too much ram

[Uemuet]

since the update of ipadapter i get this failure

Traceback (most recent call last): File "/content/drive/MyDrive/ComfyUI/execution.py", line 151, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) File "/content/drive/MyDrive/ComfyUI/execution.py", line 81, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) File "/content/drive/MyDrive/ComfyUI/execution.py", line 74, in map_node_over_list results.append(getattr(obj, func)(slice_dict(input_data_all, i))) File "/content/drive/MyDrive/ComfyUI/custom_nodes/ComfyUI_IPAdapter_plus/IPAdapterPlus.py", line 696, in apply_ipadapter work_model, face_image = ipadapter_execute(work_model, ipadapter_model, clip_vision, ipa_args) File "/content/drive/MyDrive/ComfyUI/custom_nodes/ComfyUI_IPAdapter_plus/IPAdapterPlus.py", line 257, in ipadapter_execute img_comp_cond_embeds = encode_image_masked(clipvision, image_composition) File "/content/drive/MyDrive/ComfyUI/custom_nodes/ComfyUI_IPAdapter_plus/utils.py", line 161, in encode_image_masked pixel_values = clip_preprocess(image.to(clip_vision.load_device)).float() File "/content/drive/MyDrive/ComfyUI/comfy/clip_vision.py", line 25, in clip_preprocess image = torch.nn.functional.interpolate(image, size=(round(scale image.shape[2]), round(scale image.shape[3])), mode="bicubic", antialias=True) File "/usr/local/lib/python3.10/dist-packages/torch/nn/functional.py", line 4027, in interpolate return torch._C._nn._upsample_bicubic2d_aa(input, output_size, align_corners, scale_factors) torch.cuda.OutOfMemoryError: Allocation on device 0 would exceed allowed memory. (out of memory) Currently allocated : 28.18 GiB Requested : 12.87 GiB Device limit : 39.56 GiB Free (according to CUDA): 24.81 MiB PyTorch limit (set by user-supplied memory fraction) : 17179869184.00 GiB

[cubiq]

the last update doesn't touch the clip vision encoder. the problem is somewhere else.

[Uemuet]

i have it on all workflows, i cant render anymore over 200 frames and it comes after ipadapter

[cubiq]

the latest update doesn't change the clip vision encoder, maybe you skipped a few updates but it's not the latest. Provide a minimal workflow that triggers the problem or even better try to understand which is the exact update that started showing the issue

[Uemuet]

i have the latest version of comfyui and i have it on all my style transfer workflows

[cubiq]

• provide a minimal workflow that triggers the problem
• even better try to understand which is the update that actually started the problem, it's not the latest, it must be one of the previous

[Uemuet]

it has something to do with comfy new version and pytorch 2.3

[vanche1212]

it has something to do with comfy new version and pytorch 2.3

Wait for comfyui update？