close
Skip to content

COMFYUI.md

Latest commit

 

History

History
113 lines (84 loc) · 3.3 KB

COMFYUI.md

File metadata and controls

113 lines (84 loc) · 3.3 KB
Raw

ComfyUI usage

The default Omni image runs upstream ComfyUI on one Intel XPU. Models are not bundled in the image.

Start the server

Mount an existing ComfyUI model directory and start ComfyUI directly. This is the recommended default when the workflow fits in XPU memory:

IMAGE=intel/llm-scaler-omni:0.2.0-b1
CONTAINER_NAME=comfyui

sudo docker run -itd \
    --privileged \
    --device=/dev/dri \
    --network=host \
    --shm-size=64g \
    --name="$CONTAINER_NAME" \
    --workdir=/llm/ComfyUI \
    -v /path/to/comfyui_models:/llm/ComfyUI/models \
    "$IMAGE" \
    python main.py

The release image supports Intel Arc B-series/Battlemage GPUs.

The default server is available at http://127.0.0.1:8188. Append --listen 0.0.0.0 when remote access is required, and append --enable-manager when the integrated Node Manager is needed.

DynamicVRAM for memory-constrained workflows

Use the supplied entrypoint only when a workflow has a known or observed XPU out-of-memory risk:

/llm/entrypoints/start_comfyui.sh

The entrypoint enables ComfyUI DynamicVRAM, backed by the image's pinned AIMDO XPU/Level Zero allocator, enables Node Manager, and reserves 4 GiB of XPU memory. DynamicVRAM stages, unloads, and reloads model weights to preserve activation headroom during model switching or text re-encoding. This can avoid OOM failures, but the additional weight-management work can reduce performance when the workflow already fits in XPU memory. The reserve can be changed with OMNI_COMFYUI_RESERVE_VRAM_GB when required by a specific workload.

Additional ComfyUI arguments are forwarded by the entrypoint. For example:

/llm/entrypoints/start_comfyui.sh --disable-smart-memory

Models and workflows

Place model files under the standard /llm/ComfyUI/models subdirectories used by their loader nodes. Use the model's official ComfyUI documentation for the exact file names and directory:

The focused image deliberately does not copy omni/workflows or omni/example_inputs. This prevents stale workflow snapshots from replacing maintained upstream templates.

Included custom nodes

The focused image installs pinned revisions of:

  • ComfyUI Manager;
  • VideoHelperSuite;
  • Easy-Use;
  • KJNodes;
  • CacheDiT;
  • ComfyUI-GGUF-XPU;
  • ComfyUI-nunchaku-XPU;
  • ControlNet auxiliary nodes;
  • ComfyUI-OmniXPU.

The Dockerfile is the source of truth for exact revisions. Installing or updating nodes through ComfyUI Manager changes the running container and is not part of the reproducible image build.

Omni XPU switches

ComfyUI-OmniXPU adapters are enabled by default and fall back to the original ComfyUI path when a capability or input is unsupported. Common switches are:

OMNIXPU_ENABLE=0
OMNIXPU_ATTENTION=0
OMNIXPU_NORM=0
OMNIXPU_FP8_GEMM=0
OMNIXPU_INT8_FFN=0

See ComfyUI-OmniXPU for adapter behavior, diagnostics, and opt-in legacy workarounds.

Outputs

Mount /llm/ComfyUI/output when generated files must survive container removal:

-v /path/to/comfyui_output:/llm/ComfyUI/output

Input files can similarly be mounted at /llm/ComfyUI/input.