Post
Topic
Board Bitcoin Discussion
Re: Bitcoin puzzle transaction ~32 BTC prize to who solves it
by
kTimesG
on 14/03/2025, 11:27:20 UTC
Actually is there a fully random scanning software for GPU ? Tools like keyhunt cuda or rotor they are not fully random...

What do you understand by "fully random scanning"? I hope it's not "hey, let's test a new random key at every step" lol. That would be dumb to do.

I mean fully random within range - not pseudorandom or rekey every x mkeys.

Great. The only sure thing that will happen is that you'll check the same keys, because of the birthday paradox.

Let's pick Puzzle 68. Split it in ranges of 1 keys. Truly random picks: you'll scan a previously scanned key somewhere after 2**38.5 scanned keys. After 2**67 picks (size of keyspace), you'll have scanned 64% of all the possible keys, and 36% of all keys remain unscanned.

Split it in ranges of 2**58 keys. Truly random range picks: you'll scan a previously scanned range somewhere after 23 scanned ranges. After 2**9 range picks (size of all ranges), you'll have scanned 64% of all possible ranges, and 36% of all ranges remain unscanned.

Unless you have a way to shuffle 2**67 keys, in a random order, and actually storing that order without repeats. IDK of any formula for that, and even if it would exist, it means exponentially increasing the difficulty (priv to pub key at every step because keys are no longer sequential).

Obviously you can shuffle something manageable, like 512 (or even tens of millions) of same-size ranges, and traverse them in the shuffled order. But for such low numbers you might as well pick a random range until you get one that wasn't marked as scanned.

Here’s one groundbreaking solution : scan all keys, in any order you prefer. Realise that there is no silver bullet or statistical bias.

My favorite order is to check every other 99194853094755497 private key, mod 2 ** 67. The only Cyclops-Fibonacci prime known so far.

It fits nicely with 1000 or so ranges too!