Blockchain-based hash: unlike Bytecoin PoW hash, which works on 2MB scratchpad and takes avr 400 ms, new hash will work on blockchain random data to provide operation speed. It gives faster synchronization with network and immunity from DoS attacks.
So how usefull is that innovation really and how vulnerable is it in comparison to traditional approaches? Cant the supposedly random data from the blockchain be precalculated because, well its already there and not changing much with every new block?
Im no techie, so pls have mercy on me
With traditional approach, increasing scratchpad leads to increasing time for hash calculation, since hash have to cover all scratchpad memory, to avoid precalculations. Slow PoW hash can lead to pretty big amount of problems, such as weastimg resources with calculating fake PoW, long synchronization, e.t.c. At the other hand, even pretty big fixed scratchpad (2mb for example) doesn't grant that hardware will not move to this area(remember litecoin, they was counted on it, but now they explaining that ASICs is not so bad).
Research of other PoW approaches, based on asymmetric memory consuming algorithms (ethereum dagger, Cuckoo Cycle) showed me that there always some risk that algorithm could be reinvented and PoW may become unsuitable(for ex reinvented algo may request few memory).
I've just tried to make a step back and look for another way, i want to have one hash operation fast and strong, but i want to have
mining process memory hard, because it seems to be great protection from ASIC.
I chose scratchpad memory increasing speed that gives enough protection(and even with extra margin for future) but at the same time make able to implement SPV client or mobile version without any problems.
It's pretty new approach as i know, well, we'll see how things go.
Feel free to contact me if you have any questions.
