TensorRT run slower on windows 11 than Ubuntu

Description

We run the same onnx using trtexec on Win11 and Ubuntu22.04 (same machine, dual system). On Ubuntu22.04, trtexec costs 0.9ms, while on Windows 11, trtexec costs 1.8~2.0ms. Is it a normal result? why the performance of tensorrt on windows is much slower ?

Environment

TensorRT Version: 8.6.1
GPU Type: RTX 3060 laptop
Nvidia Driver Version: 551.78
CUDA Version: 12.4
CUDNN Version: 8.9
Operating System + Version: Win11 & Ubuntu22.04
Baremetal or Container (if container which image + tag): Baremetal

Relevant Files

ik_hand.zip (2.2 MB)

Steps To Reproduce

trtexec --onnx=ik_hand.onnx --saveEngine=ik_hand.trt --workspace=4096