Performance related to password length

[devilsadvocate]

Some time ago, since oclHashcat-plus v0.15, the password length limit was increased from 15 characters to 31 characters due to popular demand at the expense of some performance drops for fast hashes.


How difficult would it be to compile an alternate version of hashcat that re-enabled the 15 character limit in order to gain performance?

I would mostly like to see it for MD5 and SHA1, but other fast algorithms, as mentioned here, https://hashcat.net/events/p13/js-ocohaaaa.pdf would be good also.

I was reading this, https://hashcat.net/forum/thread-2543.ht...erformance and was thinking how nice it would be to have some additional performance available in an alternate version of hashcat.

How labor-some would it be to bring back the zero-based optimization, for example?  What performance gains could be expected?

[atom]

Hard to estimate. Best way to find out is to modify the kernels directly. But gutfeeling tells me it's not much because of some other optimizations that went into the code since then.