lllyasviel
[Bug]: What's up with AMD cards?
Checklist
- [ ] The issue exists after disabling all extensions
- [X] The issue exists on a clean installation of webui
- [ ] The issue is caused by an extension, but I believe it is caused by a bug in the webui
- [ ] The issue exists in the current version of the webui
- [ ] The issue has not been reported before recently
- [ ] The issue has been reported before but has not been fixed yet
What happened?
install is not working on AMD card I've properly installed ROCm and Torch drivers that function with automatic1111 webui But the same does not work with Forge.
just executing ./webui.sh gives me the expected -
RuntimeError: Torch is not able to use GPU; add --skip-torch-cuda-test to COMMANDLINE_ARGS variable to disable this check
executing ./webui.sh --skip-torch-cuda-test gives me -
RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Steps to reproduce the problem
install ROCm 5.7 export TORCH_COMMAND='pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.5.0' - so correct torch drivers get installed ./webui.sh --skip-torch-cuda-test should properly execute Forge webui
What should have happened?
Webui should have started the same as automatic1111 I would have presumed
What browsers do you use to access the UI ?
No response
Sysinfo
Console logs
❯ ./webui.sh --skip-torch-cuda-test
################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye), Fedora 34+ and openSUSE Leap 15.4 or newer.
################################################################
################################################################
Running on martin user
################################################################
################################################################
Repo already cloned, using it as install directory
################################################################
################################################################
Create and activate python venv
################################################################
################################################################
Launching launch.py...
################################################################
glibc version is 2.39
ldconfig: Can't open cache file /nix/store/ddwyrxif62r8n6xclvskjyy6szdhvj60-glibc-2.39-5/etc/ld.so.cache
: No such file or directory
ldconfig: Can't open cache file /nix/store/ddwyrxif62r8n6xclvskjyy6szdhvj60-glibc-2.39-5/etc/ld.so.cache
: No such file or directory
Cannot locate TCMalloc. Do you have tcmalloc or google-perftool installed on your system? (improves CPU memory usage)
Python 3.10.14 (main, Mar 19 2024, 21:46:16) [GCC 13.2.0]
Version: f0.0.17v1.8.0rc-latest-276-g29be1da7
Commit hash: 29be1da7cf2b5dccfc70fbdd33eb35c56a31ffb7
Launching Web UI with arguments: --skip-torch-cuda-test
Traceback (most recent call last):
File "/home/martin/devel/SD/stable-diffusion-webui-forge/launch.py", line 51, in <module>
main()
File "/home/martin/devel/SD/stable-diffusion-webui-forge/launch.py", line 47, in main
start()
File "/home/martin/devel/SD/stable-diffusion-webui-forge/modules/launch_utils.py", line 541, in start
import webui
File "/home/martin/devel/SD/stable-diffusion-webui-forge/webui.py", line 17, in <module>
initialize_forge()
File "/home/martin/devel/SD/stable-diffusion-webui-forge/modules_forge/initialization.py", line 50, in initialize_forge
import ldm_patched.modules.model_management as model_management
File "/home/martin/devel/SD/stable-diffusion-webui-forge/ldm_patched/modules/model_management.py", line 122, in <module>
total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
File "/home/martin/devel/SD/stable-diffusion-webui-forge/ldm_patched/modules/model_management.py", line 91, in get_torch_device
return torch.device(torch.cuda.current_device())
File "/home/martin/devel/SD/stable-diffusion-webui-forge/venv/lib/python3.10/site-packages/torch/cuda/__init__.py", line 787, in current_device
_lazy_init()
File "/home/martin/devel/SD/stable-diffusion-webui-forge/venv/lib/python3.10/site-packages/torch/cuda/__init__.py", line 302, in _lazy_init
torch._C._cuda_init()
RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
### Additional information
_No response_
I should also note that automatic1111 only produces a warning on the no Nvidia driver check -
Warning: caught exception 'Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx', memory monitor disabled
logs for automatic1111 -
stable-diffusion-webui on master [!+?] is 📦 v0.0.0 via via 🐍 v3.10.14 via ❄️ impure (stable-diffusion-webui)
❯ ./webui.sh --skip-torch-cuda-test --share --no-half --theme dark
################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye), Fedora 34+ and openSUSE Leap 15.4 or newer.
################################################################
################################################################
Running on martin user
################################################################
################################################################
Repo already cloned, using it as install directory
################################################################
################################################################
Create and activate python venv
################################################################
################################################################
Launching launch.py...
