Closed KimoriWasTaken closed 1 year ago
I was JUST about to post this...
I saw an update last night but this morning i couldn't restart Auto1111
Did an entire reinstall to no avail.
Already up to date.
venv "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\venv\Scripts\Python.exe"
fatal: No names found, cannot describe anything.
Python 3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)]
Version: 1.5.1
Commit hash: ed20ba7f9ff593bcf012db79278930a381a55748
Launching Web UI with arguments: --autolaunch --upcast-sampling --opt-sub-quad-attention --opt-split-attention-v1 --no-half-vae --medvram
no module 'xformers'. Processing without...
no module 'xformers'. Processing without...
No module 'xformers'. Proceeding without it.
Traceback (most recent call last):
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\launch.py", line 39, in <module>
main()
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\launch.py", line 35, in main
start()
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\launch_utils.py", line 443, in start
import webui
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\webui.py", line 54, in <module>
from modules.call_queue import wrap_gradio_gpu_call, wrap_queued_call, queue_lock # noqa: F401
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\call_queue.py", line 6, in <module>
from modules import shared, progress, errors
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\shared.py", line 95, in <module>
directml_do_hijack()
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\dml\__init__.py", line 69, in directml_do_hijack
_set_memory_provider()
File "C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\dml\__init__.py", line 14, in _set_memory_provider
from modules.shared import opts, cmd_opts, log
ImportError: cannot import name 'opts' from partially initialized module 'modules.shared' (most likely due to a circular import) (C:\AI Art\Auto1111-SD\stable-diffusion-webui-directml\modules\shared.py)
I have the same issue on a previously working install, it was working like 12 hours ago. I guess something got wrong in the most recent update.
had to be, I generated one image this morning before work to test that I could run a batch... and something happened and it suddenly failed. hope for a fix in the next 2 hours :D lol (off work)
can't seem to find anything related to a change between now and ~12hours, no commit here, windows update or anything really. Guess we have to wait for someone
I have this too. I don't know enough to fix it myself. Seems like attempting to fix it might break any fixes forthcoming.
"from modules.shared import opts, cmd_opts, log ImportError: cannot import name 'opts' from partially initialized module 'modules.shared' (most likely due to a circular import) (K:\stable diffusion\webui\stable-diffusion-webui-directml\modules\shared.py) Press any key to continue . . ."
I have this too. I don't know enough to fix it myself. Seems like attempting to fix it might break any fixes forthcoming.
"from modules.shared import opts, cmd_opts, log ImportError: cannot import name 'opts' from partially initialized module 'modules.shared' (most likely due to a circular import) (K:\stable diffusion\webui\stable-diffusion-webui-directml\modules\shared.py) Press any key to continue . . ."
afaik you can mess with your own cloned repo and it won't impact the main one at all (not that you can anyway). I may try and fix it at home if it isn't working but i'm no programmer, this just seems like something that might be solved with a small change :x
I have this exact issue. Started about 12 hours ago. Waiting on fix,
I figured out the temp fix to roll back to the previous. You can switch back to the main repo later when an update hits.
Temp fix for Auto 1111 on AMD Go to folder where your 'stable-diffusion-webui-directml' folder is right click, open git bash type git checkout 4873e6a
use Auto1111 again. thank me later.
It might work on others, not sure though since I use the fork for AMD and directml, I assume nvidia would just go to the main repo and go to the last commit that was good (if they are even having issues?
@XeroCreator
OH MY GOSH THAT FIXED IT FOR NOW! THANK YOU
This is due to shoddy support of RDNA1 (and lower) cards (like RX5700XT) - torch>=2.0.0 won't run on these from my experience, at least on ROCm. This was apparently fixed in https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/11048 but someone forgot about it and forced torch version to be >=2.0.0 in requirements.txt
I figured out the temp fix to roll back to the previous. You can switch back to the main repo later when an update hits.
Temp fix for Auto 1111 on AMD Go to folder where your 'stable-diffusion-webui-directml' folder is right click, open git bash type git checkout 4873e6a
use Auto1111 again. thank me later.
It might work on others, not sure though since I use the fork for AMD and directml, I assume nvidia would just go to the main repo and go to the last commit that was good (if they are even having issues?
Though, I am a bit worried, since RDNA2 cards aren't working apparently too https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/12228 and they did before (no workaround present in shell script, so assuming that it's true) with torch>=2.0.0 so it could be due to newest torch version?
This is due to shoddy support of RDNA1 cards (like RX5700XT) Though, I am a bit worried, since RDNA2 cards aren't working apparently too #12228 and they did before (no workaround present in shell script, so assuming that it's true) with torch>=2.0.0 so it could be due to newest torch version?
I'm on RDNA3. Same problem.
I'm on RDNA3. Same problem.
Try downgrading torch to latest stable (2.0.0) version, and report if it works. torch 2.0.1 released in May, so no way that it could mess things up
Or downgrade to nightly versions before 2023-07-31, unless I'm mistaken and other requirements (or something else out of my scope) like torch-directml
or torchvision
are causing this issue
You know you're not using A1111 right? That's a forks so please don't open issues on the A1111 repository. Post your issue on there: https://github.com/lshqqytiger/stable-diffusion-webui-directml/issues The issue is invalid.
You know you're not using A1111 right? That's a forks so please don't open issues on the A1111 repository. Post your issue on there: https://github.com/lshqqytiger/stable-diffusion-webui-directml/issues The issue is invalid.
in fact I did not know. the wiki to install webui on amd gpus is on A1111: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs
so unless you read each step super extra careful you won't notice
edit: my wording sounds a little offensive, but I meant to be apologetic, as it's an oversight. I'll close the issue here and mark your answer as the solution
Is there an existing issue for this?
What happened?
git pulled as always, launch failed afterwards. Proceeded to try a clean-install, didn't work. => Unable to launch sd-webui at the moment as I don't know how to roll-back
Steps to reproduce the problem
What should have happened?
It should have launched.
Version or Commit where the problem happens
Version: 1.5.1
What Python version are you running on ?
Python 3.10.x
What platforms do you use to access the UI ?
Windows
What device are you running WebUI on?
AMD GPUs (RX 5000 below)
Cross attention optimization
Automatic
What browsers do you use to access the UI ?
Mozilla Firefox, Google Chrome
Command Line Arguments
List of extensions
clean install
Console logs
Additional information
No response