python使用onnxRuntime进行压测时，使用的是GPU计算，但是CPU也会上升，GPU和CPU都打满后停止压测，GPU资源迅速释放，但是CPU资源一直满负荷

### Describe the feature request

python使用onnxRuntime进行压测时，使用的是GPU计算，但是CPU也会上升，GPU和CPU都打满后停止压测，GPU资源迅速释放，但是CPU资源一直满负荷。

### Describe scenario use case

如上所述，使用的onnx runtime版本是1.10.0

Start by reproducing the GPU load test described in the issue with ONNX Runtime 1.10.0 and record CPU and GPU utilization before and after stopping the test. The issue names no model, execution provider, operating system, files, or tests, so the expected resource behavior and completion criteria need to be clarified first.

Written by the indexing model from the issue text.