Topic: [22TH][ANN] Australian P2Pool Backed Low-Rate Friendly Pool (Read 4214 times)

I would have thought so too, but it turned out to not. Perhaps Australia just doesn't have that many p2pool users.

Given the complaints about share difficulty, I would have assumed more interest.

Maybe people interested in p2pool want to setup their own node.

I was not able to garner interest from many miners, so I contacted all the regular miners I did have just before new year and recommended they mine somewhere else as we hadn't solved a share in over a week.

The node has been returned to operating as a standard p2pool node and the website has been updated to remove references to this experiment.

Looks like a pretty low hash rate for bitcoin.

The 0.25 bitcoin threshold is pretty high.  Maybe that is causing an issue?

The 5% admin fee has now been completely scrapped, with 100% of the profits now being distributed to miners.

P2pool hits blocks every 8 hours or so, so p2pool already contributes more variance than your pool does (since the faster the "hit" rate, the lower the variance).  My point wasn't about average payout, the point was that your variance is larger than a 14TH/s pool's would be.


Exactly. 

One way to help with that would be to offer a lower fee for miners with larger hash rate.

If you could get 2-3 people with 10GH/s each, then your share rate would be much higher than p2pool's block rate.  In that case, the variance would be almost exactly equal to p2pool's rate.

As you get more low hash rate miners, you could reduce the bonus.  Probably around 30-40GH/s your pool would contribute very little of the total variance.

Thanks for the replies.

At the moment with the p2pool network rate and the local node rate, our expected time to share is between 4 and 5 hours and statistically over the last 7 and 30 days we have been finding pretty much exactly that. Statistics are viewable at https://cryptominer.org/bitcoin/stats/

I would like to increase our percentage of the p2pool network to increase the actual payout amounts, but so far the payouts have been reasonably even when blocks are solved by the network.

I have had a number of users who have received the automatic payout now as they have reached their payout threshold, so it is working for users, and keeping the amount of dust in their wallets down.

Fair enough. 

However, for that to work, you need to earn more shares per day than p2pool gets blocks per day.  There are 3 sources of variance for your users, p2pool variance, pool variance and (local) share variance. 

If the pool variance and local share variance is much smaller than the variance of p2pool, then you get the same effective hash rate.

There is probably a formula that takes the local difficulty, the p2pool difficulty and the bitcoin main difficulty and gives the effective hash rate.

Yes, it's for the entire p2pool network (as I stated). The p2pool network rate is more important than my local node rate as that's the much bigger influence on how many blocks are solved by the network.

Shouldn't that be 14GH/s (or you mean that p2pool total)?

Just a small update to the title with the currently weekly average of the p2pool network.

LATEST NEWS
12th September 2013 - We now have an SSL certificate for our website, and all web traffic is now forced to HTTPS by default!

Congratulations to mojobobo on being the first user to receive a payout at the default payout value of 0.25BTC!

After being affected by this and taking quite a few hours yesterday afternoon and last to resolve on both production and our test environment, our mining node is back online and operating as it should.

Sorry to any inconvenience to miners, please make sure you have a backup pool configured!

Congratulations to Toresh for receiving the first automatic payout!

Have you costed or tried to run a reliable service? Hint: this is not running on some oversubscribed $10/month VPS, it's running on dedicated blade hardware in a professional data centre.

Having said that, the admin fee will be be reviewed regularly and reduced over time as the pool size increases. So far this month the 5% fee has covered about 2.5% of the monthly hosting fees.


You would need to run a separate P2Pool network to do that, which means until you will suffer from basically the same variance as solo mining until you are able to build up the volume. The main P2Pool network with 7TH/s has an average of 2 blocks a day, how many miners would you need to get even close to that?

5% fee is a little steep IMHO. As an alternative I was thinking of modifying P2Pool to create 10 times as many shares with a 10th of the difficulty but otherwise identical, would anybody be interested in running this ?

Your way will tend to give more even payouts, since lower hash rate miners could go a full shift without hitting a p2pool share.  They would have a big bonus every few shifts.

Larger miners would tend to get p2pool shares every shift.  Ofc, large hash rate miners might as well hash against p2pool directly.

I had a look at Eligius, and they don't seem to give any block finder reward (I could be wrong).  That is a pretty major pool, so maybe miners not turning in shares is not a big deal.

If you have 100 roughly even miners, then they naturally get 1% of the block reward anyway.

It is probably worth looking into how to adjust your log scheme so that you can increase the minimum difficulty.

I am not sure if the maths would be that easy.

Having said that, since you are targeting low hash rate miners, maybe it isn't worth worrying about.

Thanks again for your input and feedback, it's been a good learning experience for me!

I've run some numbers using the real statistics submitted by two miners during shift #4. Miner1 has approximately 6000MH/s and found 2 p2pool shares. Miner2 has approximately 650MH/s, or 10.8% of Miner1's rate, and did not find any shares.

The following were calculated with p2pool share diff of 25,800 and block diff of 65,750,060. TN50 is both miners set to diff 50, TN500 is both miners set to diff 500 and TNmixed is Miner1 set to diff 500 and Miner2 set to diff 50:


I also calculated using my log method with a bonus of 50% of the score for p2pool shares and of 100% for bitcoin blocks. This is shown in the LOGbon column above.

As you said, your scheme does reward submitting blocks that result in a p2pool share or block, however, I'm actually fairly happy with how my simple scheme matches up as well compared to the actual hash rate percentage.

No you get all rewards.  However, you don't get paid the full difficulty count, it is scaled based on how much the bonus the local pool actually gets.

When you get a type A block, it just proves you were working for the pool, so you get 50 "points".

