How do you enable CPU only CN, while still using using DirectML for everything else?

[xzuyn]

Using --use-cpu all works at making ControlNet use my CPU, but I can't figure out how to only have ControlNet on CPU.

I tried --lowvram, and selecting the Low VRAM button, but they still use GPU and make me have an OOM error.

I'm not familiar enough A1111 to know where to start with digging through the code.

edit: okay so I've looked through it a bit, and --use-cpu controlnet seems to run it on cpu, but now I get an error upon generating the final image.

  0%|                                                                                           | 0/20 [00:03<?, ?it/s]
03:36:01-614887 ERROR    Exception: tensor.device().type() == at::DeviceType::PrivateUse1 INTERNAL ASSERT FAILED at
                         "D:\\a\\_work\\1\\s\\pytorch-directml-plugin\\torch_directml\\csrc\\dml\\DMLTensor.cpp":31,
                         please report a bug to PyTorch. unbox expects Dml at::Tensor as inputs
03:36:01-618887 ERROR    Arguments: args=('task(ekgcy8rqe51se1z)', 'a guy', 'badnegative', [], 20,
                         2, False, False, 1, 1, 6, 1, -1.0, -1.0, 0, 0, 0, False, 512, 512, False, 0.7, 2, 'Latent', 0,
                         0, 0, [], 0, False, 'MultiDiffusion', False, True, 1024, 1024, 96, 96, 48, 4, 'None', 2,
                         False, 10, 1, 1, 64, False, False, False, False, False, 0.4, 0.4, 0.2, 0.2, '', '',
                         'Background', 0.2, -1.0, False, 0.4, 0.4, 0.2, 0.2, '', '', 'Background', 0.2, -1.0, False,
                         0.4, 0.4, 0.2, 0.2, '', '', 'Background', 0.2, -1.0, False, 0.4, 0.4, 0.2, 0.2, '', '',
                         'Background', 0.2, -1.0, False, 0.4, 0.4, 0.2, 0.2, '', '', 'Background', 0.2, -1.0, False,
                         0.4, 0.4, 0.2, 0.2, '', '', 'Background', 0.2, -1.0, False, 0.4, 0.4, 0.2, 0.2, '', '',
                         'Background', 0.2, -1.0, False, 0.4, 0.4, 0.2, 0.2, '', '', 'Background', 0.2, -1.0, False,
                         512, 64, True, True, True, False, False, 7, 100, 'Constant', 0, 'Constant', 0, 4,
                         <scripts.controlnet_ui.controlnet_ui_group.UiControlNetUnit object at 0x000001FF0331BF40>,
                         <scripts.controlnet_ui.controlnet_ui_group.UiControlNetUnit object at 0x000001FF033A26B0>,
                         <scripts.controlnet_ui.controlnet_ui_group.UiControlNetUnit object at 0x000001FF03308070>,
                         False, False, 'positive', 'comma', 0, False, False, '', 0, '', [], 0, '', [], 0, '', [], True,
                         False, False, False, 0, False, None, None, False, None, None, False, None, None, False, 50)
                         kwargs={}
03:36:01-645885 ERROR    gradio call: RuntimeError
┌───────────────────────────────────────── Traceback (most recent call last) ─────────────────────────────────────────┐
│ Z:\000_Stable_Diffusion\automatic\modules\call_queue.py:34 in f                                                     │
│                                                                                                                     │
│    33 │   │   │   try:                                                                                              │
│ >  34 │   │   │   │   res = func(*args, **kwargs)                                                                   │
│    35 │   │   │   │   progress.record_results(id_task, res)                                                         │
│                                                                                                                     │
│ Z:\000_Stable_Diffusion\automatic\modules\txt2img.py:56 in txt2img                                                  │
│                                                                                                                     │
│   55 │   if processed is None:                                                                                      │
│ > 56 │   │   processed = processing.process_images(p)                                                               │
│   57 │   p.close()                                                                                                  │
│                                                                                                                     │
│                                              ... 34 frames hidden ...                                               │
│                                                                                                                     │
│ Z:\000_Stable_Diffusion\automatic\modules\dml\amp\autocast_mode.py:32 in <lambda>                                   │
│                                                                                                                     │
│   31 │   │   op = getattr(resolved_obj, func_path[-1])                                                              │
│ > 32 │   │   setattr(resolved_obj, func_path[-1], lambda *args, **kwargs: pre_forward(op, arg                       │
│   33                                                                                                                │
│                                                                                                                     │
│ Z:\000_Stable_Diffusion\automatic\modules\dml\amp\autocast_mode.py:9 in pre_forward                                 │
│                                                                                                                     │
│    8 │   if not torch.dml.is_autocast_enabled():                                                                    │
│ >  9 │   │   return forward(*args, **kwargs)                                                                        │
│   10 │   args = list(map(cast, args))                                                                               │
└─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘
RuntimeError: tensor.device().type() == at::DeviceType::PrivateUse1 INTERNAL ASSERT FAILED at
"D:\\a\\_work\\1\\s\\pytorch-directml-plugin\\torch_directml\\csrc\\dml\\DMLTensor.cpp":31, please report a bug to
PyTorch. unbox expects Dml at::Tensor as inputs