Could this be used with word probability?

[polarathene]

Would it be possible to use a similar technique with Markov chains to apply to dictionary key selection like statsprocessor is currently doing for single letters with masks?

If out of scope and I were to try implement a processor like this myself, do I have the CPU generate these in a batch to send off to the GPU? Or does the wordlist need to be uploaded to the GPU prior and have the GPU somehow do the word concatenation generation using the markov chains?