What happened?
First of all that my first time when i try help people with it so sorry if it have mistakes.
So when you install StabilityMatrix it support 4000 series and other but 5000 not. PyTorch(problem)
Fix it with this commands (archlinux for other same but different location)
cd /opt/stabilitymatrix/Data/Packages/stable-diffusion-webui/venv
That is location of virtual env that use matrix.
./bin/pip install --pre --upgrade --no-cache-dir torch --extra-index-url https://download.pytorch.org/whl/nightly/cu128
That command will install cu128 to your virtual env.
So yea all versions of 50 series have this problem now.
ps:And remember that on many linux distros venv may be in my location or other so ye check before run commands (or it on main env).
pss: if it still have problem use --skip-torch-cuda-test to test after fixing.
psss: if it still have problem you can try this(i don't really test it so use it at your own risk):
pip install --pre --upgrade --no-cache-dir xformers --extra-index-url https://download.pytorch.org/whl/nightly/cu128
pssss:And if it don't generate image you can use(in venv dir*):
./bin/pip uninstall -y fastapi starlette
./bin/pip install "fastapi==0.95.2" "starlette==0.27.0"
Sorry for that but you may have to use 'skip install'.
And stay safe.
All list of gpu that may have problem
GeForce RTX 5050
GeForce RTX 5060
GeForce RTX 5060 Ti (8GB)
GeForce RTX 5060 Ti (16GB)
GeForce RTX 5070
GeForce RTX 5070 Ti
GeForce RTX 5080
GeForce RTX 5090
OS Archlinux
What can help to debug
pytorch install commands
Steps to reproduce
- Install StabilityMatrix
yay -S (stabilitymatrix-git/stabilitymatrix/stabilitymatrix-bin)
- install webui (my situation stable diffusion web ui )
- Start webui get error {RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.}
Relevant logs
RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.
Version
v2.14.3
What Operating System are you using?
Linux
What happened?
First of all that my first time when i try help people with it so sorry if it have mistakes.
So when you install StabilityMatrix it support 4000 series and other but 5000 not. PyTorch(problem)
Fix it with this commands (archlinux for other same but different location)
That is location of virtual env that use matrix.
That command will install cu128 to your virtual env.
So yea all versions of 50 series have this problem now.
ps:And remember that on many linux distros venv may be in my location or other so ye check before run commands (or it on main env).
pss: if it still have problem use
--skip-torch-cuda-testto test after fixing.psss: if it still have problem you can try this(i don't really test it so use it at your own risk):
pssss:And if it don't generate image you can use(in venv dir*):
Sorry for that but you may have to use 'skip install'.
And stay safe.
All list of gpu that may have problem
OS Archlinux
What can help to debug
pytorch install commands
Steps to reproduce
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.}
Relevant logs
Version
v2.14.3
What Operating System are you using?
Linux