Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 458. (Read 2591964 times)

Yes, and in that example case making any statement of "what I earned over the past couple months" is not statistically meaningful.

That example was for 322 MH/s which is insanely small given current bitcoin difficulties.  The only reason to be mining with 300 MH/s at all is if you are betting on another 100-fold increase in the BTC/USD exchange rate making your handful of satoshis valuable someday.  Or alternatively if you don't care about profits and are just mining for the fun of it.  Whatever the reason, definitely don't bother mining on p2pool with that amount of hashrate.

Ok, I guess "Threshold" is a bad word...

How about this.

There's Variance, and then there's "A Snowball's Chance in Hell".

Right now at my Hashrate based on my experience with the difficulty rises over the last few months, I've gradually crossed over into "A Snowball's Chance in Hell" of getting a regular share and payout.  

Meanwhile in BTC Guild... I currently get 12564 - 13168 shares per round with my miniscule output, so I'm virtually guaranteed a reward even if it's theoretically lower than what would have gotten if I win the Lotto and somehow out of the blue get one measly share within a 24 hour period on P2Pool.  You're right, I might as well have run Solo.... oh wait, there are other choices for us lowly bottom feeders.  

How's that sound?

A year ago, did anyone in this forum think that a miner running around 100Ghs and lower would be considered a bottom feeder?

In lenny's example that would be 17 years.

I solo mined and found a block in 30 minutes. Over the same time I only earned 0.000002 BTC on BTCGuild. Solo mining is WAY better than any pool! You get paid SO MUCH more solo mining.

that is the logical equivalent of what you're saying. kjj is right, there is no threshold. The only threshold is how much variance is worth to you. When you cross that threshold, like you found, it is no longer fun to mine on p2pool. It's ok, p2pool has high variance for small miners - and to many/most it is not worth it. I'm a huge fan of p2pool, but I stopped since my personal value of variance is exceeding my interest in p2pool

Well goody for you... you seem to be the only one... read all the posts above yours and you'll see what the threshold is.  Glad you're getting something.  I was not... so for me the decision was easy since I already had some test miners on other pools types that made more with less power over the course of months.

Good luck.  Hope you keep getting lucky.

There is no such threshold.  Well, there might be one down in the 0-10 megahash range with the CPUs where the expected number of shares earned between now and infinity is less than a half.  But 20 Ghash/sec is nowhere near that.  Once again, I'm mining with about 7 Ghash/sec, and while I see plenty of variance, sometimes going 4 or 5 days without a payout, I still get paid, and my running averages are hovering rather close to my ideal expected payout.  (I'm actually up a fair bit right now because I had shares active during the very recent string of crazy good luck, including that block with the 20 BTC fee.)

p2pool have never been good for small miners.  All that has changed is how much hashrate qualifies as a "small miner".  Bitcoin difficulty has increased 1000x in the last 18 months.  Today's 10 GH/s miner is just as screwed as a 10 MH/s miner from 18 months ago.

I would go with "over time" = at least 100 times whatever your expected time to find a share is.

I think I get it. Sounds OK to my non tech head anyway - at least it's a start.....

Quick glance at stats:

As you see, p2pool is not good for small miners anymore. Even with 30 GH/s, you getting share once in 17 hours, variance can make this value 5x smaller or bigger, and you may miss the blocks.

Right now, we have modification in p2pool share diff calculating algorithm:
If address (miner) have more than 1.67% of p2pool hashrate, increase his node's share diff, so he will find shares less likely

1.67% corresponding to ~1.4 TH/s. I check p2pool.info stats and we have about 11 miners with hashrate above that. Biggest one have almost 15TH/s, which is ~18% of p2pool hashrate.

Couple of weeks ago, when commit has been made by forrestv to modify that limit to 1.67%, it was fine - 1.67% was equivalent of about 400-500 GH/s, right now it's 3x higher.

I am suggesting to change this behaviour to formula:

* If address (miner) is finding shares in less than 60 minutes, increase his node share diff so he will find share in exactly 60 minutes

This will be independent from current pool hashrate, last implementation (percentage of pool hashrate) doesn't work anymore with such big miners we have there now.

Avg. of 24 beefy shares per day will be enough even for largest miners, they shouldn't complain about variance, as their sacrifice will enable small miners to mine again on p2pool, total pool hashrate will increase, so total variance for pool will decrease (profit for all miners).

What you guys think?

I have to agree.....High Share difficulty will cut small miners out of the picture.....Not because of MATH and odds and probability..but because its not FUN anymore......

