ImportError: arg(): could not convert default argument 'workspace: tv::Tensor' into a Python object

[ifilipis]

Issue from the current commit when built from source with DISABLE_JIT=1 f0ffe61

!!! Exception during processing !!! arg(): could not convert default argument 'workspace: tv::Tensor' in method '<class 'spconv.core_cc.csrc.sparse.convops.gemmops.GemmTunerSimple'>.run_with_tuned_result' into a Python object (type not registered yet?)
Traceback (most recent call last):
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 323, in execute
    output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
                                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 198, in get_output_data
    return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\custom_nodes.py", line 34, in load
    self.model = TrellisImageTo3DPipeline.from_pretrained(model_name)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\pipelines\trellis_image_to_3d.py", line 55, in from_pretrained
    pipeline = super(TrellisImageTo3DPipeline, TrellisImageTo3DPipeline).from_pretrained(path)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\pipelines\base.py", line 40, in from_pretrained
    k: models.from_pretrained(f"{path}/{v}")
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\models\__init__.py", line 59, in from_pretrained
    model = __getattr__(config['name'])(**config['args'], **kwargs)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\models\structured_latent_vae\decoder_mesh.py", line 108, in __init__
    SparseSubdivideBlock3d(
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\models\structured_latent_vae\decoder_mesh.py", line 43, in __init__
    sp.SparseConv3d(channels, self.out_channels, 3, indice_key=f"res_{self.out_resolution}"),
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "E:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI_TRELLIS\extensions\trellis\modules\sparse\conv\conv_spconv.py", line 11, in __init__
    import spconv.pytorch as spconv
  File "E:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\spconv\pytorch\__init__.py", line 7, in <module>
    from spconv.pytorch.core import SparseConvTensor
  File "E:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\spconv\pytorch\core.py", line 21, in <module>
    from spconv.tools import CUDAKernelTimer
  File "E:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\spconv\tools.py", line 16, in <module>
    from spconv.cppconstants import CPU_ONLY_BUILD
  File "E:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\spconv\cppconstants.py", line 15, in <module>
    import spconv.core_cc as _ext
ImportError: arg(): could not convert default argument 'workspace: tv::Tensor' in method '<class 'spconv.core_cc.csrc.sparse.convops.gemmops.GemmTunerSimple'>.run_with_tuned_result' into a Python object (type not registered yet?)

[MrForExample]

Same here, any solution?
Installed from PyPI through pip install spconv-cu124 for python 3.12 on windows 11

[MrForExample]

After I comment out the line 15 in spconv\cppconstants.py it simply throws same error for the next line:

from spconv.core_cc.csrc.sparse.all import SpconvOps
ImportError: arg(): could not convert default argument 'workspace: tv::Tensor' in method '<class 'spconv.core_cc.csrc.sparse.convops.gemmops.GemmTunerSimple'>.run_with_tuned_result' into a Python object (type not registered yet?)

[if-ai]

that is weird I did have trouble with that part of the installation but it was because I tried installing it for cu120 first all I did was make sure to run the

(gen) PS D:\ComfyUI\custom_nodes\ComfyUI-IF_Trellis> micromamba activate gen

set your vars

cmd.exe /c "C:\Program Files (x86)\Microsoft Visual Studio\2022\BuildTools\VC\Auxiliary\Build\vcvarsall.bat" x64 "&&" powershell

I pip unistall spconv-cu120

pip install spconv-cu124

Also on powershell as admin make sure to set Set-ExecutionPolicy Unrestricted

If possible use Wezterm for win
https://wezfurlong.org/wezterm/install/windows.html

winget install wez.wezterm

I am in python

(gen) PS D:\ComfyUI> python --version
Python 3.12.5
(gen) PS D:\ComfyUI>

I made a full guide for setting up from the start

https://ko-fi.com/post/Installing-Triton-and-Sage-Attention-Flash-Attenti-P5P8175434

[MrForExample]

Can confirm this error is resolved in spconv 2.3.8, pip install spconv-cu124 now works as expected for python 3.12 on windows 11