Topic: NastyPoP vs Standard P2Pool - page 6. (Read 17708 times)

I think it clearly highlights the variance inherent in standard p2pool mining.  The fact that this particular S3 has had a streak of good luck demonstrates the kinds of variance your payout system is designed to defeat.  Just because it's good luck right now means nothing as the luck can turn the other way just as quickly.  If you looked at the payout history you would see plenty of blocks for which I was not rewarded because I had no shares on the chain.

As I wrote in my analysis, the results I've posted this far are far too small a set to infer any real conclusions.  2 weeks isn't very useful, and is why I'm continuing to run this experiment.  Truth be told I'm actually hoping for a bit of bad luck so I can show how even though I found no shares to submit to the chain I am still being rewarded for my work by mining in the NastyPoP system.

P2Pool's hashrate is based upon submitted shares.  As I mentioned, that S3 has been lucky and submitted more shares than it should have, resulting in p2pool thinking it's more powerful than it is.  It's the nature of variance... if you take a look at windpath's node, you can see the payout history of that miner dating back to when I first fired it up on 8/19.  There are plenty of stretches of 0 reward blocks where I didn't have any shares on the chain.  During those times p2pool wouldn't even have known about that miner since there were no valid shares on the chain from it.  As of right now, p2pool thinks a 440GH/s miner should be expecting a payout of about 0.0043BTC, which is about 2 weighted shares on the chain.  My current expected payout is about 0.0092BTC or about 4 shares, so I'm doing better than expectations.

I'm not entirely sure what you mean?  Unless you're looking at expected payouts in the share explorer or on windpath's node... in which case, you'll see it's been a little lucky the past few days.  I can assure you, however, that the wallet address I provided belongs to a single S3 that began its mining life for me on 8/19/2014.  The miner has done pretty well for itself - earning 0.87270703BTC from p2pool since it started.  According to http://retrocalc.net, the expected earnings for that timeframe at 440GH/s is 0.783BTC, so I've made 111.46% of expectations.

Sure, you can check it out.  The address is 1Km6zTdCfZATTTVGcMabJ7BrwcopYvRp5W.  You can see it running on my backup node at 104.131.12.128:9332.

Yes.

Good write up 

Do miners get paid tx fees on Ognasty's p2pool-pop the same as the regular p2pool?

I believe you're correct about the latency killing me, but there's not a whole lot I can do unless you decided to open up a server here on the east coast of the US.

I was happy to do the writeup, and am glad to be able to help you out by providing the comparison.  One thing I forgot to mention in the OP is that if your pool finds the block, that 0.5% finder's reward is given out to the miners as well - a very nice touch!

I did pay attention to the stats as well.  NastyPoP records a very accurate hash rate graph - showing my S3 consistently at 440GH/s.  Your graphs are quite nice - showing not only the actual and mean hash rate, but also if a share was found.  Nice job on those.  Here's a screen showing my miner:

As has been alluded to in the original post and in the discussion between OgNasty and myself throughout this thread, the results of the comparison between NastyPoP and p2pool are pretty anecdotal.

I think the primary focus here should instead be upon what OgNasty and Nonnakip have done to try and solve the problem of variance inherent in p2pool mining, especially for the smaller miner.  For example, I'm using S3s for this test because, well, that's the equipment I own.  As of now, the S3 is still a pretty viable mining device to use by itself on p2pool.  The chances are pretty good that the S3 is going to have a share on the chain every time a block is found.  This means that my equipment really isn't suffering too badly from the variance problem to begin with and hence the payout normalization provided by NastyPoP shouldn't have too much of an effect.

My point here is that over time, the payouts from p2pool and NastyPoP should naturally converge, all things being equal.  There will be times when my S3 mining on the traditional node will not have any shares on the chain when blocks are found.  This will become more evident as share difficulty increases and total pool hash rate increases.  I will see blocks of no payouts, along with blocks of getting them.  With NastyPoP, I'll always see a payout for work that has been contributed, assuming of course that the combined efforts of every miner on NastyPoP have indeed found shares.

That is the primary difference between NastyPoP and traditional p2pool mining, and needs to be stressed.  Miners with lower hash rates won't have the impression that they've wasted their time and not gotten any reward for their mining.  Rather than see 5 blocks with no payouts, then 5 blocks with payouts the miner will see 10 straight blocks of payouts.

