Trellis.2 gguf Issue

[sdbrown67]

It appears to be CUDA related, but I have CUDA 12.8 installed. Not sure!!!

Error: Traceback (most recent call last):
File "C:\Program Files\Modly\resources\api\routers\generation.py", line 153, in _run_generation
output_path = await loop.run_in_executor(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\Modly\resources\python-embed\Lib\concurrent\futures\thread.py", line 58, in run
result = self.fn(*self.args, **self.kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\Modly\resources\api\routers\generation.py", line 155, in
lambda: gen.generate(image_bytes, params, progress_cb, cancel_event)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\Modly\resources\api\services\extension_process.py", line 232, in generate
raise RuntimeError(msg.get("traceback") or msg.get("message", "Unknown error"))
RuntimeError: Traceback (most recent call last):
File "D:\Program Files\Modly\extensions\trellis2\generator.py", line 664, in _run_generate
mesh_with_voxel = self._model.run(
^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\utils_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\pipelines\trellis2_image_to_3d.py", line 1239, in run
shape_slat, res = self.sample_shape_slat_cascade(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\pipelines\trellis2_image_to_3d.py", line 801, in sample_shape_slat_cascade
hr_coords = self.models['shape_slat_decoder'].upsample(slat, upsample_times=4)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\models\sc_vaes\sparse_unet_vae.py", line 1208, in upsample
h = block(h)
^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\models\sc_vaes\sparse_unet_vae.py", line 437, in forward
return self._forward(x)
^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\models\sc_vaes\sparse_unet_vae.py", line 421, in _forward
h = self.conv(x)
^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\nn\modules\module.py", line 1751, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\nn\modules\module.py", line 1762, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\modules\sparse\conv\conv.py", line 19, in forward
return backends[config.CONV].sparse_conv3d_forward(self, x)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\trellis2_gguf\modules\sparse\conv\conv_flex_gemm.py", line 48, in sparse_conv3d_forward
out, neighbor_cache = sparse_submanifold_conv3d(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\ops\spconv\submanifold_conv3d.py", line 372, in sparse_submanifold_conv3d
return SubMConv3dFunction.apply(feats, coords, shape, neighbor_cache, weight, bias, dilation)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\torch\autograd\function.py", line 575, in apply
return super().apply(*args, **kwargs) # type: ignore[misc]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\ops\spconv\submanifold_conv3d.py", line 321, in forward
output = SubMConv3dFunction._sparse_submanifold_conv_forward(feats, neighbor_cache, weight, bias)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\ops\spconv\submanifold_conv3d.py", line 166, in _sparse_submanifold_conv_forward
output = kernels.triton.sparse_submanifold_conv_fwd_implicit_gemm_splitk(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\utils\autotuner.py", line 256, in call
best_config = self._benchmark(args, kwargs, configs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\utils\autotuner.py", line 277, in _benchmark
self.kernel(*args, **kwargs, **config)
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\kernels\triton\spconv\sparse_submanifold_conv_fwd_implicit_gemm_splitk.py", line 133, in sparse_submanifold_conv_fwd_implicit_gemm_splitk
sparse_submanifold_conv_fwd_implicit_gemm_kernel[grid](
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\triton\runtime\jit.py", line 347, in
return lambda *args, **kwargs: self.run(grid=grid, warmup=False, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\utils\autotuner.py", line 67, in run
timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\flex_gemm\utils\autotuner.py", line 67, in
timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\triton\runtime\autotuner.py", line 170, in _bench
return self.do_bench(kernel_call, quantiles=(0.5, 0.2, 0.8))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\triton\testing.py", line 148, in do_bench
cache = runtime.driver.active.get_empty_cache_for_benchmark()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\venv\Lib\site-packages\triton\backends\nvidia\driver.py", line 618, in get_empty_cache_for_benchmark
return torch.empty(int(cache_size // 4), dtype=torch.int, device='cuda')
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "C:\Program Files\Modly\resources\api\runner.py", line 160, in main
output_path = gen.generate(image_bytes, params, progress_cb, cancel_evt)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\generator.py", line 620, in generate
return self._run_generate(image_bytes, params, progress_cb, cancel_event)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\Program Files\Modly\extensions\trellis2\generator.py", line 688, in _run_generate
raise RuntimeError(
RuntimeError: GPU kernel error — your RTX 50-series (Blackwell) GPU requires a ptxas compiler that supports SM 12.x.

Fix: install CUDA Toolkit 12.8 or newer from https://developer.nvidia.com/cuda-downloads
then restart Modly. The toolkit's ptxas will be picked up automatically.

Original error: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.