Quote from: Cryptos2go on February 19, 2014, 04:54:21 AM

Quote from: fcmatt on February 19, 2014, 04:29:59 AM

Quote from: Cryptos2go on February 19, 2014, 04:19:18 AM

Quote from: fcmatt on February 19, 2014, 04:00:41 AM

Quote from: drexel232 on February 19, 2014, 03:50:19 AM

Quote from: fcmatt on February 19, 2014, 03:38:33 AM

What am i missing with altcoinplex that makes it diff from other multi pools besides paying out in 3 diff coins?
Just having a proxy that splits the hash power to some coins at the same time? Instead of slamming one at a time?

Basically, little to no coin switching. The list of coins are mined constantly, but at varying percentages.

EDIT: Have to see how it performs under load, but I'm guessing that reject rates would be similar to mining a single coin directly. All day, I've been < 1%.

if their proxy works as i think it might be.. you could be mining a higher diff coin and get less rejects while another user could be mining a lower diff coin and have many rejects. I imagine they tie the miner's source IP to a stratum daemon that mines a specific coin. Therefore you could have other miners on a diff source IP mining a totally diff coin with a diff reject rate.

That makes me wonder if some users will complain their payout is smaller then others since they had many more rejects..

I would think less coin switching also and possibly the addition of "coin type" optimizations. An example might be lower share difficulty for the low difficulty fast block time coins. Assuming reject rate is based on blocks solving faster than spec, and not on latency (which of course also plays a factor) the rejects should be minimal. Until some coin switching pool without hash allocation comes along and blows the baby out of the bathtub.

You make some good points about lower share diff eliminating some rejects and less switching. That makes a lot of sense. But I could see cases where a single miner behind NAT with 20 mhash/sec might get tied to a single low diff coin.. and if you do not feed him a share diff rate of 512 or 1024.. he will be hammering you hard with a share diff of 32 or 64. That puts a ton of load on a pool server.

That is why I am trying to imagine how they are dividing up their hashing power and what the end result would be for miners on diff coins. Source IP seems the easiest to tie them to a stratum daemon.

Considering there appears to be no individual worker registration, maybe that would be a problem. Then again the current basket is made up of mostly .5~1gh/s+ coins is it not?

Don't you need individual worker's to use vardiff?

Perhaps this is a somewhat manual allocation at this point with static assignments?

I did not analyze the coins they are mining. If they are all higher diff I can see reject rates being lower for everyone but that is giving up a bit of profit ignoring the others.

I do not think you need individual workers anymore with job based vardiff. The code changed over the last 6 months. Dont hold me to that thought but I am pretty sure.

I doubt they are statically assigning them. I would guess src based proxying to stratum servers that have "weights" attached to them.. allowing certain amounts of traffic to hit them. And if I think I could make something like that work it does not impress me very much. It is basically just running a bunch of pools at once and dividing up the hash rate among them. I dont see how it will make payouts magically increase combined with the normal auto selling going on.