################################################################
glibc version is 2.39
ldconfig: Can't open cache file /nix/store/ddwyrxif62r8n6xclvskjyy6szdhvj60-glibc-2.39-5/etc/ld.so.cache
: No such file or directory
ldconfig: Can't open cache file /nix/store/ddwyrxif62r8n6xclvskjyy6szdhvj60-glibc-2.39-5/etc/ld.so.cache
: No such file or directory
Cannot locate TCMalloc. Do you have tcmalloc or google-perftool installed on your system? (improves CPU memory usage)
Python 3.10.14 (main, Mar 19 2024, 21:46:16) [GCC 13.2.0]
Version: v1.8.0
Commit hash: bef51aed032c0aaa5cfd80445bc4cf0d85b408b5
Launching Web UI with arguments: --skip-torch-cuda-test --share --no-half --theme dark
no module 'xformers'. Processing without...
no module 'xformers'. Processing without...
No module 'xformers'. Proceeding without it.
Warning: caught exception 'Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx', memory monitor disabled
CivitAI Browser+: Aria2 RPC started
Loading weights [ef8629e2c8] from /home/martin/devel/SD/stable-diffusion-webui/models/Stable-diffusion/protogenX34Photorealism_protogenX34.safetensors
Creating model from config: /home/martin/devel/SD/stable-diffusion-webui/configs/v1-inference.yaml
Running on local URL: http://127.0.0.1:7860
Applying attention optimization: InvokeAI... done.
Model loaded in 6.2s (load weights from disk: 0.6s, create model: 2.9s, apply weights to model: 2.3s, apply float(): 0.3s).
So there must be some subtle difference with Forge, stopping it from running for me
It's seems you are running on Linux. Try using by zluda Linux version might help you . You may tweak the code like SD.NExt. Then it should be working
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da7cf2b5dccfc70fbdd33eb35c56a31ffb7 Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in
main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda_init_.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda_init.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init__.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init_.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Try my link https://github.com/likelovewant/stable-diffusion-webui-forge-on-amd,make sure you have zluda support first
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init__.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init_.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Try my link https://github.com/likelovewant/stable-diffusion-webui-forge-on-amd,make sure you have zluda support first
tried your fork, got this at end of installing:
D:\stable-diffusion-webui-forge-on-amd>webui.bat --use-zluda
venv "D:\stable-diffusion-webui-forge-on-amd\venv\Scripts\Python.exe"
WARNING: ZLUDA works , You are on an amazing Journey ,Engjoy it
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: f0.0.17v1.8.0rc-1.7.0
Commit hash: 85f736f4d6f4e28b0ebe888251153d1ef4b4c18f
Cloning assets into D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets...
Cloning into 'D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets'...
fatal: unable to access 'https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git/': The requested URL returned error: 403
Traceback (most recent call last):
File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 51, in
@likelovewant @patientx
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init__.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init_.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Try my link https://github.com/likelovewant/stable-diffusion-webui-forge-on-amd,make sure you have zluda support first
tried your fork, got this at end of installing:
D:\stable-diffusion-webui-forge-on-amd>webui.bat --use-zluda venv "D:\stable-diffusion-webui-forge-on-amd\venv\Scripts\Python.exe" WARNING: ZLUDA works , You are on an amazing Journey ,Engjoy it 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: f0.0.17v1.8.0rc-1.7.0 Commit hash: 85f736f Cloning assets into D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets... Cloning into 'D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets'... fatal: unable to access 'https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git/': The requested URL returned error: 403 Traceback (most recent call last): File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 51, in main() File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 39, in main prepare_environment() File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 507, in prepare_environment git_clone(assets_repo, repo_dir('stable-diffusion-webui-assets'), "assets", assets_commit_hash) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 201, in git_clone run(f'"{git}" clone --config core.filemode=false "{url}" "{dir}"', f"Cloning {name} into {dir}...", f"Couldn't clone {name}", live=True) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 125, in run raise RuntimeError("\n".join(error_bits)) RuntimeError: Couldn't clone assets. Command: "git" clone --config core.filemode=false "https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git" "D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets" Error code: 128 Press any key to continue . . .
yeah i get the same error. annoying. i'd love to get this eventually working as well as posting issues in the correct repo, also might be a great thing too
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init__.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init_.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Try my link https://github.com/likelovewant/stable-diffusion-webui-forge-on-amd,make sure you have zluda support first
tried your fork, got this at end of installing:
D:\stable-diffusion-webui-forge-on-amd>webui.bat --use-zluda venv "D:\stable-diffusion-webui-forge-on-amd\venv\Scripts\Python.exe" WARNING: ZLUDA works , You are on an amazing Journey ,Engjoy it 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: f0.0.17v1.8.0rc-1.7.0 Commit hash: 85f736f Cloning assets into D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets... Cloning into 'D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets'... fatal: unable to access 'https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git/': The requested URL returned error: 403 Traceback (most recent call last): File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 51, in main() File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 39, in main prepare_environment() File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 507, in prepare_environment git_clone(assets_repo, repo_dir('stable-diffusion-webui-assets'), "assets", assets_commit_hash) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 201, in git_clone run(f'"{git}" clone --config core.filemode=false "{url}" "{dir}"', f"Cloning {name} into {dir}...", f"Couldn't clone {name}", live=True) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 125, in run raise RuntimeError("\n".join(error_bits)) RuntimeError: Couldn't clone assets. Command: "git" clone --config core.filemode=false "https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git" "D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets" Error code: 128 Press any key to continue . . .
