Question about the loop method

[Gxinhu]

Hi,

Thank you for contributing to such a great library.

I have a question regarding the loop method used in this repository. Based on the experiment conducted in the repository, which was also carried out by you, it appears that the 3D data array flat loop, test_stechcil_3d_flat, has relatively poor performance in both Cuda and OpenMP. I'm curious why it is still being used in this repository.

Have a nice day!