Re: [ANN] profit switching auto-exchanging pool

Quote from: Anubite on January 20, 2014, 05:50:45 AM

Quote from: dgross0818 on January 20, 2014, 05:12:41 AM

If we can kill off junk coins and make money on it, I think that's a bonus.

I can agree on that, sounds epic. Do you feel middlecoin will make the best decisions about which are pump'n'dump scamcoins to kill off and which are quality coins having implementations and advantages over LTC and BTC?

__

On to my reception and overview of 1024 pool difficulty.

For those of us with sub 2mh/s rigs, I feel we suffer. It takes my r9 290x at 962 kh/s anywhere from 45 seconds to over 2 minutes to complete one packet of work. It frequently takes my nVidia rig at 230kh/s over 10 minutes.

It appears that when a coin switches, that whole packet of work is rendered useless as cgminer starts to work on the new coin. If the coins switch enough, this adds up BIG TIME.

Below is an image of my hashing at MC.
Orange stars = Before diff 1024
Green Star = After

After having done absolutely nothing to my rigs, why else could this be happening?

I would hope that the coins with strong fundamentals will be able to survive getting hit with multi-pool traffic, however, this is certainly a new frontier, and we can't discount the possibility that we may damage a "good" coin. Then again, we have examples of forks, 51% attacks, etc with BTC, FTC and others, and those are still around even after many posted about the "critical condition" they were in at the time. Heck, BTC has had so many times in its relatively short history when it's future was thrown into question, however, thus far it's always bounced back. We will damage a small coin if we mine it with the hashpower we have, however, there are a number of ways to mitigate this, through algorithms in the coins themselves to dynamically redistributing pool hashing power so that we don't so quickly overwhelm a "small" coin. Remember how long FTC was "stuck" when the exchange rate dropped and the difficulty was crazy high? Those dedicated to it banded together and made sure transactions still went through and the coin didn't die :]

Anyways, more relevant to your issue, the "slow" miners and higher difficulty shouldn't affect payout. I know it seems that way when you're working on a share for 8 minutes and then the pool switches coins, however, unless my 2am statistics brain is shot, the probability of solving does not increase with time (The likelihood that a block WAS solved increases with time until one is, however, the instantaneous probability doesn't). If you have a miner that just received work vs one that's hashing away for 8 minutes, assuming they are otherwise identical, both have an equal likelihood of solving at any one instant in time. Otherwise it's like saying it's more likely to flip a heads after you just flipped 3 tails in a row... nope still 50/50. Sometimes a miner will take less than a second, other times it will take an hour. That's where the whole "luck" thing comes into play.

That said, there's definitely an issue in your hashing if your accepted shares dropped that much after we went to 1024. Without having experienced the same issue it's tough to hypothesize. I'll let you know tomorrow if I figure it out :] May have to fire up the 135KH/s laptop to test with to see if I can replicate it