Break points not hit in CUDA kernel code with Nsight debugger in Visual Studio 2017

I am just starting to learn Nsight CUDA debugger in Visual Studio 2017.
My GPU is NVIDIA GeForce GTX 1050 Ti with Max-Q Design.
My Nsight Version is 2021.2.1
My CUDA Toolkit is 11.4.

I was following this tutorial(Getting Started with the CUDA Debugger :: NVIDIA Nsight VSE Documentation) and using the matrixMul project.

When I clicked Nsight->CUDA start debugging(Legacy), I got a warning saying that “A CUDA context was created on a GPU that is not supported by the legacy CUDA debugger. Breakpoints will be disabled. See output window for details. Adapter: NVIDIA GeForce GTX 1050 Ti with Max-Q Design.”
Then I tried Nsight->CUDA start debugging(Next-Gen). The break points in the kernel code turned into a hollow red circle with a warning saying that the break point will not be hit.

Can I use Nsight CUDA debugger on my GPU architecture? How can I have a working breakpoint in the kernel code?