It's you who haven't been reading.  How many times must I say I was on p2pool for months and months. How long is "over time"? Will I see an odd payment in 15 years perhaps?  

Btw while I wasn't earning anything on p2pool for the last week or so, I paid my $250 daycare per week bill with earnings from BTCGuild over the previous 2 weeks and some extra that's been sitting in my wallet.  Again I was a big proponent of p2pool for a long time.  But, now difficulty has hit the threshold where I will always make more on BTCGuild with less power on it than I do at p2pool.  I'm sorry to let my node go but that's how the payouts ended up.

Take 30 days again.  We'll consider your farm as 5 units with an expected earning of 0.1 BTC per day each.

If you were fully on p2pool, you'd earn 30 days * 5 units * 0.1 per unit = 15 shares = 15 BTC. - this takes 30*5=150 unit/days

With your system:
Always on p2pool: 30 days * 0.1 per unit * 1 unit = 3 shares from your dedicated p2pool boxes = 3 BTC - this takes 30*1=30 unit/days
Because you have 3 shares, your other boxes will spend 3 days working on p2pool:  3 days * 0.1 per unit * 4 units = 1.2 shares = 1.2 BTC - this takes 3 * 4 = 12 unit/days
This leaves 108 unit/days, each of which is worth 0.1 BTC, for a total earning on the other pool of 10.8 BTC
Add them all up, and you get 15 BTC.

In reality, the second number would be a little lower, because the other pool probably charges more in fees than what you are donating for p2pool.

Is any of this making sense?  If you had exactly the right hash rate to find a new share (on average) right as your previous share was rolling off, so that you had exactly one share in the chain at a time, you'd earn X per block.  If you have half that hash rate, you still earn X per block, but you do so half as often because you only have an active share half of the time.  In reality, X isn't so constant, but it still works out to earning exactly the right amount, over time.

p2pool does not cheat small miners.

This does not work. You end up with exactly the same expected value. The days you find a p2pool share you get more than 0.5, the days you don't find a share you get only 0.4. It will exactly average out to 0.5 BTC per day over the course of several days.

Can you explain how any of PPLNS or DGM can be used by a miner to obtain more than their fair share? I'm pretty confident that either you're wrong, or you're mistaking variance for an increase in earnings.

Yes.  Some fraction of the time you will have 2, 3, 4,... active shares, but for a smaller miner, 0 and 1 will be vastly dominant, with 2 as a novelty, 3 as a rarity, etc.  At any rate, the sum of the series is 1/2 in the example.

This makes no sense.  Do you have even a rough idea of how p2pool works?

To simplify things, pretend that p2pool's hashrate is good for 1 block per day, and that the share chain only tracks the last 25 shares.  Each share is then good for 1 BTC.  Your mining rigs have a hash rate good for 0.5 BTC per day.  You will, on average, find 1 share every other day.  So, roughly half of the time, you will have an active share, and half the time you won't.  While you have a share active, you will earn 1 BTC.  While you don't have a share active, you will earn zero.  Over 30 days, you'll end up with 15 BTC either way.

And if you are unlucky you will make a lot less.  The point of pool mining is to eliminate such variance as much as possible.  You are right that it will be pretty much the same over a long enough time, but AGAIN, we mine on pools to eliminate as much variance as possible.  Nowadays, I don't think p2pool does as good a job of that for small miners as some other centralized pools.

That being said, I am not attacking p2pool, so I hope that's not what you think.  I am a big supporter of the p2pool concept, I mine on p2pool myself, and I promote p2pool to others whenever I can.  I just hope that it can be made more friendly to small miners eventually.

I don't think anyone is contending that p2pool is cheating anyone.

The point is that difficulty increases every 10-14 days, and a share a week from now may not be worth the same as a share now.  Share variance is just too big for small miners now on p2pool.

The argument that you will make the same over time is kinda true and kinda not.  You could just as well make the argument that you will make the same solo mining as you will in a pool - but it's not really true because difficulty isn't static.  If you get paid for your shares NOW, at a rate set by current difficulty, you will make more BTC - unless you are very lucky.

What part of "over time" and "in the long run" do you not understand?  Did you miss the part where I'm running a third of the power that you are running?

I don't really care about you, but I'm not a big fan of you spreading lies in here.  P2pool does NOT cheat people with low hashrates.

I guess if you really need that $7 each day for daycare, it may make sense for you to mine somewhere where you can get the steady payout.  But you would earn more over time on p2pool.