dete the the word " fast" /modules) /launch_utils.py , change
Same problem, but on Windows 10.
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] Version: f0.0.17v1.8.0rc-latest-276-g29be1da7 Commit hash: 29be1da Launching Web UI with arguments: --skip-torch-cuda-test --opt-sdp-attention --upcast-sampling --opt-channelslast Traceback (most recent call last): File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 51, in main() File "E:\Downloads\webui_forge_cu121_torch21\webui\launch.py", line 47, in main start() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules\launch_utils.py", line 541, in start import webui File "E:\Downloads\webui_forge_cu121_torch21\webui\webui.py", line 17, in initialize_forge() File "E:\Downloads\webui_forge_cu121_torch21\webui\modules_forge\initialization.py", line 50, in initialize_forge import ldm_patched.modules.model_management as model_management File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 122, in total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) File "E:\Downloads\webui_forge_cu121_torch21\webui\ldm_patched\modules\model_management.py", line 91, in get_torch_device return torch.device(torch.cuda.current_device()) File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init__.py", line 769, in current_device lazy_init() File "E:\Downloads\webui_forge_cu121_torch21\system\python\lib\site-packages\torch\cuda__init_.py", line 298, in _lazy_init torch._C._cuda_init() RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
Try my link https://github.com/likelovewant/stable-diffusion-webui-forge-on-amd,make sure you have zluda support first
tried your fork, got this at end of installing:
D:\stable-diffusion-webui-forge-on-amd>webui.bat --use-zluda venv "D:\stable-diffusion-webui-forge-on-amd\venv\Scripts\Python.exe" WARNING: ZLUDA works , You are on an amazing Journey ,Engjoy it 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: f0.0.17v1.8.0rc-1.7.0 Commit hash: 85f736f Cloning assets into D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets... Cloning into 'D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets'... fatal: unable to access 'https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git/': The requested URL returned error: 403 Traceback (most recent call last): File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 51, in main() File "D:\stable-diffusion-webui-forge-on-amd\launch.py", line 39, in main prepare_environment() File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 507, in prepare_environment git_clone(assets_repo, repo_dir('stable-diffusion-webui-assets'), "assets", assets_commit_hash) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 201, in git_clone run(f'"{git}" clone --config core.filemode=false "{url}" "{dir}"', f"Cloning {name} into {dir}...", f"Couldn't clone {name}", live=True) File "D:\stable-diffusion-webui-forge-on-amd\modules\launch_utils.py", line 125, in run raise RuntimeError("\n".join(error_bits)) RuntimeError: Couldn't clone assets. Command: "git" clone --config core.filemode=false "https://githubfast.com/AUTOMATIC1111/stable-diffusion-webui-assets.git" "D:\stable-diffusion-webui-forge-on-amd\repositories\stable-diffusion-webui-assets" Error code: 128 Press any key to continue . . .
delete the the word " fast" after the github on line 444, , in the file modules /launch_utils.py, it's might not work in some countries .or reclone it . I update the file .
What is the output of?
lspci 2>/dev/null | grep -E "VGA|Display"
You exporting TORCH_COMMAND doesn't work because it gets overwritten in the webui.sh script anyways, it should detect your card and install the correct pytorch version. You have the cuda version installed.
How I got it to work with ROCM on my rx6700
Nuke the existing venv
rm -rf ./venv
Create a new venv
python -m venv venv
Activate the venv
source venv/bin/activate
Install Pytorch
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7
or optionally nightly
pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/rocm6.0
launch the webui
./webui.sh
delete the the word " fast" after the github on line 444, , in the file modules /launch_utils.py, it's might not work in some countries .or reclone it . I update the file .
