Closed skullmonkey666 closed 1 month ago
try running "patchzluda.bat"
and if it still gives error for the last option try running :
"fixforrx580.bat"
which installs a previous torch because latest versions have a problem with rx480-580.
Thank you for the answer. I did run patchzluda.bat many times as well as fixfor580.bat. fixfor580.bat just open a new console window and stops. So I entered the commands manualy, but without any luck. I'm out of ideas....
U need to manually patch zluda yourself. the patchzluda batch script runs into errors and instantly closes. My guess is curl fails to download the zluda files. Anyways... go to this link and it'll download zluda https://github.com/lshqqytiger/ZLUDA/releases/download/rel.11cc5844514f93161e0e74387f04e2c537705a82/ZLUDA-windows-amd64.zip
then take the cublas.dll, cusparse.dll, and nvrtc.dll from the zluda files and rename them into cublas64_11.dll, cusparse64_11.dll, and nvrtc64_112_0.dll Now take these renamed zluda files (three), and replace them inside of your comfyui-zluda\venv\Lib\site-packages\torch\lib
Hi ItsNifer, I just did as you suggested without any luck. Still the same error
Microsoft Windows [version 10.0.19045.4894] (c) Microsoft Corporation. Tous droits réservés.
G:\AI\ComfyUI-Zluda>comfyui.bat *** Checking and updating to new version if possible remote: Enumerating objects: 48, done. remote: Counting objects: 100% (48/48), done. remote: Compressing objects: 100% (38/38), done. remote: Total 48 (delta 18), reused 32 (delta 9), pack-reused 0 (from 0) Unpacking objects: 100% (48/48), 57.15 KiB | 352.00 KiB/s, done. From https://github.com/patientx/ComfyUI-Zluda bb3fc16..4eb0983 master -> origin/master Updating bb3fc16..4eb0983 Fast-forward api_server/routes/internal/internal_routes.py | 9 ++++- app/user_manager.py | 23 +++++++++++ comfy/controlnet.py | 56 +++++++++++++++------------ comfy/k_diffusion/sampling.py | 11 ++++++ comfy/lora.py | 2 +- comfy/model_management.py | 2 +- comfy/sd.py | 2 +- comfy_extras/nodes_custom_sampler.py | 22 +++++++++++ extra_model_paths.yaml.example | 2 + folder_paths.py | 7 +++- tests-unit/utils/extra_config_test.py | 4 +- utils/extra_config.py | 5 ++- 12 files changed, 111 insertions(+), 34 deletions(-)
[START] Security scan [DONE] Security scan
ComfyUI startup time: 2024-09-20 15:41:16.752111 Platform: Windows Python version: 3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)] Python executable: G:\AI\ComfyUI-Zluda\venv\Scripts\python.exe ComfyUI Path: G:\AI\ComfyUI-Zluda Log path: G:\AI\ComfyUI-Zluda\comfyui.log
Prestartup times for custom nodes: 7.2 seconds: G:\AI\ComfyUI-Zluda\custom_nodes\ComfyUI-Manager
Traceback (most recent call last):
File "G:\AI\ComfyUI-Zluda\main.py", line 88, in
G:\AI\ComfyUI-Zluda>
I just checked and it appears that my CPUs are of Westmere architecture. Could it be the issue ? According to the requirement it should be Haswell https://rocm.docs.amd.com/projects/install-on-windows/en/latest/reference/system-requirements.html#supported-gpus-win Am I right ?
operating system?Ram? wich gpu? gpu vram?which cpu?
i see your log, and try to updating torch to 2.4.1. It may be solved this problem.
Your question
First of all, I'm a complete noob with AI as much as github requests, so please forgive any mistake I could make. I'm having an issue installing ComfyUI as I get an error : OSError: [WinError 1114] Une routine d’initialisation d’une bibliothèque de liens dynamiques (DLL) a échoué. Error loading "G:\AI\ComfyUI-Zluda\venv\lib\site-packages\torch\lib\cublas64_11.dll" or one of its dependencies.
Latest AMD Drivers are installed, I'm running Python 3.10.11 and I think paths have been set correctly. I've downloaded and installed ROCm libraries for my GPU from here : https://github.com/DrENDzZ/ROCmLibs-Fallback (gfx803) My GPU is an AMD Radeon RX480, my OS is windows10 and my computer is a Macpro 5,1 but I don't know if that part is relevant. Any help would be gretly appreciated.
Regards
Logs
Other
No response