workflow hanging at DDIM sampler step

[Draxredd]

using safetensor versions of SDXL

Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
making attention of type 'vanilla' with 512 in_channels
missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}
left over keys: dict_keys(['denoiser.log_sigmas', 'denoiser.sigmas'])
v_prediction False

then hangs at DDIM sampler start, kinda works with unipc/karras

maybe directML related :
torch.Size([1, 1280]) 4096 4096 0 0 4096 4096
torch.Size([1, 1280]) 4096 4096 0 0 4096 4096
15%|████████████▊ | 2/13 [00:04<00:23, 2.16s/it]E:\StableDiffusion\ComfyUI\comfy\extra_samplers\uni_pc.py:666: UserWarning: The operator 'aten::_linalg_solve_ex.result' is not currently supported on the DML backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at D:\a_work\1\s\pytorch-directml-plugin\torch_directml\csrc\dml\dml_cpu_fallback.cpp:17.)
rhos_c = torch.linalg.solve(R, b)

[SytanSD]

This is a very strange issue. I am sorry for the delay, I recently got a new PC and have been all over the place trying to get everything working on it again

As for this issue, I have honestly never seen any issues like this before. This would likely be something better discussed with Comfy himself, as I don't believe this is a workflow issue, but rather an install/driver issue

[mr-bo-jangles]

Switching to Euler_a and normal scheduler fixes this for me. I am also using DirectML as a backend