I hope anyone who is reading this thread can find some useful information and make some informed decisions on how they want to proceed.  I applaud OgNasty and Nonnakip for taking the time to devise and implement their solution.

TL;DR - Comparison of payouts of NastyPoP and NastyP2P.  Here's the current chart showing payouts over time:



Expected Payouts vs Actuals



Raw spreadsheet:

https://www.dropbox.com/s/vv7r7ejq2pquynl/pop_vs_p2p.xlsx?dl=0

A couple weeks ago I wrote a post comparing p2pool's payouts to those of BAN, clearly showing that p2pool has paid out an average of 112.63% of expectations over the lifetime of the BAN pool.  BAN pays 110% PPS, so p2pool has beaten it by 2.63%.  You can see my post here: https://bitcointalksearch.org/topic/p2pool-vs-ban-875644.

After reading my thread, OgNasty asked if I would do a similar comparison between p2pool and his newly created NastyPoP pool.  For those of you who do not know what that is, let me briefly describe it.  OgNasty has been running his own p2pool node for a long time at nastyfans.org.  During the existence of his pool, he's offered some pretty interesting additions to standard mining, including the ability to purchase "seats", which effectively give the owner of a seat a membership to the nasty fan club.  The seats themselves are special 1 ounce silver coins that are actually pretty cool.  You can check them out here: https://nastyfans.org/mint.

As anyone who has been involved with p2pool is aware, one of the biggest problems with the pool is variance.  Specifically, as the pool grows, individual miners suffer greater and greater variance.  This is precisely the opposite of what happens in a more traditional pool like BTCGuild, where the larger the pool gets, the steadier the payouts become.

OgNasty and Nonnakip decided to do something about it and introduced NastyPoP.  Effectively, this is the first real attempt by anyone to try and tackle the p2pool variance problem.  Bitmain's p2p AntPool is also out there, but it seems to be stuck in limbo, so who knows.  So how does it work?  Well, with a regular p2pool node you connect your miner by providing a bitcoin wallet address as the username and anything as a password.  If and when your miner finds a share with a difficulty greater than the target share chain difficulty, your share gets added to the chain.  When the pool finds a block, you get paid for the shares you have in the current payout list of the chain.

NastyPoP is a bit different.  When you connect, you still provide your bitcoin wallet address, but you append -PoP to the end of it.  This signifies to the nasty pool that you are going to use their proprietary payout system.  Underneath the scenes, OgNasty and Nonnakip have created a sort of "pool on top of a pool".  Everybody who is connected using the -PoP suffix is in reality mining to the node's payout address.  You can try this on your own node.  Simply provide something that isn't a valid wallet address as the user name, and if you find a share, it is the node's default payout address that gets the credit for the share.

Well, here's where the custom code comes into play.  OgNasty and Nonnakip have devised a way to put a PPLNS payout system very similar to a traditional pool in place on top of p2pool's backbone.  They are tracking shares for your miner just like a traditional pool would.  Every miner who has the -PoP suffix shares in the block reward when p2pool finds a block, based upon how many valid shares that miner has contributed during the payout shifts.  I don't have the exact details on how long shifts are, or what the share clip length is, and as I wrote, the code is proprietary and not yet open-sourced.  Payouts from NastyPoP are on Fridays at 19:00 UTC.  They are sent just like any other transfer and require only 6 confirmations.  P2Pool payouts happen as soon as the block is found and require 101 confirmations.

How does this reduce your variance?  Simply put, your miner is paid out in small increments based upon your contributions, rather than being dependent on finding a share to add to the share chain.  That job is left to the node's payout address.  All in all it's quite a clever implementation of reducing p2pool's variance for smaller miners.

One thing that's very important to note here about NastyPoP: you can't take your work with you if you move to a traditional p2pool node.  No other node out there has any clue whatsoever of the work you've been doing on NastyPoP.  In this fashion, NastyPoP behaves much more like a centralized pool the likes of BTCGuild than a p2pool node.  I think it's great that OgNasty and Nonnakip have taken the initiative and are trying to help solve the variance problem.  Unfortunately, the solution they've come up with defeats the very nature of the decentralized p2pool.  Yes, it helps eliminate variance, but at the cost of forcing you to remain tied to the NastyPoP node.

