(venv-comfy) (...)$ python3 main.py --listen --port 9000 --reserve-vram 1.0 --use-pytorch-cross-attention&
[1] 2330733
(venv-comfy) (...)$ [START] Security scan
[ComfyUI-Manager] Using `uv` as Python module for pip operations.
Using Python 3.12.3 environment at: venv-comfy
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2026-01-08 15:49:43.295
** Platform: Linux
** Python version: 3.12.3 (main, Nov  6 2025, 13:44:16) [GCC 13.3.0]
** Python executable: /mnt/storage/workspace/scripts/ComfyUI/venv-comfy/bin/python3
** ComfyUI Path: /mnt/storage/workspace/scripts/ComfyUI
** ComfyUI Base Folder Path: /mnt/storage/workspace/scripts/ComfyUI
** User directory: /mnt/storage/workspace/scripts/ComfyUI/user
** ComfyUI-Manager config path: /mnt/storage/workspace/scripts/ComfyUI/user/__manager/config.ini
** Log path: /mnt/storage/workspace/scripts/ComfyUI/user/comfyui.log
Using Python 3.12.3 environment at: venv-comfy
Using Python 3.12.3 environment at: venv-comfy

Prestartup times for custom nodes:
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/rgthree-comfy
   2.6 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-Manager

Checkpoint files will always be loaded safely.
Total VRAM 32076 MB, total RAM 64039 MB
pytorch version: 2.9.1+cu128
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 5090 : cudaMallocAsync
Using async weight offloading with 2 streams
Enabled pinned memory 60836.0
working around nvidia conv3d memory bug.
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': False, 'disabled': True, 'unavailable_reason': 'libcublasLt.so.13: cannot open shared object file: No such file or directory', 'capabilities': []}
Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
Using pytorch attention
Python version: 3.12.3 (main, Nov  6 2025, 13:44:16) [GCC 13.3.0]
ComfyUI version: 0.8.0
ComfyUI frontend version: 1.35.9
[Prompt Server] web root: /mnt/storage/workspace/scripts/ComfyUI/venv-comfy/lib/python3.12/site-packages/comfyui_frontend_package/static
Total VRAM 32076 MB, total RAM 64039 MB
pytorch version: 2.9.1+cu128
Set vram state to: NORMAL_VRAM
Device: cuda:0 NVIDIA GeForce RTX 5090 : cudaMallocAsync
Using async weight offloading with 2 streams
Enabled pinned memory 60836.0

[rgthree-comfy] Loaded 48 exciting nodes. 🎉

[rgthree-comfy] ComfyUI's new Node 2.0 rendering may be incompatible with some rgthree-comfy nodes and features, breaking some rendering as well as losing the ability to access a node's properties (a vital part of many nodes). It also appears to run MUCH more slowly spiking CPU usage and causing jankiness and unresponsiveness, especially with large workflows. Personally I am not planning to use the new Nodes 2.0 and, unfortunately, am not able to invest the time to investigate and overhaul rgthree-comfy where needed. If you have issues when Nodes 2.0 is enabled, I'd urge you to switch it off as well and join me in hoping ComfyUI is not planning to deprecate the existing, stable canvas rendering all together.

ComfyUI-GGUF: Allowing full torch compile

----------------------------------------------------------------------
[ComfyUI-Manager] NOTICE: Legacy backup exists
  - Your old Manager data was backed up to:
      /mnt/storage/workspace/scripts/ComfyUI/user/__manager/.legacy-manager-backup
  - Please verify and remove it when no longer needed.
----------------------------------------------------------------------

### Loading: ComfyUI-Manager (V3.39.1)
[ComfyUI-Manager] network_mode: public
[ComfyUI-Manager] ComfyUI per-queue preview override detected (PR #11261). Manager's preview method feature is disabled. Use ComfyUI's --preview-method CLI option or 'Settings > Execution > Live preview method'.
### ComfyUI Revision: 4484 [fc0cb10b] *DETACHED | Released on '2026-01-07'
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/alter-list.json
-------------------------------------------------
  Node 'Pipeline SuperScaler' chargé avec succès
-------------------------------------------------
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/model-list.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/github-stats.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/extension-node-map.json
[ComfyUI-Manager] default cache updated: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json
Warning: Could not load sageattention: No module named 'sageattention'
sageattention package is not installed, sageattention will not be available
WanVideoWrapper WARNING: FantasyPortrait nodes not available due to error in importing them: No module named 'onnx'

