CUDA cores, SMs, Grids, Blocks and Threads . . . What is the relation?

I have started developing CUDA programs on my Jetson TX2, but I am completely unfamiliar with the terminology for parallel programming. I will appreciate some help understanding the terminology

First, I would like to understand the relation between CUDA cores, SMs, Grids, Blocks and Threads. All of these concepts are jumbled in my head right now.

Second, I would like to know how many SMs are in the TX2 architecture. Is it just one?

Hi raisintoe,

Please refer to CUDA Programming Guide, you should be able to get moreunderstanding from there: