Some strange bug lives deep in our code. The execution of Task 4 sometimes fails with a higher number of participants and sometimes works. I think it is due to this, that we get these unreasonable results. Actually the durations of task 4 and 5 should be similar for a low number. For high numbers task 5 should do better because the key sharing of task 4 has a exponentially growth.
- Task 4 = 0.046 s
- Task 5 = 23.07 s
- Task 4 = 0.059 s
- Task 5 = 41.26 s