It doesn't run after automatic update to 0.7.2

[2026-01-10 09:32:38.177] [info] Adding extra search path checkpoints C:\Stable_Diffusion_ComfyUI\models\checkpoints
Adding extra search path classifiers C:\Stable_Diffusion_ComfyUI\models\classifiers
Adding extra search path clip C:\Stable_Diffusion_ComfyUI\models\clip

[2026-01-10 09:32:38.179] [info] Adding extra search path clip_vision C:\Stable_Diffusion_ComfyUI\models\clip_vision
Adding extra search path configs C:\Stable_Diffusion_ComfyUI\models\configs
Adding extra search path controlnet C:\Stable_Diffusion_ComfyUI\models\controlnet
Adding extra search path diffusers C:\Stable_Diffusion_ComfyUI\models\diffusers
Adding extra search path diffusion_models C:\Stable_Diffusion_ComfyUI\models\diffusion_models
Adding extra search path embeddings C:\Stable_Diffusion_ComfyUI\models\embeddings
Adding extra search path gligen C:\Stable_Diffusion_ComfyUI\models\gligen
Adding extra search path hypernetworks C:\Stable_Diffusion_ComfyUI\models\hypernetworks
Adding extra search path loras C:\Stable_Diffusion_ComfyUI\models\loras
Adding extra search path photomaker C:\Stable_Diffusion_ComfyUI\models\photomaker
Adding extra search path style_models C:\Stable_Diffusion_ComfyUI\models\style_models
Adding extra search path text_encoders C:\Stable_Diffusion_ComfyUI\models\text_encoders
Adding extra search path unet C:\Stable_Diffusion_ComfyUI\models\unet
Adding extra search path upscale_models C:\Stable_Diffusion_ComfyUI\models\upscale_models
Adding extra search path vae C:\Stable_Diffusion_ComfyUI\models\vae
Adding extra search path vae_approx C:\Stable_Diffusion_ComfyUI\models\vae_approx
Adding extra search path animatediff_models C:\Stable_Diffusion_ComfyUI\models\animatediff_models
Adding extra search path animatediff_motion_lora C:\Stable_Diffusion_ComfyUI\models\animatediff_motion_lora
Adding extra search path animatediff_video_formats C:\Stable_Diffusion_ComfyUI\models\animatediff_video_formats
Adding extra search path ipadapter C:\Stable_Diffusion_ComfyUI\models\ipadapter
Adding extra search path liveportrait C:\Stable_Diffusion_ComfyUI\models\liveportrait
Adding extra search path insightface C:\Stable_Diffusion_ComfyUI\models\insightface
Adding extra search path layerstyle C:\Stable_Diffusion_ComfyUI\models\layerstyle
Adding extra search path LLM C:\Stable_Diffusion_ComfyUI\models\LLM
Adding extra search path Joy_caption C:\Stable_Diffusion_ComfyUI\models\Joy_caption
Adding extra search path sams C:\Stable_Diffusion_ComfyUI\models\sams
Adding extra search path blip C:\Stable_Diffusion_ComfyUI\models\blip
Adding extra search path CogVideo C:\Stable_Diffusion_ComfyUI\models\CogVideo
Adding extra search path xlabs C:\Stable_Diffusion_ComfyUI\models\xlabs
Adding extra search path instantid C:\Stable_Diffusion_ComfyUI\models\instantid
Adding extra search path custom_nodes C:\Stable_Diffusion_ComfyUI\custom_nodes
Adding extra search path download_model_base C:\Stable_Diffusion_ComfyUI\models
Adding extra search path custom_nodes C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes
Setting output directory to: C:\Stable_Diffusion_ComfyUI\output
Setting input directory to: C:\Stable_Diffusion_ComfyUI\input
Setting user directory to: C:\Stable_Diffusion_ComfyUI\user

[2026-01-10 09:32:38.955] [info] [START] Security scan
[DONE] Security scan
** ComfyUI startup time:
[2026-01-10 09:32:38.957] [info] 2026-01-10 09:32:38.954
** Platform: Windows
** Python version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]
** Python executable: C:\Stable_Diffusion_ComfyUI.venv\Scripts\python.exe
** ComfyUI Path: C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI
** ComfyUI Base Folder Path: C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI

[2026-01-10 09:32:38.965] [info] ** User directory:
[2026-01-10 09:32:38.966] [info] C:\Stable_Diffusion_ComfyUI\user
** ComfyUI-Manager config path: C:\Stable_Diffusion_ComfyUI\user__manager\config.ini
** Log path: C:\Stable_Diffusion_ComfyUI\user\comfyui.log

[2026-01-10 09:32:39.502] [info] [ComfyUI-Manager] Skipped fixing the ‘comfyui-frontend-package’ dependency because the ComfyUI is outdated.

[2026-01-10 09:32:39.504] [info] [PRE] ComfyUI-Manager

[2026-01-10 09:32:39.506] [info]
Prestartup times for custom nodes:
0.0 seconds: C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\rgthree-comfy
0.0 seconds: C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\custom_nodes\ComfyUI-Easy-Use

[2026-01-10 09:32:40.505] [info] Checkpoint files will always be loaded safely.

[2026-01-10 09:32:40.614] [info] Total VRAM 24463 MB, total RAM 130479 MB

[2026-01-10 09:32:40.616] [info] pytorch version: 2.9.1+cu130
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 5090 Laptop GPU : cudaMallocAsync

[2026-01-10 09:32:40.631] [info] Using async weight offloading with 2 streams

[2026-01-10 09:32:40.633] [info] Enabled pinned memory 58715.0

[2026-01-10 09:32:40.634] [info] working around nvidia conv3d memory bug.