Yes !! At least a working zluda implementation. Neither the og sd-next nor any other implementation worked this well, my usual gen takes around 100 secs , it is down to 70 secs. The gen times falls down to 10 secs / it when I gen a lot in comfyui and in ideal times it is around that but it flactuates a lot, 10sec/it is achievable but not all the time. But here I am almost always 10 or below 10. So consistent, at lastly the memory management of forge beats everything hands down.
patientx I don't know your gpu number , I have tested it in several gpu , including the igpu or some old card , gfx1012, gfx1035, gfx 1103 , the speed should be at 1.5 it/s to 1.1 it/s for image 512-512px no matter how much varm of dedicted gpu are , 4g or 16g . the gen time is same . 10 secs/it still way to slow unless you generate large file with Hire-fix on. the speed have a room to improve so far.
patientx I don't know your gpu number , I have tested it in several gpu , including the igpu or some old card , gfx1012, gfx1035, gfx 1103 , the speed should be at 1.5 it/s to 1.1 it/s for image 512-512px no matter how much varm of dedicted gpu are , 4g or 16g . the gen time is same . 10 secs/it still way to slow unless you generate large file with Hire-fix on. the speed have a room to improve so far.
6600 and this is not 512x512 , this is sdxl. 1024x1024. I already get over 1it/s with comfyui with standard settings. The speed on generations is not very much changed but the consistency is, meaning I get the same gen speeds whether I started the app just a few seconds ago or doing it for half an hour. This is not the case for directml for me , so overall good time gains.
BTW, reactor doesn't work with this in CPU or CUDA mode. Both of them gives errors. (problem is at face restoring part I think)
patientx I don't know your gpu number , I have tested it in several gpu , including the igpu or some old card , gfx1012, gfx1035, gfx 1103 , the speed should be at 1.5 it/s to 1.1 it/s for image 512-512px no matter how much varm of dedicted gpu are , 4g or 16g . the gen time is same . 10 secs/it still way to slow unless you generate large file with Hire-fix on. the speed have a room to improve so far.
6600 and this is not 512x512 , this is sdxl. 1024x1024. I already get over 1it/s with comfyui with standard settings. The speed on generations is not very much changed but the consistency is, meaning I get the same gen speeds whether I started the app just a few seconds ago or doing it for half an hour. This is not the case for directml for me , so overall good time gains.
BTW, reactor doesn't work with this in CPU or CUDA mode. Both of them gives errors. (problem is at face restoring part I think)
I think the 6600 have better performace that that . you may need to rebuild your rocblas or check other available rocblas. I have see someone share a new tensile in sd.next discord for 6600 which have better perfomace . my apu for sdxl 1024 could be 4s/it. and there should have a room to improve the speed . 6600 is better that a igpu for sure . anyway ,try if you want to improve your speed .also reacor not work .you may test the faceswaplab .
patientx I don't know your gpu number , I have tested it in several gpu , including the igpu or some old card , gfx1012, gfx1035, gfx 1103 , the speed should be at 1.5 it/s to 1.1 it/s for image 512-512px no matter how much varm of dedicted gpu are , 4g or 16g . the gen time is same . 10 secs/it still way to slow unless you generate large file with Hire-fix on. the speed have a room to improve so far.
6600 and this is not 512x512 , this is sdxl. 1024x1024. I already get over 1it/s with comfyui with standard settings. The speed on generations is not very much changed but the consistency is, meaning I get the same gen speeds whether I started the app just a few seconds ago or doing it for half an hour. This is not the case for directml for me , so overall good time gains. BTW, reactor doesn't work with this in CPU or CUDA mode. Both of them gives errors. (problem is at face restoring part I think)
I think the 6600 have better performance that that . you may need to rebuild your rocblas or check other available rocblas. I have see someone share a new tensile in sd.next discord for 6600 which have better perfomace . my apu for sdxl 1024 could be 4s/it. and there should have a room to improve the speed . 6600 is better that a igpu for sure . anyway ,try if you want to improve your speed .also reacor not work .you may test the faceswaplab .
Thanks for your fork once again. I experimented with cmd arguments, "--pin-shared-memory" made it slower, but with "--cuda-stream" the gen time fell down to 8sec/it , and I also added those new lib files for 6600 now it is around 6.5 sec/it.
If I use lightning lora / lightning models with cfg 1 , the gen time fells down to 4 sec/it.
And one more thing to add, my gpu (6600) and my cpu (3600x) is power limited , both @40w. (electricity costs are too high in my country) So all these numbers before and after are with this limits.
EDIT: "Released the limits" on the gpu and holy s**** , 2.4 sec / it @ 100w ... It is the same speed I was getting with sd 1.5 one year ago when I started with this same hardware. With sd1.5 , it is 2.84 it / sec . Just wow...
PS : please continue updating your fork if possible with all these new info added there are a lot of people who are eager to try this zluda thing with amd gpu's. Thanks for the hard work.
Glad to see the improve . A solar panel with a 100w or less plus a Inverter might help you get free electricity. Sure , I will keep update . However , the are only 4 files I touched , and not so much code changes . it would be easier for anyone who could keep it .