Post
Topic
Board Development & Technical Discussion
Re: Crypto question: Breaking ECDSA for all key-pairs simultaneously?
by
keystroke
on 13/09/2012, 16:54:54 UTC
Maybe I'll turn my spare GPUs to cracking ECDSA when I convert to ASICs.

GPUs are no good at point multiplication from what I was able to discover on the very limited amount of data on the subject a few months back. Parallelization is not a performance enhancement.

I always thought address reuse was mainly to protect privacy somewhat - would you mind explaining this more? How does address re-use let the public key be known?

It is about privacy, for the most part. When you send a transaction with an address that is only known to the network as a hash, you must give your public key, or your public key will be derived from the signature. Then the network knows the public key for that address. But as RIPEMD160 is "160 bits of security" vs. the effective 128-bits of security of a 256-bit elliptic curve, it is 32 bits more secure in a sense, but not really against a brute-force attack as it just adds another step of first converting private keys into public keys (a relatively slow operation compared to hashing) then hashing (very fast) to see if it matches the hash. But since a RIPEMD160 hash is not necessarily just a hash of a public key (could be scripts or something else to throw it off, or one of several signature algorithms in the future), the address space being larger does make it somewhat more secure under some circumstances.

Thanks Etlase2! Where is the best place to read about these internals? I am going to study https://en.bitcoin.it/wiki/Transactions but is anywhere else good? Besides the source code Cheesy