Glad I could help!

You're welcome!

I have Rx 5700xt only able to generate 200x200 pixels before running into out of memory errors even with that parameter what sizes can you generate

@ I have an RX 550 4GB and I can easily generate 512x512 images (up to 1024x1024 on Linux with native ROCm and optimizations). The 5000 series has some unique problems; try using the parameters: "--opt-sub-quad-attention --medvram" and see if it works. With these parameters, you should be able to generate a 512x512 image in about 30 seconds.

@@MrJohnboyofsj it also depends on the model you are using. For instance if you use a cartoon model of some kind you may create higher resolution images, but using flux and stable may not be so successful.

Yes, I have been using Linux for gaming for about 8 years, and I can say that currently the experience is very good. Even on weak hardware, the games run very close to Windows, sometimes even better. However, be careful with online games as they tend to be problematic due to anti-cheat measures.

In Linux, drivers usually come with the kernel. In the case of AMD GPUs, everything works out of the box. The only thing you should note is to use the most recent kernel if your GPU has just been released (less than 6 months ago).

Sorry, but I haven't done a Forge tutorial on Windows yet. I was actually thinking about doing an update on the Windows tutorials. Keep an eye on the channel; it should be out next week.

@@Luinux-Techwaiting for

Yes, it works, but you will need to install it on a distro with native ROCm. PyTorch binaries are no longer compiled with support for older GPUs, so you have to use PyTorch compiled for these GPUs: github.com/LinuxMadeEZ/PyTorch-GFX803.

@@Luinux-Tech thanks i will try to do that using your guide👍 but still would apreciate an update since WebUI Forge (and its amdgpu directml fork) lack documentation

This is just a bug in DirectML. Don't worry, it will use all the VRAM it needs. You can check in the Task Manager.

Hehehe, the Lora used is: civitai.com/models/118892/shizune-naruto-lora and the model name is: "anyloraCheckpoint_bakedvaeBlessedFp16".

Actually never mind. I just needed to add the --highvram flag. Thanks for the tutorial.

SDL or "Non GUI" builds are builds without a graphical interface, in which you interact by passing commands to the binary through the terminal.

Xformers is not necessary in Forge, the built-in functionality is faster without it.

Are you using python 3.10? Newer versions may cause this problem.

No, it comes bundled with pytorch

You misspelled the command. The correct command is: "py -m venv venv".

@Luinux-Tech yeah.....I got everything to work correctly...do recommend some other models to try for comphy ui? As the model u used in this video doesn't give good pictures

I can't guarantee it, but try using the flag "HSA_OVERRIDE_GFX_VERSION=10.3.0".

I do, I have other videos where I show how to do it in Linux as well.

This is just a bug in DirectML. Don't worry, it will use all the VRAM it needs. You can check in the Task Manager.

@Luinux-Tech yeah u are right friend, but only sometimes the image generation works, most of the time it just gives me an error about lack of memory and won't let me continue. Is there any way to fix it? also ty for ur quick response

@@miauxdd6789 Can you tell me what the resolution of the images you are trying to generate is and what the Stable diffusion model is?

@Luinux-Tech its hunyuan for videos, 320x320 steps 25, framerate 24, frames 49, and other option that I don't remember in 64 that is for vram (less is minus vram)

Thanks, yes, I have a video about SD Forge planned. I'll try to make it before the end of the year. Regarding ComfyUI, you'll be surprised to know that it's much faster than A1111. The only downside is that the interface is more complex.

@@Luinux-Tech Thats why im interested by Forge too, for somes its even faster than ComfyUI or at least similar and the interface reminds me alot A1111.

Explain your error in detail, inform about your hardware(CPU + GPU), and specify whether you used WebUI or ComfyUI.

Activate the venv and run the command "python" or "py", this will start a python console(interative mode), then run "import torch", and then check if torch is seeing your gpu with the command: "torch.cuda.get_device_name(0)" does this command return the name of your GPU?

@Luinux-Tech Yes i've already fix that but there is another error could not locate tensor, the first one is around 400k bytes and after i use the --lowvram its change to 100k bytes any fix for this?

​@@Goddsindra can you tell me how did you fix the first error, it was working when i launched first 2 times but after that it started using cpu every time (using RX 6600) already tested (venv) PS C:\Users\minid\Documents\stable-diff> python >>> import torch >>> torch.cuda.get_device_name(0) Traceback (most recent call last): File "<stdin>", line 1, in <module> File "C:\Users\minid\Documents\stable-diff\venv\lib\site-packages\torch\cuda\__init__.py", line 435, in get_device_name return get_device_properties(device).name File "C:\Users\minid\Documents\stable-diff\venv\lib\site-packages\torch\cuda\__init__.py", line 465, in get_device_properties _lazy_init() # will define _get_device_properties File "C:\Users\minid\Documents\stable-diff\venv\lib\site-packages\torch\cuda\__init__.py", line 305, in _lazy_init raise AssertionError("Torch not compiled with CUDA enabled") AssertionError: Torch not compiled with CUDA enabled >>> print(torch.version.cuda) None >>> print(torch.cuda.is_available()) False >>>

im use cpu r5 5600 +gpu rx6600

Are you using Python 3.10.6?

@@Luinux-Tech now working . im downgrade python

Thanks, glad to help

it should work normally. Unfortunately, I can't guarantee it because I haven't tested it on APUs.

You are using the wrong path, check if you are in the correct folder and if there are no misspelled words, for example in your comment it says "\Scrips" instead of "\Scripts".