Device call in global function Parallel or serial?

Accelerated Computing CUDA CUDA Programming and Performance
1

If I have more than one device call in the global function, will they be executed in parallel or one after another?

Thanks

2

Sequence. I think so. :)