Files

T

Frank 1e9d921b95 docs: research ComfyUI on ROCm and Vulkan for Strix Halo (gfx1151)

Live-verified on Dankclaw 2026-08-14:
- official rocm/comfyui image is built PYTORCH_ROCM_ARCH=gfx942;gfx950
  (datacenter only) -> torch.cuda.is_available() False on gfx1151
- AMD gfx1151-tagged PyTorch build (rocm/vllm:...gfx1151...pytorch_2.9.1)
  -> avail True, Radeon 8060S, cap (11,5); needs /dev/kfd + /dev/dri
- torch.backends.vulkan is compiled out of standard wheels; no ComfyUI
  --vulkan flag exists; Vulkan is a dead end for ComfyUI
- Lemonade Server does image gen via wrapped stable-diffusion.cpp, whose
  HIP kernels include gfx1151 -> works where ComfyUI's torch cannot

New pages: ComfyUI ROCm and Vulkan, Lemonade and sd.cpp; ComfyUI.md
qualified; compose stays --cpu until a gfx1151-enabled torch image is built

2026-08-13 20:40:47 -07:00

2.6 KiB

Raw Blame History

ComfyUI

ComfyUI is a node-based AI image generation engine. Workflows are built from nodes that load models, sample latents, and post-process images; everything is scriptable through a web UI and an API.

Host: Dankclaw — Hostname: comfyui.dankclaw.local — Port: 8188 — Docs: https://github.com/Comfy-Org/ComfyUI

Facts (from research, primary source)

Item	Value
Image	`ghcr.io/comfyanonymous/comfyui:latest` (community CPU build)
Port	8188 (host) → 8188 (container)
Models volume	bind mount to the model collection
CPU flag	`--cpu` — confirmed in `cli_args.py`: "To use the CPU for everything (slow)"
Known issue	ROCm wheels do not ship kernels for gfx1151 (Strix Halo) — CPU backend required

Why CPU on Dankclaw

Dankclaw is an AMD Strix Halo APU (gfx1151). Precompiled PyTorch ROCm wheels do not ship kernels for this iGPU, and the official rocm/comfyui image is built for datacenter archs only (PYTORCH_ROCM_ARCH=gfx942;gfx950). This matches the knowledge base entry: the running instance uses --cpu.

Update (2026-08-14): the blanket claim "ROCm unavailable on Strix Halo" is now qualified. AMD publishes gfx1151-tagged PyTorch builds that work on this hardware, and sd.cpp/Lemonade ship gfx1151 kernels today. See ComfyUI ROCm and Vulkan and Lemonade and sd.cpp. The CPU config below remains correct until a gfx1151-enabled torch image is built.

Command Line (from `cli_args.py`)

Flag	Meaning
`--cpu`	use CPU for everything
`--cpu-vae`	run the VAE on CPU only
`--lowvram`	text encoders on CPU (dynamic VRAM default handles this)
`--listen`	bind address (0.0.0.0 for LAN)
`--port`	listen port (8188)

Volumes

Container path	Host path (convention)
`/workspace`	`/opt/comfyui` (models, input, output, custom_nodes)

Start Fresh

The model collection starts from scratch. The first boot creates the directory structure; models are added per the knowledge base procedure.

Healthcheck

The web UI responds on the root path /. ComfyUI also exposes /system_stats.

Gotchas

Expect minutes per image on CPU; that is the documented behavior for this APU.
ComfyUI-Manager installs custom nodes into custom_nodes/ — persist that directory.
Do not pull the CUDA image; it fails to find an NVIDIA driver.

2.6 KiB Raw Blame History