Thanks for posting that. It's going to take a little more time to understand what's wrong. For the first dump, the column array has the wrong value for column 0, which makes absolutely no sense. Also the X/Y values produced for (0,0)-(0,31) are the expected values for (1, 32)-(1,63), which would suggest that somewhere, the pointer to the temporary result buffer is off by 8192 bytes. I can't find any similarities on expected vs. produced Z values. The second dump doesn't have any X/Y or Z similarities. The zeros at the end are kinda interesting too, also suggesting a bad pointer somewhere.
I'm going to add a specific test case for column-major load/store of bignums, and give you something new to try. Hang on and thanks again!
The problem with this position is that the shorter address is random happenstance, not the goal or the defined problem. By that position I could say that my generated address 1CoinsLoLBY6o9khnW95MkbW2eEZQaxTRa beats it, by having eight characters that mean something, although that was also randomness and not something that was searched for. Or I could change my forum nick to 6o9, and say it has 13 digits.
If vanitygen had an option for finding only short addresses with a vanity phrase, then it would really be something to find a 25 character address with a word you searched for in it.
Good points!
I guess it's possible to cheat like that. Maybe we should adopt Boggle or Scrabble rules, with case-sensitive bonuses? Perhaps forum handle inclusion should be limited to the beginning of a prefix?
Is a 25 character address even possible, short of having a long sequence of 1s?