Might just be quicker to hash out new addresses en-masse then use a checking script to check if any of your generated addresses have balances already. You have a better chance hitting the lottery 20 weeks in a row, than finding any sizable amount in billions of keys.
-MisterCoin
That's how brainflayer works, does the address generation in batches, then turns it loose and compares the hits in a bloom filter which is a single cycle check, production of addresses from priv-key (guesses) is best done in 8192 batches, having a bloom filter on board GPU allows 150m/sec, no problem on 1060 class card
I suggest studying 'vanity gen' gutting all the vanity bullshit and hacking the gpu code to just generate batches and verify in the bloom-filter that a high-value address was found, then print the private-key, the hardest part of this task is getting the bloom filter on board the gpu, limit is 1/2mb for 1060, and about 1gb ( malloc() alloc per chunk ), so with a 1080 you bloom can do about 50 million addresse verification on one check.
www.inflection.topThe worst thing about 'bitcrack' is that you can't find anything doing a linear brute force search, you block-generation of private-key candidates must be an intelligent search, and the public key guesses must be based on a known DLP algo
It's impossible to hit specific private keys using linear brute force, if all the atoms in universe ( 10**72 ) were cpus using this code (bitcrack), they would still have difficulty find a key match, if you only have 10k gpu class machines running, then it would take 10**48 years to find your key.
There is a reason crypto scientists call it 'brute force', brute is code for moron, only a moron de-crypts in this manner.
Thanks for sharing! I think indeed that this might be a better approach. I am giving this a try as well.