Extrapolating from BrowserUk's estimate in Re^5: Search for identical substrings (58 hours for the 300/3k data set) and my estimate that this code is about 7000 times faster than that, the total run time should be of the order of 30 seconds. If it is more than an hour something is very wrong. Even if it is more than a few minutes our understanding is flawed or there is a bug that wasn't shown by the six string data set.