I am curious about the large fluctuations in the total network hashrate. I mean 20,000 GH/s is the size of Slush's entire pool, how does that much compute power come and go so quickly? Its like someone with 40 BFL Mini Rigs is dropping in and out of the network.
Network hashrate is calculated from the count of blocks found, there's no other way to determine it. But block frequency varies according to the luck. As we can't determine current luck, we have to take it into account always being 100 %. Sure this is not true, and so we are observing variations in network hashrate that don't exist in fact.
Certainly I don't want to say there aren't any variations. They are. But I can't tell how big they really are. Can someone go further and calculate the probable contribution of this effect to the total variation?
- Miners' workers can tell their hashrate exactly.
- The pools calculate their hashrates from shares of difficulty 1 or more. With ASICs comming, there'll be more of "more", and the accuracy will go down. Well only absolutely, maybe, and the relative accuracy can be maintained, as the hashrate increases also.
- The whole network has no means to determine its total hashrate other than from the blocks found, as stated above. [edit:] With increasing difficulty, variations will increase and the accuracy will continue going down.
[edted after
n0creativity's reply]:
The pools get 2
32 less haashes from the miners than the miners themselves calculate. (Or even less, when client difficulty set to > 1.) But they don't obtain them equaly in time, rather with variable frequency according to the luck.
The whole network only gets blocks of current difficulty from both pools and solo miners. (So currently it gets other 19M times less hashes than the pools do.) There's nothing like shares at the network level. Shares' only purpose is to enable reward distribution among pools' clients.
And again, luck (variation) applies.