Could you elaborate on this? I noticed a while ago that some people were trying to solve the puzzles by collecting RIPEMD160's hashes of addresses with a certain number of prefix matches, and somehow using that information to generate areas in the puzzle keyspace to scan which (they thought) would have a higher probability of having another address with the same number of prefix matches too. Isn't that the exact same thing as using vanitysearch to find addresses with matching prefixes (i.e. 1BY8GQbnue), and taking the difference between the prefix addresses found to try and find the next/nearest address with the same prefix?

Am I missing something? Is there a more efficient/faster method to search more keys in less time, besides vanitysearch with GPUs? Or is vanitysearch + GPU's already the best available option out there?

Thanks.