Out of Memory on Bernini

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

I was running Bernini-R and I got a OOM error.
I did reduce the output size and length down to 256x256 65f and still got the issues.
I did managed to run it once a week ago, on 720p 65f. since then I did update comfyui.

### Actual Behavior

OOM error.

### Steps to Reproduce

[cars_bernidi.json](https://github.com/user-attachments/files/29457133/cars_bernidi.json)

### Debug Logs

```powershell
D:\ComfyUI_windows_portable>.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --disable-all-custom-nodes
[INFO] setup plugin alembic.autogenerate.schemas
[INFO] setup plugin alembic.autogenerate.tables
[INFO] setup plugin alembic.autogenerate.types
[INFO] setup plugin alembic.autogenerate.constraints
[INFO] setup plugin alembic.autogenerate.defaults
[INFO] setup plugin alembic.autogenerate.comments
[WARNING] WARNING: You need pytorch with cu130 or higher to use optimized CUDA operations.
[INFO] Found comfy_kitchen backend triton: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'apply_rope_split_half', 'apply_rope_split_half1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8']}
[INFO] Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'apply_rope_split_half', 'apply_rope_split_half1', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'gemv_awq_w4a16', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8']}
[INFO] Found comfy_kitchen backend cuda: {'available': True, 'disabled': True, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'apply_rope_split_half', 'apply_rope_split_half1', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'gemv_awq_w4a16', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'quantize_svdquant_w4a4', 'scaled_mm_svdquant_w4a4', 'stochastic_rounding_fp8']}
[INFO] Checkpoint files will always be loaded safely.
[INFO] Total VRAM 24575 MB, total RAM 32709 MB
[INFO] pytorch version: 2.10.0+cu128
[INFO] Set vram state to: NORMAL_VRAM
[INFO] Device: cuda:0 NVIDIA GeForce RTX 3090 : cudaMallocAsync
[INFO] Using async weight offloading with 2 streams
[INFO] Enabled pinned memory 13083.0
[INFO] working around nvidia conv3d memory bug.
[INFO] Using pytorch attention
aimdo: src-win/cuda-detour.c:38:INFO:aimdo_setup_hooks: installing 6 hooks
aimdo: src-win/cuda-detour.c:28:DEBUG:install_hook_entries: hooks successfully installed
aimdo: src-win/shmem-detect.c:80:INFO:comfy-aimdo WDDM adapter match: NVIDIA GeForce RTX 3090 runtime_luid=00000000:00011fdf dxgi_luid=00000000:00011fdf
aimdo: src/control.c:248:INFO:comfy-aimdo inited for GPU: NVIDIA GeForce RTX 3090 (VRAM: 24575 MB)
[INFO] DynamicVRAM support detected and enabled
[INFO] Python version: 3.12.10 (tags/v3.12.10:0cc8128, Apr 8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
[INFO] ComfyUI version: 0.26.0
[INFO] comfy-aimdo version: 0.4.10
[INFO] comfy-kitchen version: 0.2.10
[INFO] comfyui-frontend-package version: 1.45.19
[INFO] comfyui-workflow-templates version: 0.10.2
[INFO] comfyui-embedded-docs version: 0.5.5
[INFO] comfy-kitchen version: 0.2.10
[INFO] comfy-aimdo version: 0.4.10
[INFO] [Prompt Server] web root: D:\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
[INFO] Asset seeder disabled
[INFO] Skipping loading of custom nodes
[INFO] Context impl SQLiteImpl.
[INFO] Will assume non-transactional DDL.
[INFO] Using RAM pressure cache.
[INFO] Starting server

[INFO] To see the GUI go to: http://127.0.0.1:8188
[ERROR] Exception in callback _ProactorBasePipeTransport._call_connection_lost(None)
handle:
Traceback (most recent call last):
File "asyncio\events.py", line 88, in _run
File "asyncio\proactor_events.py", line 165, in _call_connection_lost
ConnectionResetError: [WinError 10054] Une connexion existante a dû être fermée par l’hôte distant
[INFO] got prompt
[WARNING] invalid prompt: {'type': 'missing_node_type', 'message': "Node 'ID #341' has no class_type. The workflow may be corrupted or a custom node is missing.", 'details': "Node ID '#341'", 'extra_info': {'node_id': '341', 'class_type': None, 'node_title': None}}
[INFO] got prompt
[INFO] Using pytorch attention in VAE
[INFO] Using pytorch attention in VAE
[INFO] VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[INFO] Found quantization metadata version 1
[INFO] Using MixedPrecisionOps for text encoder
[INFO] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[INFO] Requested to load WanTEModel
[INFO] Model WanTEModel prepared for dynamic VRAM loading. 6419MB Staged. 0 patches attached. Force pre-loaded 73 weights: 488 KB.
[INFO] Model WanTEModel prepared for dynamic VRAM loading. 6419MB Staged. 0 patches attached. Force pre-loaded 73 weights: 488 KB.
[INFO] Requested to load WanVAE
[INFO] Model WanVAE prepared for dynamic VRAM loading. 241MB Staged. 0 patches attached. Force pre-loaded 60 weights: 61 KB.
[INFO] Model WanVAE prepared for dynamic VRAM loading. 241MB Staged. 0 patches attached. Force pre-loaded 60 weights: 61 KB.
[INFO] Model WanVAE prepared for dynamic VRAM loading. 241MB Staged. 0 patches attached. Force pre-loaded 60 weights: 61 KB.
[INFO] Found quantization metadata version 1
[INFO] Detected mixed precision quantization
[INFO] Using mixed precision operations
[INFO] Native ops: , emulated ops: nvfp4, mxfp8, float8_e4m3fn, float8_e5m2
[INFO] model weight dtype torch.float16, manual cast: torch.float16
[INFO] model_type FLOW
[INFO] Found quantization metadata version 1
[INFO] Detected mixed precision quantization
[INFO] Using mixed precision operations
[INFO] Native ops: , emulated ops: nvfp4, mxfp8, float8_e4m3fn, float8_e5m2
[INFO] model weight dtype torch.float16, manual cast: torch.float16
[INFO] model_type FLOW
[INFO] Requested to load WAN21
[INFO] 0 models unloaded.
[INFO] Model WAN21 prepared for dynamic VRAM loading. 14268MB Staged. 894 patches attached. Force pre-loaded 160 weights: 1603 KB.
0%| | 0/3 [00:09

Start with the attached cars_bernidi.json workflow and the OOM traceback, following the execution path through comfy_extras/nodes_custom_sampler.py, comfy/sample.py, and comfy/samplers.py. Compare the reported 24 GB device limit and 61.87 GiB allocation request with the 256x256, 65-frame reproduction; done means the cause is identified and the workflow no longer fails unexpectedly under the stated setup.

Written by the indexing model from the issue text.