Closed m8pple closed 7 years ago
One of the defensive measures used in the tests is that if things take too long, then
they are killed (using /usr/bin/timeout
). The timeout is set long enough that any
reasonable implementation should complete, but short enough that a few infinite loops
don't cause the entire batch to take a whole day.
I don't think I put any logic in to print a specific message for timeouts, so that's my best guess for what's happening here. Test 34 should take <10 secs in most case, so the execution limit is set to 60 seconds. So I'd investigate:
std::pow
is very expensive compared to a complex
multiply, so if that is still in the inner loop it will be much slower.Thank you for the detailed explanation!