Is the all `t` test cases run for each subtask?

subtask

#1

Is the program run once for each subtask(all t test cases per subtask) or are the subtasks divided into ranges of test cases?


#2

Depends on the problem.
Sometimes the first way, sometimes the other way.


#3

And is this usually explicit in the problem statement?