Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 315. (Read 2591971 times)

So I've been following this experiment since you started.  Let's examine what you've done vs the other guy.
1) You setup a rig virtually identical in hashing power to his to keep things equal on that level
2) You set your rig to only submit shares that would solve a block, whereas his rig was set to difficulty defined by the node.
3) The difficulty you set is approximately 1000 times higher than what he has.

Over time, both of your payouts will even out.  Each of your rigs will take approximately the same time to find a block, and each has an equal chance of doing so.  The difference is that while he continues to submit shares to the chain, you do not until you find a block.  This means he'll have 1000 shares to your 1, give or take.  However, your 1 share will be worth about 1000 times what his is.  When the block is found, and the payouts are generated, all of the shares on the chain are weighted to determine how much of the 99.5% of the block reward each share gets.  The XJO chain is 4320 shares for payout, and the reward is 16 coins spread over those shares.

Since he's just as likely as you to find the block, the "finder's reward" will be split evenly between the two of you.  This leaves us the shares on the chain.  As I stated in the previous paragraph, he'll have far more shares on the chain, but each will be worth about 1000th of yours.  Upon payout, you're going to get about the same coins.  Yes, there will be times where you find a bunch of blocks and have more shares on the chain than you should.  There will also be times when you have no shares on the chain because he's been finding the blocks, and you haven't found any.

In conclusion, your experiment simply has not run long enough to provide proof of your claim.  If you were to let it run, you'd see it play out as I have stated.

grrr looks like this other 200 g rig has pulled up the stumps and is leaving that node wonders why   so no point in leaving the rig on there to run the test without the control there

Ya may want to look at the p2 stats as there are ph users that do pop up on the network there none on there atm but there were a couple on the p2 network a couple days back will grab a screenshot the next time they pop up

If you would have read the rest, you would've noticed that I attributed that to normal variance (and provided a real world example of the variance on BTC p2pool). You happened to have gotten "lucky".  If you let it run a week, I'm sure those payments would've evened themselves out.  That's how the system (sharechain and blockchain) works.  I've heard of pools taking close to 7 times expected to find blocks and I was mining on eligius when we had a 26 second block instead of our normal hour long expected.  You got lucky in finding the appropriate difficulty, given time it would've evened out.  That's how both the blockchain and sharechained are designed to work.  A 6 hour window on any p2pool node is not enough time to prove anything.  Now comeback when you've mined a week or more, constantly, then you'll have proof, until then you just got lucky, that's how the system works.  Hell, it works the same as every non-PPS pool out there, only we have (and you used) larger share targets, so variance is a huge factor.

And again, see as this thread is for the BTC p2pool, if someone comes on and mines with several Petahash, rather than solo mining with it, they should have their heads examined, as that's what would be required to to even come close to duplicating what you did on XJO.

My UI reports that I currently have 77 shares, and it's been in the 60-80 range pretty consistently for the last several days. I got regular payouts for all the blocks found yesterday (and, in fact, for the last few weeks).  However, I didn't get paid for today's two blocks. Why would that be?

think about your statement "To start with you were setting a 40k diff on a node where a normal share is in the 20-40 difficutly range, of course you're going to take the majority of the payments when you submit shares that 1000 to 2000 times normal diff.  At this moment that particular node is reporting a diff of 4 for a share so again, you'll takeover the sharechain the second you submit a couple 40k diff shares. "   think very carefully you are correct in what you are saying and hence the higher payouts  <<<<<< as that share carrys a higher weight than a share worth 1000 times less I will contune to run the miner on that node and people will be able to see the effect this has on earnings between myself and the other 200 g rig on there via looking at the 2 payments address.....     <<<<   in 24 hours I will bet ya I am 70 - 100 up percent wise on the other 200 g rig earning wise as they are putting shares in at the lesser value so do not carry as much value to them


Hence the flaw call on the p2 payout system as I should NOT be murdering a rig running at about the same hash by 70-100 percent earning wise when in theory it should be a 50/50 split or close to it.....    Just by setting a manual diff at the block diff  this would also apply to anyone that set above the share diff for example on that 20-40 range coin if they set to 10000 for example they would also out do the 200 g rig on the base diff earning wise due to again putting shares in that carry more weight to them....   As if the 200g user puts in say 50 shares at the base diff it will not factor in the value of them shares.... so they could put in 20 at the base diff and the other 30 at block hits but it only factors them shares at the base rate......


Where the user set to the block hit put 30 shares in at the higher diff setting is going to to see them shares treated as 40 k shares hence the massive diff in the payout rates as they would have to payout at 16 coins over the pplns due to only 1 share to the block which is reflected in the higher earnings due to the diff setting = the block diff  so it not poss for there to be other shares to the block when the user is using a block diff...  eg the same as a solo mine where ya have only 1 share to a block 0

I can 100% prove your statements to be wrong.  To start with you were setting a 40k diff on a node where a normal share is in the 20-40 difficutly range, of course you're going to take the majority of the payments when you submit shares that 1000 to 2000 times normal diff.  At this moment that particular node is reporting a diff of 4 for a share so again, you'll takeover the sharechain the second you submit a couple 40k diff shares.