[2026-01-10 09:32:41.261] [info] Found comfy_kitchen backend triton: {‘available’: False, ‘disabled’: True, ‘unavailable_reason’: ‘ImportError: DLL load failed while importing libtriton: 지정된 모듈을 찾을 수 없습니다.’, ‘capabilities’: }

[2026-01-10 09:32:41.263] [info] Found comfy_kitchen backend eager: {‘available’: True, ‘disabled’: False, ‘unavailable_reason’: None, ‘capabilities’: [‘apply_rope’, ‘apply_rope1’, ‘dequantize_nvfp4’, ‘dequantize_per_tensor_fp8’, ‘quantize_nvfp4’, ‘quantize_per_tensor_fp8’, ‘scaled_mm_nvfp4’]}
Found comfy_kitchen backend cuda: {‘available’: True, ‘disabled’: False, ‘unavailable_reason’: None, ‘capabilities’: [‘apply_rope’, ‘apply_rope1’, ‘dequantize_nvfp4’, ‘dequantize_per_tensor_fp8’, ‘quantize_nvfp4’, ‘quantize_per_tensor_fp8’, ‘scaled_mm_nvfp4’]}

[2026-01-10 09:32:41.393] [error] Traceback (most recent call last):
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\main.py”, line 177, in

[2026-01-10 09:32:41.395] [error] import execution
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\execution.py”, line 16, in
from latent_preview import set_preview_method
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\latent_preview.py”, line 5, in
from comfy.sd import VAE
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy\sd.py”, line 13, in
import comfy.ldm.genmo.vae.model
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy\ldm\genmo\vae\model.py”, line 13, in
from comfy.ldm.modules.attention import optimized_attention
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy\ldm\modules\attention.py”, line 30, in
raise e
File “C:\Users\mayit\AppData\Local\Programs\ComfyUI\resources\ComfyUI\comfy\ldm\modules\attention.py”, line 23, in
from sageattention import sageattn
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\sageattention_init_.py”, line 1, in
from .core import sageattn, sageattn_varlen
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\sageattention\core.py”, line 2, in
import triton
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\triton_init_.py”, line 8, in
from .runtime import (
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\triton\runtime_init_.py”, line 1, in
from .autotuner import (Autotuner, Config, Heuristics, autotune, heuristics)
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\triton\runtime\autotuner.py”, line 11, in

[2026-01-10 09:32:41.397] [error] from .. import knobs
File “C:\Stable_Diffusion_ComfyUI.venv\Lib\site-packages\triton\knobs.py”, line 15, in
from triton._C.libtriton import getenv, getenv_bool # type: ignore
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ImportError: DLL load failed while importing libtriton: The registered module could not be found.

[2026-01-10 09:32:41.711] [error] [W110 09:32:41.000000000 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (function operator ())

-

There were no issues with version 0.7.1 before the update. Is there a way to roll back to the previous version, at least with comfyui not working at all?

Since the update is automatic, I couldn’t block it. This problem occurred right after I turned my computer off last night and turned it back on today, leaving me with no time to deal with it.

업데이트 전인 0.7.1 버전에서는 전혀 문제가 없었는데, 최소한 comfyui가 아예 작동하지 않는 현 상태에서 이전 버전으로 롤백하는 방법이라도 없을까요?

업데이트가 자동이다 보니, 막지도 못 하고 어제 저녁 컴퓨터를 끈 이후, 오늘 다시 켠 후에 바로 나타난 현상이라 대처할 시간이 없었습니다.

3 Likes

I am having the same problem.

2 Likes

Same here.
Seems to be it’s all started after pytorch 2.9.1 + cu130 update. I got cu130, in PATH, still not working.
upd1: Definitely dependency issue, maybe devs try to wrote some system requirements for desktop version, like Python version and such stuff which not install automatically (mostly)?
upd2: In nearest topic read that comfy told about outdated videodriver, so I updated it and build 0.7.1 worked. Still don’t know what to do with nodes which requires torch2.8.2 + cu128.

He-he, go back to old builds ^_^/.

1 Like

i have the same issue here. when you say i have to go back to old build , do you downgrade only python ? i’m very new since 3-4- month on comfyui. if i ask to GPT is says to …

(.venv) C:\ComfyUI>
pip uninstall torch torchvision torchaudio -y
pip install torch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 --index-url https://download.pytorch.org/whl/cu121

python -c “import torch; print(torch.version)”

and then it could be solve.

Is that correct or i have to download V 0.7.1. Releases · Comfy-Org/ComfyUI · GitHub unzipe and restart everything. ?

In case of GPT, he provides outdated versions of stuff like torch and cuda, because he doesn’t know that newer versions are exists.

Hm… Your link leads to portable versions, so they had own python and venv, just unzip and use archive depending which cuda you have installed before (archive without markings - cu130).

I spoke about desktop version where it’s hard to start working my old venv with torch2.8.2+cu128 (and tons of useful stuff like sage attention and tensorrt, and custom_nodes) on updated to torch2.9+cu130 desktop version (even if I do clean install and got to work 0.7.2 desktop build - many of custom_nodes just don’t work with it). So I just installed previous version 0.6.2 and recover my working venv from backup I’ve done before all updatings.

For portable version it may work from a box of you have cuda 12.4, 12.8 or 13.0 installed.

更新你的nvidia驱动

更新你的nvidia驱动

This did not resolve the issue.