How to disable 'smart memory management' when using GGUF?

[bbaudio-2025]

Custom Node Testing

• I have tried disabling custom nodes and the issue persists (see how to disable custom nodes if you need help)

Your question

I'm running Scail2 GGUF Q4, with the higher resolution 1280x720, comfyui seems enable VRAM offload automatically even there is some free VRAM. I have tried add '--disable-smart-memory' in bat file but not worked.

Logs

[INFO] Total VRAM 24563 MB, total RAM 65422 MB
[INFO] pytorch version: 2.7.0+cu128
[INFO] xformers version: 0.0.30
[INFO] Set vram state to: NORMAL_VRAM
[INFO] Disabling smart memory management
[INFO] Device: cuda:0 NVIDIA GeForce RTX 3090 Ti : cudaMallocAsync
[INFO] Using async weight offloading with 2 streams
[INFO] Enabled pinned memory 26168.0
[INFO] Using xformers attention
[WARNING] Unsupported Pytorch detected. DynamicVRAM support requires Pytorch version 2.8 or later. Falling back to legacy ModelPatcher. VRAM estimates may be unreliable especially on Windows
[INFO] Python version: 3.12.7 (tags/v3.12.7:0b05ead, Oct  1 2024, 03:06:41) [MSC v.1941 64 bit (AMD64)]
[INFO] ComfyUI version: 0.24.0
[INFO] comfy-aimdo version: 0.4.9
[INFO] comfy-kitchen version: 0.2.10
[INFO] comfyui-frontend-package version: 1.45.15
[INFO] comfyui-workflow-templates version: 0.9.98
[INFO] comfyui-embedded-docs version: 0.5.3
[INFO] comfy-kitchen version: 0.2.10
[INFO] comfy-aimdo version: 0.4.9
[INFO] [Prompt Server] web root: D:\AI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\comfyui_frontend_package\static
[INFO] Asset seeder disabled

Other

No response

[coderabbitai]

🔗 Related PRs

#14183 - Revert deprecation of non-dynamic smart memory (CORE-152 (revert)) [merged]
#14390 - main: force cudnn.benchmark to false (CORE-284) [merged]

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[rattus128]

--reserve-vram -999 --disable-smart-memory.

Will force it to try and load everything for only one model at a time. Totally unsupported and this has no guard against ooming.

[bbaudio-2025]

--reserve-vram -999 --disable-smart-memory.

Will force it to try and load everything for only one model at a time. Totally unsupported and this has no guard against ooming.

Yes it works, thankyou