Now, I can prove that variance over a short period of time can easily.  Here is a link to my personal mining address on p2pool and the payouts I have received recently. http://minefast.coincadence.com/miner.php?id=1EEoGzEbUYmak9y5wegvqEqd939ATZy7Bp  As you can see from August 4th to August 6th my payouts double, although my hashrate never changed.  During this period I had 8 shares found during the 72 hour window, and with the difficulty at the time, I probably should've had 4 (4.5M to 5M shares).  Just because your miner has gotten lucky, does not mean that it's not due to variance.  I don't see why much of this even matters, because by the time someone has the hashing power to even set their idividual difficulty at the bitcoin difficulty, why would you be mining on p2pool, it would be MUCH more profitable to just run your own bitcoind and solo mine.  And actually, there are advantages to the pool as a whole if large miners would set their difficulty to such that they would only submit shares a share from every hour to ever few, I'd personally probably shoot to ever 4 hrs if I had the hashing power, it lowers the number of shares submitted to the pool and thus helps control the difficulty re-targeting by not constantly submitting shares.

Hello to all you p2pool node operators. I've forked p2pool, added an adaptive difficulty patch and exposed some additional bitcoind RPC calls. You might want to check it out:

https://github.com/jramos/p2pool

I've also added an updated web interface. It provides:

• Audio and text notifications for new shares and blocks (on most modern browsers)
• Automatic conversion of BTC to USD
• Integration with Blockchain.info APIs
• Detailed information about each share, block and transaction
• All the graphs you know and love

https://github.com/jramos/p2pool-node-status

Live demo here: http://bitcoin.ramosresearch.com/static/

As a few seen 1st hand yesterday I can manulate the payments on the p2 pool that is a fact<<<<< (any user can)   As two rigs running at the same hash should be getting roughly the same payout NOT a 70-100 percent diffence in payout and was been clearly shown in that test on a alt coin it would not be hard for the top btc hashes etc a farm or 2 to change everyone payouts as proven yesterday via the alt coin test using the p2 pool set up I am more than happy to set the rig up on any p2 alt coin node that has a low diff block hit so I can mirror a farm on the btc P2 network with the live test again for others to see this 1st hand and the flaw in the p2 system via setting that manaul diff to the block size or any joe bloggs can test themselfs It does not matter if you are on the node for 2 hours or 24hrs there no way 2 rigs running at the same speed should see a 100 percent diff in payouts when the second rig myself had only been on there for 6 hrs and still coming up to speed in terms of the pplns shares but yet was able to murder the earnings of the other user for the whole 6hrs I was on there and they had been there for days....(hate to see what the diff would of been after 24 hr or 2 days when I was fully charged up on the pplns side of things)   By only putting blocks in as shares   vs the other user that was at a guess using the default settings of using a payout adresses only    vs myself using the payout address/40000 command which was the block hit on the coin that was been test.....


I am about to carry out another test on the same p2 node with an old jap miner (5-6 gh) this is going to be intersting to watch to see what happens when that set to the block hit value NOW in theory this miner should only see about very small payments of about 1 coin but if my hunch is right on what going to happen it going to be a hell lot higher....    Atm there is a miner of 200g on the node and another running at 30 odd g....   In theory with there been 240 g in total on there and the coin block size of 16 coin per block I should only see payouts of 0.5 coins per block as I have only .03 percent of the hash time will tell how much this bug can be played....

change of plans since the 200g rig there that can be used as a control I will throw an ant over there on the same payout address etc as yesterday again set to only put blocks in as shares to repeat the test from yesterday so people know i am the user with the jsn address and well the results are starting to mirror yesterdays

Also as a footnote there is another rig there running at roughly 800 g and I am also murdering that rigs earnings as can be seen on that xjo p2 node there something wrong when a 200g rig is also murdering a 800g rig earnings by just having the /# set

fire000 - I'm not sure running your test for such a short period is viable.  I know you keep saying there's issues with p2pool payouts and the system as a whole and you're not making as much compared to a proprietary mining pool, but take a look at the below thread:

https://bitcointalksearch.org/topic/benchmark-p2pool-vs-btcguild-vs-eligius-416933

Share weights, diff variences, everything else aside, there's someone who has an apples to apples test, running over 7 months now, comparing p2pool with 2 other well known, well run pools.  I tend to believe his numbers, as I personally am seeing no difference, if not a slight increase in my own earnings from switching off of BTCGuild to here.  I am not seeing any dip, and I'm happy with my mining outputs.