That's all fine and dandy, but how does it compare?  To test this out, I pointed an Antminer S3 at the NastyPoP node, and another S3 at my own standard p2pool node.  Each S3 is configured to run at stock speeds (440GH/s), and I set a static pseudo share difficulty of 500 on both.  Both are powered by an EVGA 1300 G2 and hardwired into an always-on gigabit home network with more than enough up and down bandwidth.

Let's take a look at the ping times from my miner to the pool:

--- nastyfans.org ping statistics ---
57 packets transmitted, 57 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 115.153/144.510/277.534/45.255 ms

An average of 144.51 would stop me from pointing a miner there if it were a standard p2pool node.  Let's see where the server is.  Running a traceroute shows the last hop in Germany.  Well, that explains the ping time - I had to go to Europe .  Our friends on that side of the pond would probably be better served.

Another thing I noticed was a much higher reject rate from the NastyPoP pool than I do on my own nodes.  This could be due to the latency, but it turned into about 12-14% rejects, whereas I typically see less than 1% on my own nodes.

Alright, so how did the payouts work?  Here are the results:

11/28 - 12/5
NastyPoP - 0.03379787BTC
P2Pool - 0.03589959BTC
Expected - 0.0385BTC

12/5 - 12/12
NastyPoP - 0.03238691BTC
P2Pool - 0.03937531BTC
Expected - 0.0387BTC

12/12 - 12/19
NastyPoP - 0.03382075BTC
p2pool - 0.07363736BTC
Expected - 0.0389BTC

12/19 - 12/26
My p2pool node - 0.05615571BTC
NastyPoP - 0.05015601BTC
Expected - 0.0393BTC

12/26 - 1/2
My p2pool node - 0.03205607BTC
NastyP2P - 0.03861873BTC
NastyPoP - 0.02812263BTC
Expected - 0.0388BTC

1/2 - 1/9
My p2pool node - 0.024337627BTC
NastyP2P - 0.06870469BTC
NastyPoP - 0.04531921BTC
Expected - 0.0381BTC

1/9 - 1/16
My p2pool node - 0.00419189BTC
NastyP2P - 0.01805329BTC
NastyPoP - 0.02152612BTC
Expected - 0.0367BTC
P2Pool 7 day luck - 62.25%

1/16 - 1/23
My p2pool node - test discontinued due to hardware failure
NastyP2P - 0.03989456BTC
NastyPoP - 0.02833329BTC
Expected - 0.0352BTC

1/23 - 1/30
NastyPoP - 0.01963295BTC
NastyP2P - 0.02061486BTC
Expected - 0.0361BTC
Luck - 81.85%

1/30 - 2/6
NastyPoP - 0.05905592BTC
NastyP2P - 0.02003162BTC
Expected - 0.0375BTC
Luck - 93.32%

2/6 - 2/13
NastyPoP - 0.00435959BTC
NastyP2P - 0.00938869BTC
Expected - 0.0361BTC
Luck - 14.3%

2/13 - 2/20
NastyPoP - 0.00698833BTC
NastyP2P - 0.01822003BTC
Expected - 0.0348BTC
Luck - awful

2/20 - 2/27
NastyPoP - 0.03798674BTC
NastyP2P - 0.04202049BTC
Expected - 0.0338BTC
Luck - 100.64%

2/27 - 3/6
NastyPoP - 0.02048889BTC
NastyP2P - 0.00942830BTC
Expected - 0.0332BTC
Luck - 51.98%

3/6 - 3/13
NastyPoP - 0.05138458BTC
NastyP2P - 0.05908961BTC
Expected - 0.0328BTC
Luck - 131.5%

3/13 - 3/20
NastyPoP - 0.03402797BTC
NastyP2P - 0.05156673BTC
Expected - 0.0327BTC
Luck - 101.66%

3/20 - 3/27
NastyPoP - 0.02843568BTC
NastyP2P - 0.01970915BTC
Expected - 0.033BTC
Luck - 90.54%

3/27 - 4/3
NastyPoP - 0.03301631BTC
NastyP2P - 0.05550941BTC
Expected - 0.0332BTC
Luck - 97.55%

4/3 - 4/10
NastyPoP - 0.02866930BTC
NastyP2P - 0.03528256BTC
Expected - 0.0318BTC
Luck - 113.4%

