Is Unique Thread Id guaranteed for each Kernel Call in CUDA?

Question

I have recently started to work with Cuda, I have multithread, multiprocess coding experience on C++, Java and Python. In PyCuda I see example codes like this,