I'd suggest running your own test (long enough to normalize) to see whether your experiences are the same.  We can argue about diff numbers, share weight all day, but come up with a similar test, with real world numbers, and then it's a valid argument.  The thread referenced above is what got me to switch to p2pool, after researching and seeing real numbers of someone else.

Was this alt coin using the standard 3 day PPLNS system?  If so, a 6 hour test isn't a fair comparison.  They should even out over time, and I don't think 6 hours is long enough.


That's an interesting theory.  It means if everyone did it, we'd all be solo mining.  It also means until you get that block size share, you won't get anything.  You could go years without a payout.

M

Wait for the screen shots it will blow people away....   There a flaw in the payout system on the p2 network as a few seen 1st hand last night while I was running the live test as the comments over the last page or so highlight.....  

2 Rigs hashing at the same speed....   there was a massive difference in the payouts 1 was me hashing on the block diff   and the other was joe bloggs the public on a min share diff....    At one stage there the payouts I was getting were well over double the other user so 100 percent higher and at no time in the 6 hour test did my payout fall below the other users most of the time they were between 70 -100 percent higher than that of the other user hashing at the same rate as me..... so in theory it should of been a 50/50 split not what the results showed if even surpised me how much the difference was in the payouts between the miners

Anyone can test this on a ALT coin and see it 1st hand what that manual overwrite does to the p2 in terms of payouts and having it set to only put blocks in as shares vs a user running at the node min diff,


IT also backs the claim up that if a user was to set to the block diff there is only 1 share for that block as the system would have to pay out the block value to the user if they were set to 19.7g and hit as they would have to be paid =25 btc over the ppnls to reward them for that 1 share that they have set as their diff as at 19.7g that equals 25 btc in terms of payout value on that one share....    They also see more that the 25 btc as they collect in the other rds while those shares disappear...   As highlighted in the payout numbers and the earning diffence between the 2 users on the test I did last night on an alt

Ure wrong, because not every share have same power - it depends on pool power.

after that test last night and letting all the payouts roll into that address there appears to be yet another flaw in the p2 system if one sets to the block diff on a coin....    

if you look at that address above it has been credited 269 coins based on 12 shares I put in which all where blocks now simple maths 12 * 16 = 192 coins    so off that I have been overpaid about 78 coins.... (feel sorry for the other user that was on the node at the time on the same hash rate as I have taken this guys earnings)   <<<<< this can not be the block finders bounus as I was the only share in the block so no matter what the payout = 16 coins as the bonus comes from the block I will rerun the test in the next day or so and screen shot it and gave the miner rig stats so people can see the block finds vs payout address and also will screen shot the node to show what effect it has on other miners on the node running at the same speed etc in terms of their payouts as afew seen last night while I was doing the live test there was a massive different in the payouts between a user putting in shares at the min share rate and myself only putting shares in at blocks via the manual diff setting

Yeah.. that's what I'm starting to finally come to terms with. Just leave it alone. Just hard watching lucky miners with twice the payouts and less HR. I just need to stop looking I guess.
0.04217808 so far today with under 1TH/s is pretty damn good/lucky.

My best share on one s3 in the last 24hrs is 1.5 million LMAO. My undervolted S1 is better.

There's really no need.  In a nutshell, here's the deal:

Setting your share difficulty to anything LESS than the p2pool minimum is pointless.  You can't submit a share to the chain that does not meet the minimum difficulty requirement.
Setting your share difficulty to anything MORE than the p2pool minimum will mean you expect to find fewer shares, but when you do the share is weighted more (i.e. it's worth more).

Hey thanks for the info..
So currently I'm showing 5202394.44 share diff on the pool. I should try something like /5203000 and just keep an eye on the current diff for adjustment?

the /#   needs to be above the current diff on the node for the share....    So for example if the current diff is say 5500000 a share that setting needs to above that number so for example that /#    would need to be set at /5500001 <<<< or higher    the higher that number the more value the share will carry....    


Just a footnote you will hit less shares BUT they carry a higher weigth to them as they are above the min share value and it adjusts the payments to match the value of the shares you are putting as can be seen over on that test

So address/40000. I need to go back and read up. How often would you adjust that number and do you base that off the current share diff on P2Pool? I need to try this.

via the manual override command example    server -u JSnJaDXZMtNRvdSSrLQXhc9bJsthKF3zn2/40000 -p x

do not worry about the address bit as that a non btc address   the part after it the /40000 that is a diff setting you can apply a diff to you miner over the normal share value.....     Now if you hit at the high diff you have set it will carry more weight that the lower default diff....    As can be seen over on the pool I have a miner sitting on atm where the effects of using that diff setting can be seen clear as days and the effect it has on earnings if it set higher that the min share value.