Thanks a lot. After thinking a bit more about my solution I found that I checked some cases several times. Now my solution works and finishes the test also in about three seconds.

considering the requirement, you have to find what kind of checks you do that are actually doomed to lead to a dead end. Prune your implementation from those cases. The reference solution (mine) passes the tests in 3,7s, compilation time included. So you're clearly missing one (or several) important "simplifications". ;)

I am currently trying to solve this Kata using Java. But my code seems to be way to slow. I am only able to get one fifth through the random test of 500 examples. And right now I have no more ideas to make my code seriously faster. What I have found out is that especially the unsolvable examples of widths of 100 are consuming a lot of time. Have others also encountered such problems?