TensorRT Conversion Fails with "Assertion major < 10 failed" and AttributeError: enter When Converting ONNX Model
Hi all,
I'm trying to convert ONNX models to TensorRT engines for the WarpTuber project using the provided script (scripts/all_onnx2trt.bat), but I'm running into some issues. Here are the relevant log and error messages:
[TRT] [I] No importer registered for op: GridSample3D. Attempting to import as plugin.
[TRT] [I] Successfully created plugin: GridSample3D
...
[TRT] [I] BuilderFlag::kTF32 is set but hardware does not support TF32. Disabling TF32.
[TRT] [E] 2: [helpers.h::nvinfer1::smVerHex2Dig::571] Error Code 2: Internal Error (Assertion major < 10 failed. )
Traceback (most recent call last):
File "C:\baidunetdiskdownload\program\WarpTuber\scripts\onnx2trt.py", line 161, in
main(args)
File "C:\baidunetdiskdownload\program\WarpTuber\scripts\onnx2trt.py", line 140, in main
builder.create_engine(
File "C:\baidunetdiskdownload\program\WarpTuber\scripts\onnx2trt.py", line 132, in create_engine
with self.builder.build_engine(self.network, self.config) as engine, open(engine_path, "wb") as f:
AttributeError: enter
Environment:
Windows 11
Python (from venv)
NVIDIA GPU 5080 (with CUDA)
TensorRT version: (default from WarpTuber, let me know if you need the exact version)
Running via scripts/all_onnx2trt.bat
What I tried:
The script seems to find and create the GridSample3D plugin.
The error [TRT] [E] ... Assertion major < 10 failed. appears, followed by a Python AttributeError: enter.
There are also DeprecationWarnings about using set_memory_pool_limit and build_serialized_network.
Questions:
What does the "Assertion major < 10 failed" error mean? Is it related to my GPU architecture or TensorRT version?
How can I fix the AttributeError: enter when using with self.builder.build_engine(...) as engine?
Are the DeprecationWarnings critical, or can they be ignored for now?
Has anyone successfully converted ONNX models with custom ops (like GridSample3D) to TensorRT on Windows? Any tips?
Any help or suggestions would be greatly appreciated! If you need more logs or environment details, let me know.
Thank you!