inference takes longer time than previous version failure of TensorRT 10.3 when running trt model on GPU orin

Description

I tried to run my model on GPU, but it fails with the problem below:
It takes longer time when doing the inference than Trt8.6.2.
Here is the log of Trt10.3:

trt10.3.0.txt

vs the log of Trt8.6.2:

trt8.6.2.txt

Environment

TensorRT Version: 10.3.0

NVIDIA GPU: orin

NVIDIA Driver Version: 540.4.0

CUDA Version: 12.6

CUDNN Version:

Operating System: Ubuntu 22.04.5 LTS

Python Version (if applicable): 2.7

Tensorflow Version (if applicable):

PyTorch Version (if applicable):

Baremetal or Container (if so, version):

jetpack : 6.2

Relevant Files

Model link:

Steps To Reproduce

Commands or scripts:

Have you tried the latest release?:

Can this model run on other frameworks? For example run ONNX model with ONNXRuntime (polygraphy run <model.onnx> --onnxrt):

Start with the attached trt10.3.0.txt and trt8.6.2.txt logs, then review the reported TensorRT, Orin, JetPack, CUDA, driver, and operating-system versions. Reproduction requires the missing model link and exact commands or scripts; done means a reproducible comparison that identifies the source of the inference-time regression.