4/10 - 4/17
NastyPoP - 0.00028594BTC
NastyP2P - 0BTC
Expected - 0.0313BTC
Luck - 0%

4/17 - 4/24
NastyPoP - 0.05554895BTC
NastyP2P - 0.05281205BTC
Expected - 0.0321BTC
Luck - 165.47%

4/24 - 5/1
NastyPoP - 0.02111442BTC
NastyP2P - 0.02415729BTC
Expected - 0.0325BTC
Luck - 59.99%

5/1 - 5/8
NastyPoP - 0.03294026BTC
NastyP2P - 0.05447989BTC
Expected - 0.0325BTC
Luck - 127.84%

5/8 - 5/15
NastyPoP - 0.03472755BTC
NastyP2P - 0.02216391BTC
Expected - 0.0325BTC
Luck - 84.97%

5/15 - 5/22
NastyPoP - 0.05258228BTC
NastyP2P - 0.03171723BTC
Expected - 0.032BTC
Luck - 182.64%

5/22 - 5/29
NastyPoP - 0.04249927BTC
NastyP2P - 0.03263267BTC
Expected - 0.0307BTC

5/29 - 6/5
NastyPoP - 0.03711295BTC
NastyP2P - 0.03414086BTC
Expected - 0.0312BTC

6/5 - 6/12
NastyPoP - 0.04351160BTC
NastyP2P - 0.04518917BTC
Expected - 0.03144BTC
Luck - 122.76%

6/12 - 6/19
NastyPoP - 0.01964088BTC
NastyP2P - 0.02788807BTC
Expected - 0.030107BTC
Luck - 65.12%

6/19 - 6/26
NastyPoP - 0.02159208BTC
NastyP2P - 0.0193531BTC
Expected - 0.030107BTC
Luck - 60.76%

6/26 - 7/3
NastyPoP - 0.02828584BTC
NastyP2P - 0.01623161BTC
Expected - 0.030263BTC
Luck - 81.3%

7/3 - 7/10
NastyPoP - 0.02039533BTC
NastyP2P - 0.02521461BTC
Expected - 0.030289BTC
Luck - 71.40%

7/10 - 7/17
NastyPoP - 0.02054230BTC
NastyP2P - 0.02884854BTC
Expected - 0.03493028BTC
Luck - 71.40%

7/17 - 7/24
NastyPoP - 0.04771791BTC
NastyP2P - 0.04040042BTC
Expected - 0.03465856BTC
Luck - 162.79%

7/24 - 7/31
NastyPoP: 0.04298881BTC
NastyP2P: 0.03145364BTC
Expected: 0.03401418BTC
Luck - 134.83%

7/31 - 8/7
NastyPoP: 0.01487946BTC
NastyP2P: 0.01145337BTC
Expected: 0.02962897BTC
Luck - 55.38%

8/7 - 8/14
NastyPoP: 0.03914884BTC
NastyP2P: 0.07990332BTC
Expected: 0.02941349BTC
Luck - 241.06%

8/14 - 8/21
NastyPoP: 0.03640067BTC
NastyP2P: 0.04331323BTC
Expected: 0.02939197BTC
Luck - 138.27%

8/21 - 8/28
NastyPoP: 0.01170063BTC
NastyP2P: 0.02753223BTC
Expected: 0.02857788BTC
Luck - 54.07%

8/28 - 9/4
NastyPoP: 0.0455039BTC
NastyP2P: 0.04888988BTC
Expected: 0.02846873BTC
Luck - 180.68%

9/4 - 9/11
NastyPoP: 0.01466154BTC
NastyP2P: 0.00815205BTC
Expected: 0.02719480BTC
Luck - 63.24%

Running totals
NastyPoP: 1.54926483BTC
NastyPoP from 12/26: 1.39910329BTC
NastyP2P: 1.48230244BTC (test started 12/26)
Expected earnings: 1.69383675BTC
Expected earnings from 12/26: 1.53839914BTC
My P2Pool Node: 0.265653547BTC - discontinued 1/16

NastyPoP vs Expected Earnings from 11/28: 91.46%
NastyPoP vs Expected Earnings from 12/26: 90.95%
NastyP2P vs Expected Earnings from 12/26: 96.35%
My P2Pool Node vs Expected Earnings from 11/28 - 1/16: 98.76% - discontinued