lllyasviel f06ba8e60b Significantly reduce thread abuse for faster model moving ...

This will move all major gradio calls into the main thread rather than random gradio threads.
This ensures that all torch.module.to() are performed in main thread to completely possible avoid GPU fragments.
In my test now model moving is 0.7 ~ 1.2 seconds faster, which means all 6GB/8GB VRAM users will get 0.7 ~ 1.2 seconds faster per image on SDXL.

2024-02-08 10:13:59 -08:00

..

config.py

i

2024-01-31 15:18:02 -08:00

controlnet.py

qol

2024-01-31 23:21:58 -08:00

diffusers_patcher.py

elegant patcher

2024-02-02 00:34:18 -08:00

forge_alter_samplers.py

Euler A Turbo

2024-02-05 14:23:42 -08:00

forge_clip.py

Revert "safer device"

2024-02-06 05:27:07 -08:00

forge_loader.py

use better context manager to fix potential problems

2024-02-08 02:00:54 -08:00

forge_sampler.py

memory_peak_estimation_modifier

2024-02-04 19:36:58 -08:00

forge_util.py

Update forge_util.py

2024-01-31 15:51:43 -08:00

forge_version.py

Update forge_version.py

2024-02-06 17:55:14 -08:00

gradio_compile.py

i

2024-01-26 00:32:54 -08:00

initialization.py

set CUDA_VISIBLE_DEVICES with args

2024-02-05 23:53:11 -08:00

main_thread.py

Significantly reduce thread abuse for faster model moving

2024-02-08 10:13:59 -08:00

ops.py

backend

2024-01-31 09:46:24 -08:00

patch_basic.py

qol

2024-01-31 23:21:58 -08:00

shared.py

better diffusers handling

2024-02-03 17:16:21 -08:00

supported_controlnet.py

masked cnet

2024-02-01 21:48:58 -08:00

supported_preprocessor.py

ini mask support

2024-02-01 21:21:41 -08:00

unet_patcher.py

Allow controlnet_model_function_wrapper

2024-02-07 20:06:23 -08:00