
        operator wasn't built - see `python -m xformers.info` for more info
    `smallkF` is not supported because:
        max(query.shape[-1] != value.shape[-1]) > 32
        xFormers wasn't build with CUDA support
        operator wasn't built - see `python -m xformers.info` for more info
        unsupported embed per head: 64

Additional information

NotImplementedError: No operator found for memory_efficient_attention_forward with inputs: query : shape=(2, 1024, 10, 64) (torch.float32) key : shape=(2, 1024, 10, 64) (torch.float32) value : shape=(2, 1024, 10, 64) (torch.float32) attn_bias : <class 'NoneType'> p : 0.0 decoderF is not supported because: xFormers wasn't build with CUDA support attn_bias type is <class 'NoneType'> operator wasn't built - see python -m xformers.info for more info [email protected] is not supported because: xFormers wasn't build with CUDA support dtype=torch.float32 (supported: {torch.bfloat16, torch.float16}) operator wasn't built - see python -m xformers.info for more info tritonflashattF is not supported because: xFormers wasn't build with CUDA support dtype=torch.float32 (supported: {torch.bfloat16, torch.float16}) operator wasn't built - see python -m xformers.info for more info triton is not available cutlassF is not supported because: xFormers wasn't build with CUDA support operator wasn't built - see python -m xformers.info for more info smallkF is not supported because: max(query.shape[-1] != value.shape[-1]) > 32 xFormers wasn't build with CUDA support operator wasn't built - see python -m xformers.info for more info unsupported embed per head: 64

(Stable Diffusion)

Franches13 avatar Jan 26 '25 22:01 Franches13

I can't install xFormers

I don't believe that's the case "xformers wasn't build with CUDA support" if it isn't installed then it will be a completely different error message I believe it is installed just not correctly

I may have seen similar error messages cause by a a miss match of version between xformers and pytorch if you want people to actually help you then you would want to provide your sysinfo

sysinfo is a json file generated by web UI, not two lines, read the issue template instructions


because you did not provide sysinfo, the following is pure guesswork my guess some distro of linux with python 3.12~ as default, irrc the default pytorch that the default version we use is not support on 3.12 and so you install a new version manually then you use --xformers which by default installs a version of xformers meant for the old version of pytorch

if my guess is correct then I will try I will try installing xformers yourself manually, overriding what web UI does automatically or install an old version of python, 3.11 or 3.10 then run web ui usin it

w-e-w avatar Feb 04 '25 17:02 w-e-w