When you hit a type B block, you win a p2pool share for the pool.  Normally, most pools pay the block finder 0.5% of the block reward as an incentive to submit the share.

In this case, that is 0.5% of 22000 (the p2pool share difficulty).  They get that in addition to their 50 points.  That gives them 160 points.

However, if a miner hits an actual block, then p2pool pays 0.5% of the block reward (25 BTC) to the local pool and they are entitled to 0.5% of that as the finder bonus.

That works out at 0.5% of (0.5% of 68 million).  They get that in addition to the p2pool share bonus and the basic payout 50 + 110 + 1700.


Yes, the advantage is that miners who select 500 difficulty still have the same expected payouts.  They get 500 points for each of their >500 difficulty shares, which they find 10X times less often.


True, you pay less for the actual winners, but not that much less.

Miners might withhold winning shares.  The bigger the bonus the less likely they would do that.  However, I don't know if it is actually a big deal.

Paying out more for bitcoin blocks could be worth doing, since you get the 0.5% reward from p2pool when you do that.

50: 3.912
22000: 10.00 (2.56 times 50's reward)
68 million: 18.04 (4.61 times 50's reward)

With what I suggested

50: 50
22000: 160 (3.2 times 50's reward)
68 million: 1860 (37.2 times 50's reward)

I've been thinking about your scheme overnight and just want to clarify something, are you suggesting that each of the local share, p2pool share and blocks only score the minimum for that type of share?

Eg,

A = > 50 < 22,000 but only scores 50
B = > 22,000 < 68,000,000 but only scores as 22,000
C = > 68,000,000 but scores as 68,000,000

So, finding a local share at say diff 10,000 still scores 50 points?

With my log scheme, you do already get a higher score for finding shares and blocks:

I think some pools have a minimum payout though.  If the fee is 0.0005, then not much paying out if the total is less than 0.001 (and that is 50% fee if you do).


Ahh ok, I assumed log((difficulty) / (minimum diff))


You might want to consider just having 3 types of shares based on their difficulty.

A) 50 = local share
B) 22000 = p2pool share
C) 68 million = bitcoin block

The miner who hits B should be given an incentive to turn the share in.

However, type C shares (very rare) would earn the local pool 0.5% of a BTC block reward (the p2pool finder fee), which is worth 0.125 BTC.

A should pay 50
B should pay 50 + 22000 * (0.5%) = 160
C should pay 50 + 22000 * (0.5%) + ((68 million) * 0.5%) * 0.5% = 1860

(You need to fill in the actual difficulty for both bitcoin blocks and p2pool shares)

A person who hits a share just gets 50 "points"
A person who hits the p2pool share also gets 0.5% of the p2pool reward (so 0.5% of 22000 "points")
A person who hits the bitcoin blocks also gets 0.5% of the finder bonus that p2pool pays (which is itself 0.5% of a block)


I think your log system makes it more difficult, since you have to do the math to make sure all difficulty selections by the miner give the same average reward.

Under the scheme I suggested miner who picks difficulty 500 would get

A pays 500 (since only difficulty > 500 shares are accepted)
B pays 500 + 22000 * (0.5%) = 610
C pays 500 + 22000 * (0.5%) + ((68 million) * 0.5%) * 0.5% = 2310

Thanks for the excellent questions!


I hadn't thought about that. Other pools I've mined in (Slush and BitMinter) don't do that to my knowledge, but in the context of P2Pool, it's probably a very good idea!


No, a diff 50 share does count - a diff 50 share has a score of around 3.9. Even my miners with only 2 USB Erupters (around 600-650MH/s) are averaging 130 pseudo-shares that are higher than diff 50 per 12 hour shift.


At the moment it's the natural log rounded to 5 decimal places, ie the score for a diff 50 share is:



This is also a good idea, although I'll have to have a think about how to implement it.

Do you have a timeout if a miner leaves the pool?  For example, if the miner submits no shares for 7 days, it will be considered to have left the pool and so be paid.


A log payout means that someone who submits a 50 difficulty share gets nothing?

Is it log2(50) ?

It would be useful to give a system where miners can choose higher share difficulty.  This will put less load on your server.  However, it needs to work so that they get the same payout as everyone else.

Announcement: Australian P2Pool Backed Low-Rate Friendly Pool

CryptoMiner is pleased to announce the release of our modified P2Pool mining node based in Adelaide, Australia specifically designed to provide low-rate (<10GH/s) P2Pool users with a lower variance reward than standard P2Pool nodes. We have the benefit of the P2Pool network hashrate (6TH/s) with scoring based on diff 50 pseudo-shares rather than 22k diff P2Pool shares.

Work is broken up into 12 hour shifts (starting a midnight and noon UTC) and miners are scored for all submitted pseudo-shares with a diff >= 50. For each shift, the system calculates the user's shift score and percentage of the total score submitted for that shift. The miner is credited with their calculated percentage of the total amount received by the pool during that shift, minus a 5% admin fee.

Payouts are automatically made when the miner's credit balance reaches BTC0.25. By registering for an account on the CryptoMiner website, the user is able to modify their payout address and threshold and also see statistics about their miners. Please note that registering on the website is completely optional as the node will act as a standard P2Pool node, paying out in BTC0.25 minimums instead of every block.

Website: https://cryptominer.org/bitcoin/
Statistics: https://cryptominer.org/bitcoin/stats/
Scoring Information: https://cryptominer.org/bitcoin/scoring/

LATEST NEWS

• 14th October 2013 - Removed 5% admin fee - all payouts are now 100% of the mined percentage,
  
• 12th September 2013 - We now have an SSL certificate for our website, and all web traffic is now forced to HTTPS by default!