|
Title: ASICBoost and the Status Quo Post by: 25hashcoin on May 25, 2017, 08:49:06 AM https://blog.bitcoinhk.org/asicboost-and-the-status-quo-96de404400b6
Great read. Very important. Title: Re: ASICBoost and the Status Quo Post by: dinofelis on May 25, 2017, 09:34:15 AM https://blog.bitcoinhk.org/asicboost-and-the-status-quo-96de404400b6 Great read. Very important. I already talked about that, but in as much as the theoretical asic boost implementation is really winning in efficiency (it does, on paper, but the electronics my be more complicated), this is much ado about nothing. Note that I'm adding the above caveat, because I too, accepted the theoretical efficiency gain of ASIC boost as established, but then some experienced electronics designer jumped in to point out that it is not because on paper, you can save yourself from repeating 25% of the calculations, that you can make hardware that is 25% more efficient, if the householding of those calculations is much more involved (which it is). So, *under the assumption* that a hardware implementation of ASICboost is genuinly more efficient (with the above caveat), then this is just "technology improvement". The hard part of this is that there's a patent deposited on it, but it is not bitmain that detains it, it is a core dev as far as I understood. There's no "cheat" in asic boost, the principle is published, it is just a smarter way of organizing the calculations ; note that the standard way of calculating the hashes is already doing something similar. The standard calculation doesn't calculate full hashes either for each trial and re-uses previously calculated results. But ASIC boost is a smarter scheme, that uses the particular split-up of the two blocks to be hashed in the header, to re-use results in a smarter way, gaining 25% in the amount of individual operations to be done (but has more complex data paths). As I pointed out earlier, PoW being a silly cryptographic protection scheme where the attacker needs to spend *the same* amount of work as the good guy (in most cryptographic schemes, the attacker needs to spend MUCH MUCH more work than the good guy), at least, the good guy should use all of the available technology and knowledge to not have to spend MORE work than the attacker. This would be the case if miners wouldn't but attackers would use ASIC boost. Also, if the "fairness dogma" would be that "hashes have to be calculated each time completely and no results can be re-used", then the standard algorithm is already cheating on that hypothetical dogma (which would be vastly stupid given the previous paragraph). So the only problem with ASIC boost, if it is true that hardware based upon it is more efficient, is that there's a patent taken on it. Title: Re: ASICBoost and the Status Quo Post by: franky1 on May 25, 2017, 10:00:10 AM I already talked about that, but in as much as the theoretical asic boost implementation is really winning in efficiency (it does, on paper, but the electronics my be more complicated), this is much ado about nothing. Note that I'm adding the above caveat, because I too, accepted the theoretical efficiency gain of ASIC boost as established, but then some experienced electronics designer jumped in to point out that it is not because on paper, you can save yourself from repeating 25% of the calculations, that you can make hardware that is 25% more efficient, if the householding of those calculations is much more involved (which it is). So, *under the assumption* that a hardware implementation of ASICboost is genuinly more efficient (with the above caveat), then this is just "technology improvement". The hard part of this is that there's a patent deposited on it, but it is not bitmain that detains it, it is a core dev as far as I understood. There's no "cheat" in asic boost, the principle is published, it is just a smarter way of organizing the calculations ; note that the standard way of calculating the hashes is already doing something similar. The standard calculation doesn't calculate full hashes either for each trial and re-uses previously calculated results. But ASIC boost is a smarter scheme, that uses the particular split-up of the two blocks to be hashed in the header, to re-use results in a smarter way, gaining 25% in the amount of individual operations to be done (but has more complex data paths). As I pointed out earlier, PoW being a silly cryptographic protection scheme where the attacker needs to spend *the same* amount of work as the good guy (in most cryptographic schemes, the attacker needs to spend MUCH MUCH more work than the good guy), at least, the good guy should use all of the available technology and knowledge to not have to spend MORE work than the attacker. This would be the case if miners wouldn't but attackers would use ASIC boost. Also, if the "fairness dogma" would be that "hashes have to be calculated each time completely and no results can be re-used", then the standard algorithm is already cheating on that hypothetical dogma (which would be vastly stupid given the previous paragraph). So the only problem with ASIC boost, if it is true that hardware based upon it is more efficient, is that there's a patent taken on it. the first post of yours i can completely agree with you on. we should not temporarily halt efficiency gains, by bitcoiners, where by private outside badguys could take the advantage. bitcoin needs to use the most efficient ways to mine. |