intel / openvino-ai-plugins-gimp

GIMP AI plugins with OpenVINO Backend
Apache License 2.0
507 stars 55 forks source link

WinError 126: Error Loading fbgemm.dll or one of it's dependencies #128

Closed HaywardUW closed 3 months ago

HaywardUW commented 3 months ago

Hello, I have built this out on several systems without fail. Starting last week I am getting the following error on multiple systems:

This is a development version of GIMP. Debug messages may appear here.

(gimp-2.99.exe:9112): GLib-GIO-WARNING **: 13:59:19.793: win32 session dbus binary not found set device 'System Aggregated Pointer' to mode: disabled Skipping potential plug-in 'C:\Users\TestLab\GIMP\gimpenv3\Lib\site-packages\gimpopenvino\plugins\plugin_utils.py': plug-ins must be installed in subdirectories. Skipping potential plug-in 'C:\Users\TestLab\GIMP\gimpenv3\Lib\site-packages\gimpopenvino\plugins__init__.py': plug-ins must be installed in subdirectories. Error in set_translation_domain. This is expected if running GIMP 2.99.11 or later Error in set_translation_domain. This is expected if running GIMP 2.99.11 or later Error in set_translation_domain. This is expected if running GIMP 2.99.11 or later C:\Users\TestLab\openvino-ai-plugins-gimp\weights..\gimp_openvino_run_sd.json not found, loading defaults No stable-diffusion model server found to kill Traceback (most recent call last): File "C:\Users\TestLab\GIMP\gimpenv3\Lib\site-packages\gimpopenvino\tools\stable-diffusion-ov-server.py", line 13, in import torch File "C:\Users\TestLab\GIMP\gimpenv3\lib\site-packages\torch__init__.py", line 148, in raise err OSError: [WinError 126] The specified module could not be found. Error loading "C:\Users\TestLab\GIMP\gimpenv3\lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies.

Any help is much appreciated.

HaywardUW commented 3 months ago

I was able to resolve this by installing VS Community, pulling libomp140.x86_x64.dll from Windows\System32 and placing it in ...GIMP\gimpenv3\Lib\site-packages\torch\lib.

gblong1 commented 3 months ago

interesting. i did notice this today, and was able to resolve it by modifying the torch revisions to be torch>=2.2.0,<2.4.0.