Import times for custom nodes:
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/websocket_image_save.py
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyMath
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/comfyui-kjnodes
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/comfyui_essentials
   0.0 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/rgthree-comfy
   0.1 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-GGUF
   0.1 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-LTXVideo
   0.3 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-SuperScaler
   0.4 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-Manager
   0.8 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/comfyui-videohelpersuite
   2.5 seconds: /mnt/storage/workspace/scripts/ComfyUI/custom_nodes/ComfyUI-WanVideoWrapper

Context impl SQLiteImpl.
Will assume non-transactional DDL.
No target revision found.
Starting server

To see the GUI go to: http://0.0.0.0:9000
To see the GUI go to: http://[::]:9000
FETCH ComfyRegistry Data: 5/118
FETCH ComfyRegistry Data: 10/118
got prompt
Using a slow image processor as `use_fast` is unset and a slow processor was saved with this model. `use_fast=True` will be the default behavior in v4.52, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with `use_fast=False`.
Loaded processor from /mnt/storage/workspace/scripts/ComfyUI/models/text_encoders - enhancement enabled
`torch_dtype` is deprecated! Use `dtype` instead!
Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:00<00:00, 92.37it/s]
FETCH ComfyRegistry Data: 15/118
FETCH ComfyRegistry Data: 20/118
FETCH ComfyRegistry Data: 25/118
FETCH ComfyRegistry Data: 30/118
FETCH ComfyRegistry Data: 35/118
CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
Requested to load _LTXVGemmaTextEncoderModel
FETCH ComfyRegistry Data: 40/118
FETCH ComfyRegistry Data: 45/118
FETCH ComfyRegistry Data: 50/118
FETCH ComfyRegistry Data: 55/118
FETCH ComfyRegistry Data: 60/118
FETCH ComfyRegistry Data: 65/118
loaded completely; 29018.80 MB usable, 27885.96 MB loaded, full load: True
Token indices sequence length is longer than the specified maximum sequence length for this model (1043 > 1024). Running this sequence through the model will result in indexing errors
Setting `pad_token_id` to `eos_token_id`:1 for open-end generation.
FETCH ComfyRegistry Data: 70/118
FETCH ComfyRegistry Data: 75/118
FETCH ComfyRegistry Data: 80/118
Found quantization metadata version 1
Detected mixed precision quantization
Using mixed precision operations
model weight dtype torch.bfloat16, manual cast: torch.bfloat16
model_type FLUX
unet unexpected: ['audio_embeddings_connector.learnable_registers', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.k_norm.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.q_norm.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_k.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_k.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_out.0.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_out.0.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_v.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.attn1.to_v.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.ff.net.0.proj.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.ff.net.0.proj.weight', 'audio_embeddings_connector.transformer_1d_blocks.0.ff.net.2.bias', 'audio_embeddings_connector.transformer_1d_blocks.0.ff.net.2.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.k_norm.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.q_norm.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_k.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_k.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_out.0.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_out.0.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_q.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_q.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_v.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.attn1.to_v.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.ff.net.0.proj.