How to install sd in any amd gpu. Torch not is not able to use gpu error fixed.

Not ROCm or Zluda, CPU. How to install stable diffusion in any amd gpu. Torch not found error fixed
There's commands: .\venv\Scripts\activate.bat
Install library: pip install torch-directml
COMMANDLINE_ARGS=--skip-torch-cuda-test --use-directml --lowvram --opt-sub-quad-attention --disable-nan-check.
stable diffusion, ai, generative ai, sdxl,sd,torch is not able to use gpu,how to run stable diffusion in amd gpu, stable diffusion amd, torch error, stable diffusion error,

[@googlengineer]

If you are installed latest version of SD, there might be problem with that arguments f.e: launch.py unrecognized arguments : --bla bla, just remove that argument in webui-user.bat file or go to github repo and take a look issues section

[@ArjunoWWH]

Hai, thx for the solution but i have another question, why my cpu and gpu only load about 5% in total when generating that causing long waiting time to finish the render?

[@shalapai2990]

I don't know man, I did everything as you said and I still have the same error

[@ИльяСоболев-р3п]

ONNX failed to initialize: Failed to import diffusers.pipelines.auto_pipeline because of the following error (look up to see its traceback).
How to fix?

[@wintersz668]

Hi, I've copy and pasted the command and it just says this
'.\venv\Scripts\activate.bat' is not recognized as an internal or external command,
operable program or batch file.
please help

[@Azaan-UK]

Hi, Just go to the venv folder, there scripts, and open cmd there, and type activate, it'll automatically activate it

[@ismailcelik4327]

Reis laptopta iki tane ekran kartı var birisi amd diğeri nvidia. Günlük kullanımda amd kullanıyor oyunda ise nvidia ya geçiyor. Ancak automatic1111 de sadece cpu kullanıyor bir türlü çözemedim.

[@OdrikaAnusmita]

What to do?? File "C:\Users\parve\miniconda3\envs\LivePortrait\lib\site-packages\torch\cuda\__init__.py", line 284, in _lazy_init
raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

[@googlengineer]

try to reinstall venv folder

[@aivisanta6838]

DLL load failed while importing onnxruntime_pybind11_state: How to solve this problem? Help please🙏

[@googlengineer]

should be problem with directX drivers, update or downgrade your gpu driver

[@aivisanta6838]

thanks🙏@@googlengineer

[@Viral_Shorts66642]

Allah razı olsun sayende hallettik

[@Zooloo38]

Failed
launchpy error: unrecognized arguments: --use-directml

[@googlengineer]

read pin

[@thelegendoof8744]

@@googlengineerdoesn’t help

[@nikhilkumar_7]

Why i m getting the same error even i have a nvidia graphics card 4gb not a amd. Should i follow this method or not?

[@googlengineer]

its depend on OS. torch is especially for nvidia try different versions of torch library or take a look that github.com/AUTOMATIC1111/stable-diffusion-webui/issues/15614#issuecomment-2221500167

[@ohmboneee2235]

But why u use --skip-torch-cuba if u use torch directml ?

[@googlengineer]

Cuda is nvidia technology, zluda is alternative for cuda in amd gpu's, also torch doesn't support amd gpu's that's why we need to use torch-directml. It's pretty complex and If it works don't touch it

[@AnonSkyBlu]

i have radeon rx 570 can i run stable diffusion or not?

[@googlengineer]

yes you can

[@AnonSkyBlu]

@@googlengineer and question here should i install zluda ? since it's alternative for cuda in amd gpu

[@googlengineer]

@@AnonSkyBlu just watch the video

[@RowmitHSN]

Getting images with only noise, Not able to generate properly

[@googlengineer]

should be problem with gpu, try to update gpu driver

[@RowmitHSN]

@@googlengineer There was problem with certain models like dreamshapner redownloading them fixed it

[@nampham7110]

launch.py: error: unrecognized arguments: --use-directml

[@googlengineer]

read pin

[@FondoNoruego]

AttributeError: module 'torch' has no attribute 'dml' :(

[@googlengineer]

try to reinstall venv folder

[@thatguy5]

you lost me for a long time at the virtual environment. Apparently in Win 10 its a little different, but I managed to figure it out. I've still noticed that SD runs far slower than comfyui for me. Thats for another day.

[@dyro7868]

how to do in windows 10 becouse its not working for me

[@maximlukyanov6561]

Thank you so much😁

[@DamilolaKarpov13]

Спасибо, бро! 2 дня искал решение проблемы

[@Maverick-hg8dp]

thank you

[@Asae.-.]

Not working at all sadly

[@googlengineer]

What's the problem

[@sanchezcruzjoserodrigo8355]

didn't work for me :(

[@googlengineer]

Why :(

[@sanchezcruzjoserodrigo8355]

@@googlengineer I think I skipped a step or maybe I downloaded something wrong. I'm not sure but after setting the commands and text, the program doesn't work. Can you a tutorial from 0. Step by step

[@googlengineer]

@@sanchezcruzjoserodrigo8355 Unfortunately I can't. it's so easy just install specific version of python, git and clone repository and then watch the video again

[@sanchezcruzjoserodrigo8355]

@@googlengineer okey i'll uninstall everything and try again

[@InteligenciaIndependienteL-u2i]

i have this same issue but on NVIDIA

[@googlengineer]

take a look at this issue github.com/AUTOMATIC1111/stable-diffusion-webui/issues/15614#issuecomment-2221500167

[@InteligenciaIndependienteL-u2i]

@@googlengineer hey thank you, but yeah did see that, no luck ;(

[@ZakiAsir]

bless you you kind soul

[@streamer7222]

thx

[@rafikimad3436]

thank youuuuuu

[@oLoboBrancoTV]

YOU ARE FUCKING AWESOME! LIKE + SUB

[@GarteGamingOfficial]

allah bless you bro

[@googlengineer]

Thx

[@deswing88]

i love you