#Better A1111 - CUDA 12

39 messages · Page 1 of 1 (latest)

fresh grove
#

Here comes new template for automatic1111 web UI!

Includes:

  • fast sync (5-10 min)
  • extensions install like on your pc
    Not includes:
  • Any models (you need bring own models recomend using network storage)
  • No preinstalled extensions (so you can costumize how you want)

Link: https://runpod.io/console/deploy?template=l0augaikfv&ref=vfker49t

tawny skiff
#

do you know how to download ADetailer onto that?

fresh grove
#

Just like you would install any extension

#

@tawny skiff for sure recomend using network storage if you plan use long term

tawny skiff
fresh grove
#

In web ui you have extensions tab 😄

tawny skiff
#

ah okay ty

tawny skiff
#

keeps showing this when I click on the port

fresh grove
#

Did you let it sync?

tawny skiff
#

i've waited a while yes

#

is there something else I need to do?

#

cpu is also down to 0%

fresh grove
#

weird

tawny skiff
#

ive had this issue with all the templates

#

I have no clue why its not working

fresh grove
#

not sure why it uses python3.10

#

works fine for me

sudden river
#

also using network storage?

tawny skiff
sudden river
#

I tested and all works fine

tawny skiff
#

I see. Do you know how it might be an issue on my end? and how to fix it?

fresh grove
#

not sure what you did

#

using network storage or volume?

tawny skiff
#

I simply loaded it up and tried to connect to port

#

network volume, I believe. The permanent one that can connect to any pod in the EU region

#

I have used them previously but they have stopped working all of a sudden

fresh grove
#

If you want we can jump on call and you can show me

#

If you would prefer like this if not can we via text

tawny skiff
fresh grove
#

fell free to dm me

tawny skiff
#

done

nocturne axle
#

Any plans to add CUDA >12.8 support for 5090 cards? 🙂

fresh grove
#

No need to use CUDA 12.8 just install PyTorch Nightly

nocturne axle
zenith adder
#

Sorry for hijacking this thread but... if I want to install Cuda nightly, I need to install it into a virtual environment, is that correct? I'm using ComfyUI, and that template had no venv folder so I'm guessing it is a standalone version. Can i convert it into a venv version somehow? I tried removing cuda and installing the nightly outside of any venv, but it kept putting the files onto the local storage and not my network storage, so I got "disk full" all the time.