Updatet with newer base image, ROCm 7.1.1 and ComfyUI v.0.3.76

[Griznah]

Hi! I couldn't get Qwen Image or Z-Image-Turbo to work with your version so I updated the container. Merge if you please.
My image is also ~8GB smaller than your original one.

[joaomorenorf]

I had to add the lib requests to the requirements to be able to run the model downloader.

The code loads successfully, but when I run the example, it segfaults.

comfyui-rocm  | [ComfyUI] Model downloader starting (MODEL_DOWNLOAD=default)
comfyui-rocm  | [ComfyUI] Downloading default models (1 total)...
comfyui-rocm  | [ComfyUI] Stable Diffusion 1.5 already exists, skipping
comfyui-rocm  | [ComfyUI] default downloads completed: 1/1 successful
comfyui-rocm  | [ComfyUI] Model downloader completed successfully
comfyui-rocm  | [ComfyUI] Starting ComfyUI on port 8188...
comfyui-rocm  | Checkpoint files will always be loaded safely.
comfyui-rocm  | Total VRAM 30027 MB, total RAM 60055 MB
comfyui-rocm  | pytorch version: 2.9.1+rocm7.1.1.git351ff442
comfyui-rocm  | Set: torch.backends.cudnn.enabled = False for better AMD performance.
comfyui-rocm  | AMD arch: gfx1103
comfyui-rocm  | ROCm version: (7, 1)
comfyui-rocm  | Forcing FP32, if this improves things please report it.
comfyui-rocm  | Set vram state to: NORMAL_VRAM
comfyui-rocm  | Device: cuda:0 AMD Radeon Graphics : native
comfyui-rocm  | Enabled pinned memory 57052.0
comfyui-rocm  | Using split optimization for attention
comfyui-rocm  | Python version: 3.12.3 (main, Jan  8 2026, 11:30:50) [GCC 13.3.0]
comfyui-rocm  | ComfyUI version: 0.3.76
comfyui-rocm  | ComfyUI frontend version: 1.32.10
comfyui-rocm  | [Prompt Server] web root: /opt/venv/lib/python3.12/site-packages/comfyui_frontend_package/static
comfyui-rocm  | Total VRAM 30027 MB, total RAM 60055 MB
comfyui-rocm  | pytorch version: 2.9.1+rocm7.1.1.git351ff442
comfyui-rocm  | Set: torch.backends.cudnn.enabled = False for better AMD performance.
comfyui-rocm  | AMD arch: gfx1103
comfyui-rocm  | ROCm version: (7, 1)
comfyui-rocm  | Forcing FP32, if this improves things please report it.
comfyui-rocm  | Set vram state to: NORMAL_VRAM
comfyui-rocm  | Device: cuda:0 AMD Radeon Graphics : native
comfyui-rocm  | Enabled pinned memory 57052.0
comfyui-rocm  | setup plugin alembic.autogenerate.schemas
comfyui-rocm  | setup plugin alembic.autogenerate.tables
comfyui-rocm  | setup plugin alembic.autogenerate.types
comfyui-rocm  | setup plugin alembic.autogenerate.constraints
comfyui-rocm  | setup plugin alembic.autogenerate.defaults
comfyui-rocm  | setup plugin alembic.autogenerate.comments
comfyui-rocm  | Context impl SQLiteImpl.
comfyui-rocm  | Will assume non-transactional DDL.
comfyui-rocm  | No target revision found.
comfyui-rocm  | Starting server
comfyui-rocm  | 
comfyui-rocm  | To see the GUI go to: http://0.0.0.0:8188
comfyui-rocm  | got prompt
comfyui-rocm  | model weight dtype torch.float32, manual cast: None
comfyui-rocm  | model_type EPS
comfyui-rocm  | Using split attention in VAE
comfyui-rocm  | Using split attention in VAE
comfyui-rocm  | VAE load device: cuda:0, offload device: cpu, dtype: torch.float32
comfyui-rocm  | Requested to load SD1ClipModel
comfyui-rocm  | loaded completely; 95367431640625005117571072.00 MB usable, 235.84 MB loaded, full load: True
comfyui-rocm  | CLIP/text encoder model load device: cpu, offload device: cpu, current: cpu, dtype: torch.float16
comfyui-rocm  | /workspace/ComfyUI/comfy/samplers.py:1134: UserWarning: expandable_segments not supported on this platform (Triggered internally at /pytorch/c10/hip/HIPAllocatorConfig.h:36.)
comfyui-rocm  |   self.sigmas = self.calculate_sigmas(steps).to(self.device)
comfyui-rocm  | Requested to load BaseModel
comfyui-rocm  | loaded completely; 27734.06 MB usable, 3278.81 MB loaded, full load: True
comfyui-rocm exited with code 139 (restarting)