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.ff.net.0.proj.weight', 'audio_embeddings_connector.transformer_1d_blocks.1.ff.net.2.bias', 'audio_embeddings_connector.transformer_1d_blocks.1.ff.net.2.weight', 'video_embeddings_connector.learnable_registers', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.k_norm.weight', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.q_norm.weight', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_k.bias', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_k.weight', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_out.0.bias', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_out.0.weight', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.bias', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_q.weight', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_v.bias', 'video_embeddings_connector.transformer_1d_blocks.0.attn1.to_v.weight', 'video_embeddings_connector.transformer_1d_blocks.0.ff.net.0.proj.bias', 'video_embeddings_connector.transformer_1d_blocks.0.ff.net.0.proj.weight', 'video_embeddings_connector.transformer_1d_blocks.0.ff.net.2.bias', 'video_embeddings_connector.transformer_1d_blocks.0.ff.net.2.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.k_norm.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.q_norm.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_k.bias', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_k.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_out.0.bias', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_out.0.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_q.bias', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_q.weight', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_v.bias', 'video_embeddings_connector.transformer_1d_blocks.1.attn1.to_v.weight', 'video_embeddings_connector.transformer_1d_blocks.1.ff.net.0.proj.bias', 'video_embeddings_connector.transformer_1d_blocks.1.ff.net.0.proj.weight', 'video_embeddings_connector.transformer_1d_blocks.1.ff.net.2.bias', 'video_embeddings_connector.transformer_1d_blocks.1.ff.net.2.weight']
FETCH ComfyRegistry Data: 85/118
FETCH ComfyRegistry Data: 90/118
VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
no CLIP/text encoder weights in checkpoint, the text encoder model will not be loaded.
Requested to load LTXAV
FETCH ComfyRegistry Data: 95/118
FETCH ComfyRegistry Data: 100/118
Unloaded partially: 20191.75 MB freed, 7704.52 MB remains loaded, 337.50 MB buffer reserved, lowvram patches: 0
FETCH ComfyRegistry Data: 105/118
FETCH ComfyRegistry Data: 110/118
FETCH ComfyRegistry Data: 115/118
FETCH ComfyRegistry Data [DONE]
[ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes
FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]
[ComfyUI-Manager] All startup tasks have been completed.
loaded completely; 23004.58 MB usable, 20541.27 MB loaded, full load: True
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:02<00:00,  2.87it/s]
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:01<00:00,  1.77it/s]
Requested to load AudioVAE
loaded completely; 2450.96 MB usable, 415.20 MB loaded, full load: True
Processing temporal chunk: 0:16 (16 latent frames)
Processing VAE decode tile at row 0, col 0:
  Position: (0:7, 0:7)
  Size: 7x7
Requested to load VideoVAE
Unloaded partially: 562.50 MB freed, 7142.02 MB remains loaded, 337.50 MB buffer reserved, lowvram patches: 0
loaded completely; 2596.27 MB usable, 2331.69 MB loaded, full load: True
Processing VAE decode tile at row 0, col 1:
  Position: (0:7, 3:10)
  Size: 7x7
Processing VAE decode tile at row 0, col 2:
  Position: (0:7, 6:13)
  Size: 7x7
Processing VAE decode tile at row 0, col 3:
  Position: (0:7, 9:16)
  Size: 7x7
Processing VAE decode tile at row 1, col 0:
  Position: (3:10, 0:7)
  Size: 7x7
Processing VAE decode tile at row 1, col 1:
  Position: (3:10, 3:10)
  Size: 7x7
Processing VAE decode tile at row 1, col 2:
  Position: (3:10, 6:13)
  Size: 7x7
Processing VAE decode tile at row 1, col 3:
  Position: (3:10, 9:16)
  Size: 7x7
Processing VAE decode tile at row 2, col 0:
  Position: (6:13, 0:7)
  Size: 7x7
Processing VAE decode tile at row 2, col 1:
  Position: (6:13, 3:10)
  Size: 7x7
Processing VAE decode tile at row 2, col 2:
  Position: (6:13, 6:13)
  Size: 7x7
Processing VAE decode tile at row 2, col 3:
  Position: (6:13, 9:16)
  Size: 7x7
Processing VAE decode tile at row 3, col 0:
  Position: (9:16, 0:7)
  Size: 7x7
Processing VAE decode tile at row 3, col 1:
  Position: (9:16, 3:10)
  Size: 7x7
Processing VAE decode tile at row 3, col 2:
  Position: (9:16, 6:13)
  Size: 7x7
Processing VAE decode tile at row 3, col 3:
  Position: (9:16, 9:16)
  Size: 7x7
Prompt executed in 195.59 seconds