DirectML or AMD GPU crashes on IP Adapter

[marekbujko]

Hi,
I have a problem running IP Adapter on Comfy UI with AMD GPU: RX 480 8GB.
I followed these instructions (json projects are in attachments):
https://civitai.com/articles/2345/aloeveras-instant-lora-no-training-15-read-new-info

When the process gets to:
Requested to load BaseModel
Loading 1 new model

GPU VRAM gets to 7.6 GB and then after a while a warning is displayed and ComfyUI crashes (exits in the terminal):
warnings.warn('User provided device_type of 'cuda', but CUDA is not available. Disabling')
[F D:\a_work\1\s\pytorch-directml-plugin\torch_directml\csrc\engine\dml_util.cc:118] Invalid or unsupported data type.

(venv) C:\Users\User\Documents\ComfyUI\ComfyUI>

Start commands:
python main.py --directml

Log:
(venv) C:\Users\User\Documents\ComfyUI\ComfyUI>python main.py --directml
** ComfyUI start up time: 2023-11-21 17:43:04.766768

Prestartup times for custom nodes:
0.0 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Manager

Using directml with device:
Total VRAM 1024 MB, total RAM 16350 MB
Set vram state to: NORMAL_VRAM
Device: privateuseone
VAE dtype: torch.float32
Using sub quadratic optimization for cross attention, if you have memory or speed issues try using: --use-split-cross-attention

Loading: ComfyUI-Impact-Pack (V4.31.1)

Loading: ComfyUI-Impact-Pack (Subpack: V0.3)

Loading: ComfyUI-Inspire-Pack (V0.47)

Loading: ComfyUI-Manager (V1.1.3)

ComfyUI Revision: 1707 [6ff06fa] | Released on '2023-11-21'

Downloading anime face detector...
Failed to download lbpcascade_animeface.xml so please download it in C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\IPAdapter-ComfyUI.

Import times for custom nodes:
0.0 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\IPAdapter-ComfyUI
0.0 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\Derfuu_ComfyUI_ModdedNodes
0.1 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Inspire-Pack
0.6 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Manager
1.3 seconds: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Impact-Pack

Starting server

To see the GUI go to: http://127.0.0.1:8188

FETCH DATA from: C:\Users\User\Documents\ComfyUI\ComfyUI\custom_nodes\ComfyUI-Manager\extension-node-map.json
got prompt
model_type EPS
adm 0
Using split attention in VAE
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
Using split attention in VAE
missing {'cond_stage_model.clip_l.logit_scale', 'cond_stage_model.clip_l.text_projection'}
left over keys: dict_keys(['model_ema.decay', 'model_ema.num_updates', 'cond_stage_model.clip_l.transformer.text_model.embeddings.position_ids'])
Requested to load CLIPVisionModelWithProjection
Loading 1 new model
Requested to load SD1ClipModel
Loading 1 new model
Requested to load BaseModel
Loading 1 new model
0%| | 0/32 [00:00<?, it/s]
C:\Users\User\Documents\ComfyUI\ComfyUI\venv\lib\site-packages\torch\amp\autocast_mode.py:204: UserWarning: User provided device_type of 'cuda', but CUDA is not available. Disabling
warnings.warn('User provided device_type of 'cuda', but CUDA is not available. Disabling')
[F D:\a_work\1\s\pytorch-directml-plugin\torch_directml\csrc\engine\dml_util.cc:118] Invalid or unsupported data type.