You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Inference without enabling the ADetailer script will use 24 threads every time.
When ADetailer script is enabled in this session, the first pass (where it generates with original prompts and settings) of the first generated image will use 24 threads, the "ADetailer" pass (where it masks the generated image and inpaints it) will only use 8 threads. All subsequent inferences will only use 8 threads from now on, even if I disable ADetailer.
Steps to reproduce
Have a CPU that has more than 8 threads
Launch webui in CPU only mode
Generate any image w/ ADetailer script
Observe the CPU usage during each pass
Screenshots
No response
Console logs, from start to end.
Python 3.10.12 (main, Nov 6 2024, 20:22:13) [GCC 11.4.0]
Version: v1.10.1
Commit hash: 82a973c04367123ae98bd9abdf80d9eda9b910e2
ControlNet init warning: Unable to install insightface automatically. Please try run `pip install insightface` manually.
Launching Web UI with arguments: --no-download-sd-model --listen --enable-console-prompts --enable-insecure-extension-access --use-cpu all --precision autocast --no-half --skip-torch-cuda-test
No module 'xformers'. Proceeding without it.
Warning: caught exception 'No CUDA GPUs are available', memory monitor disabled
[-] ADetailer initialized. version: 24.11.1, num models: 10
....(Everything is normal with no exceptions)
Using the latest version of both SDWebUI (82a973c) and adetailer (v24.11.1).
The CPU is an Intel Xeon w7-3455 (24C48T), but I think any CPU with more than 8 cores should be able to reproduce this problem.
Describe the bug
sd-webui is ran in CPU only mode:
Inference without enabling the ADetailer script will use 24 threads every time.
When ADetailer script is enabled in this session, the first pass (where it generates with original prompts and settings) of the first generated image will use 24 threads, the "ADetailer" pass (where it masks the generated image and inpaints it) will only use 8 threads. All subsequent inferences will only use 8 threads from now on, even if I disable ADetailer.
Steps to reproduce
Screenshots
No response
Console logs, from start to end.
List of installed extensions
adetailer
tagcomplete
dynamic prompt
controlnet
webui-state
The text was updated successfully, but these errors were encountered: