Generate Engine failure of TensorRT 8.6.1.6 when running onnx to engine on GPU 5060 #4541
Description
Description
I try to run engine file generate on GPU 5060, but get the following error
[08/05/2025-22:04:29] [I] [TRT] ----------------------------------------------------------------
[08/05/2025-22:04:29] [I] [TRT] Input filename: /home/oem/Downloads/weight/superpoint_v1.onnx
[08/05/2025-22:04:29] [I] [TRT] ONNX IR version: 0.0.7
[08/05/2025-22:04:29] [I] [TRT] Opset version: 13
[08/05/2025-22:04:29] [I] [TRT] Producer name: pytorch
[08/05/2025-22:04:29] [I] [TRT] Producer version: 2.1.0
[08/05/2025-22:04:29] [I] [TRT] Domain:
[08/05/2025-22:04:29] [I] [TRT] Model version: 0
[08/05/2025-22:04:29] [I] [TRT] Doc string:
[08/05/2025-22:04:29] [I] [TRT] ----------------------------------------------------------------
[08/05/2025-22:04:29] [W] [TRT] onnx2trt_utils.cpp:374: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[08/05/2025-22:04:29] [I] [TRT] BuilderFlag::kTF32 is set but hardware does not support TF32. Disabling TF32.
[08/05/2025-22:04:29] [E] [TRT] 2: [helpers.h::smVerHex2Dig::694] Error Code 2: Internal Error (Assertion major >= 0 && major < 10 failed. )
Superpoint engine build failed
Environment
TensorRT Version: 8.6.1.6
NVIDIA GPU: 5060
NVIDIA Driver Version:570.133.07
CUDA Version: 11.8
CUDNN Version:cudnn-linux-x86_64-8.9.6.50_cuda11-archive
Operating System: Ubuntu 20.04
Relevant Files
Model link: