I am thinking that the "finished" state could be interpreted as
"task finished executing, waiting for other tasks to reach my
barrier". From a debug POV, it's code and data are still on the
physical core, so it makes sense to show it in the multicore
view. However other tasks that have finished before, over a
previous barrier, their core has been freed and may have been
re-allocated to something else.
I have cleaned-up the example to reflect this; i.e. removed
"finished" tasks from previous barriers.
I am thinking that the "finished" state could be interpreted as "task finished executing, waiting for other tasks to reach my barrier". From a debug POV, it's code and data are still on the physical core, so it makes sense to show it in the multicore view. However other tasks that have finished before, over a previous barrier, their core has been freed and may have been re-allocated to something else.
I have cleaned-up the example to reflect this; i.e. removed "finished" tasks from previous barriers.