I participated in Codeforces Round #713 (Div. 3)
I had solved the Problem 1512G — Short Task, during the contest, which I believe at least ran on 3 pretests, as many participants got
WA on TC 3, during the contest.
My During-Contest AC Code ran in
Pretests Passed Verdict.
The intended complexity for the problem, was
O(10^7 log(10^7)), as described in the editorial.
My solution's complexity was the same.
After system test, my during-contest solution got TLE'ed on
TC 3, the case which was included in pretests as well. (Code)
After system test, I submitted the same solution, without changing a single word/line, it got
AC'ed in 1965 ms. (Code).
I saw the leaderboard and seems like I am not the only one to have a TLE'ed solution.
I truly agree that
~100ms non-deterministic difference in run-time is possible. But, if the intended solution requires TL so close to
2 sec., then it would have been better to put the TL as
2.5s or 3s. Also, this would have helped Python/Java solutions, which TLE'ed, to pass, and also the non-deterministic difference in run-time wouldn't have affected the solutions, with "intended" complexity.
Please, look upon this suggestion MikeMirzayanov.
Correct me, if I have wrote something wrong or contradictory.