CUDA & OpenCL similarities
I am going through the book "Programming massively parallel processors" and in chapter 4 it mentions that,
"in general, a grid is organized as a 2D array of blocks. Each array of blocks is organized into 3D array of threads."
Now if i map the above statement to the corresponding OpenCL concept , i can write as follows:
"In general, a NDRange(index space) is organized as 2D array of work-groups. Each array of work-groups is organized into 3D array of work-items."
Am i right in the mapping?
Can we have NDRange organized as 3D array of work-groups and in turn each work-group organized into 3D array of work-items?