Hi,
If I run the same solution multiple times, I get totally different results in performance. I repeat, this happens without changing the code.
Only the call to the function returning the solution should be measured (both CPU and RAM). I know that running the same piece of code again and again is not 100% deterministic because of different overheads (like scheduling, etc.), but still differences are really big. Maybe I'm missing something. Take a look at the two images, comparing especially the memory usage.

I think something should be done about this, since it misinforms people on their solution performance. Also, many people here use tricks to speed up their code (like the C++ streams sync one). If the measurement were correct, those tricks would not help that much, which is more fair.

1st run

2nd run