02-24-2015, 01:24 AM
Not a very astute observation. Obviously you have to give the GPU enough work to do. If you only feed 120 candidates to an algorithm that can pull somewhere in the KH/s range, then you're obviously starving the GPU. But this doesn't really matter if that's the attack you want to run, since the attack will complete instantly. The reported speed is irrelevant.
This varies by algorithm though. For fast hashes it doesn't matter how large your wordlist is, you'll never gain acceleration if you don't have a modifier. For extremely slow hashes, you might be able to achieve full acceleration with less than five candidates.
And as long as your mask is properly constructed, you will always achieve full acceleration on mask attacks since candidates are generated on the GPU.
This varies by algorithm though. For fast hashes it doesn't matter how large your wordlist is, you'll never gain acceleration if you don't have a modifier. For extremely slow hashes, you might be able to achieve full acceleration with less than five candidates.
And as long as your mask is properly constructed, you will always achieve full acceleration on mask attacks since candidates are generated on the GPU.