[Bug]: Segmentation fault (core dumped) on torch-2.0.0.dev20230209+rocm5.2

[KEDI103]

Is there an existing issue for this?

• I have searched the existing issues and checked the recent builds/commits

What happened?

(venv) b_cansin@b-cansin-ubuntu:/media/b_cansin/ai/ai/stable-diffusion-webui$ ./webui.sh

################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye)
################################################################

################################################################
Running on b_cansin user
################################################################

################################################################
Repo already cloned, using it as install directory
################################################################

################################################################
python venv already activate or run without venv: /media/b_cansin/ai/ai/stable-diffusion-webui/venv
################################################################

################################################################
Launching launch.py...
################################################################
Using TCMalloc: libtcmalloc_minimal.so.4
fatal: No names found, cannot describe anything.
Python 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0]
Version: 1.6.0
Commit hash: 7ba3923
Launching Web UI with arguments:
./webui.sh: line 256: 516482 Segmentation fault (core dumped) "${python_cmd}" -u "${LAUNCH_SCRIPT}" "$@"

Steps to reproduce the problem

Installed pytorch/pytorch#103973 (comment) for suggest to get work on pytorch 2 for SDXL
and
Launching Web UI with arguments:
./webui.sh: line 256: 516482 Segmentation fault (core dumped) "${python_cmd}" -u "${LAUNCH_SCRIPT}" "$@"

What should have happened?

I also tried with Invoke AI it worked but for AUTOMATIC1111 not worked. It should be work like Invoke AI.

Sysinfo

AMD Radeon VII
AMD® Fx(tm)-9590 eight-core processor × 8 (not supporting pci atomics)
32 gb ram
Ubuntu 22.04.3 LTS x64

What browsers do you use to access the UI ?

Mozilla Firefox

Console logs

(venv) b_cansin@b-cansin-ubuntu:/media/b_cansin/ai/ai/stable-diffusion-webui$ ./webui.sh

################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye)
################################################################

################################################################
Running on b_cansin user
################################################################

################################################################
Repo already cloned, using it as install directory
################################################################

################################################################
python venv already activate or run without venv: /media/b_cansin/ai/ai/stable-diffusion-webui/venv
################################################################

################################################################
Launching launch.py...
################################################################
Using TCMalloc: libtcmalloc_minimal.so.4
fatal: No names found, cannot describe anything.
Python 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0]
Version: 1.6.0
Commit hash: 7ba3923d5b494b7756d0b12f33acb3716d830b9a
Launching Web UI with arguments: 
./webui.sh: line 256: 516482 Segmentation fault      (core dumped) "${python_cmd}" -u "${LAUNCH_SCRIPT}" "$@"
(venv) b_cansin@b-cansin-ubuntu:/media/b_cansin/ai/ai/stable-diffusion-webui$ python launch.py
fatal: No names found, cannot describe anything.
Python 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0]
Version: 1.6.0
Commit hash: 7ba3923d5b494b7756d0b12f33acb3716d830b9a
Launching Web UI with arguments: 
Segmentation fault (core dumped)

Additional information

I need pytorch 2 for SDXL in webui but can't get offical one because of pci atomics missing this topic goes so long
pytorch/pytorch#103973
they suggest me this
pytorch/pytorch#103973 (comment)
https://download.pytorch.org/whl/nightly/rocm5.2/torch-2.0.0.dev20230209%2Brocm5.2-cp310-cp310-linux_x86_64.whl
https://download.pytorch.org/whl/nightly/rocm5.2/torchvision-0.15.0.dev20230209%2Brocm5.2-cp310-cp310-linux_x86_64.whl
https://download.pytorch.org/whl/nightly/rocm5.2/torchaudio-2.0.0.dev20230209%2Brocm5.2-cp310-cp310-linux_x86_64.whl
but this give me that error but in webui also test of invoke ai working.
also when I installing it I also noticed this

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
torchvision 0.14.1+rocm5.2 requires torch==1.13.1, but you have torch 2.0.0.dev20230209+rocm5.2 which is incompatible.
torchaudio 0.13.1+rocm5.2 requires torch==1.13.1, but you have torch 2.0.0.dev20230209+rocm5.2 which is incompatible.
open-clip-torch 2.20.0 requires protobuf<4, but you have protobuf 4.24.4 which is incompatible.
Successfully installed torch-2.0.0.dev20230209+rocm5.2

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
open-clip-torch 2.20.0 requires protobuf<4, but you have protobuf 4.24.4 which is incompatible.
Successfully installed torchvision-0.15.0.dev20230209+rocm5.2

[Jmvars]

I get the same, but I'm on 7900 XT which is technically not supported. I have tried every conceivable way both in this wiki and AMD procedures to get webui up and running on Linux but it refuses to work, I think I'll give up for now and wait for AMD to officially support my card before trying again.

I was able to get it to run but it was unacceptably slow. I'm able to run it on Windows with directml and get blazing fast generation but Loras and embeddings don't work, weights don't seem to work and plugins like Adetailer don't work. Sad that these are my options.

I knew AMD was slow on SD but I did not expect it to be this bad.

[KEDI103]

Okey heres after months of trying fix for missing pci atomics problem:
pytorch/pytorch#103973 (comment)
Gona be added on pre nighty build on next week for now that one fixing it.
pytorch/pytorch#103973 (comment)
But this needs also this fix for AUTOMATIC1111
#13985 (comment)
I am closing this one because no need this.