Yes, I figured it out, its impossible to fix it by saving 6 algo, so..
Yes, by using AddDirect you can effectively reduce the problem to case 1 but you loose all benefits, it is like generating random number, add it to the starting key and check the address.
IMHO, the best way is as you have done, keep only case 1 and can keep the group optimization.
(by the way, why does release sm30 load sm60 cores, is there no error here?)
Release SM30 is especially here for debugging purpose.
COMPARE
if runtime right:
196s / 125s = ~1,56
if hashrate right:
195MK/s / 105Mk/s = ~1,85
VanitySearch win!
(issue created "need symmetry when calculating batch inversion" github.com/brichard19/BitCrack/issues/188, await..)
if runtime right:
196s / 125s = ~1,56
if hashrate right:
195MK/s / 105Mk/s = ~1,85
VanitySearch win!
(issue created "need symmetry when calculating batch inversion" github.com/brichard19/BitCrack/issues/188, await..)
Thanks for the comparison
