[1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool

[phillipsjk]

Between 12% hardware error rate, 20% penalty for no long polling, and another 10% for slow CPU lag (probably 3 seconds), I am getting 43% local DOA though.

Hard numbers?

Hardware error rate:
Blade thinks it is hashing: 10777 MHash/s
P2Pool sees an actual rate of: 9534MH/s
(1-(9534/10777))x100%=11.5%

No long polling penalty
Blade has 32 relatively independent hashing chips.
Blade handles 147 shares per minute.
147 shares/32chip.minute=4.59375shares/chip.minute
(60seconds/minute)/(4.59375shares/chip.minute)=13.1seconds.chip/share
Expected DOA: (13.1seconds.chip/share)/(2*30 second.share-interval)*100%=22%(chip/share-interval)

Slow CPU lag:
Probably the number you are most interested in, but the most shaky as well.
P2Pool reports the "Bitcoind GetBlockTemplate Latency" is an average of  0.454s
With a 10 minute sample period, P2Pool appears to use ( 17.19%/2.34%=7.35x) the CPU time bitcoind does.
7.35x0.454s=about 3.3seconds of CPU lag.
Penalty with 30 second round time: 3.3/30*100%=11.1%
The network probably has some lag as well.

Are the numbers sane?
The above percentages (neglecting hardware error rate) add up to an expected local DOA of 33.1%
P2Pool reports a local DOA rate (over last ~12 hours) of: (DOA mean 3.45GH/s)/(Total mean 9.48GH/s)x100%=36% DOA

My DOA number will vary by 12%; that is why I thought the hardware error rate was maybe counted.

Conclusion
I have a slow CPU, and including zero fee transactions does no generally push my bitcoind latency over 1 second (with 500000 byte blocks). Yes, including zero-fee transactions will increase your latency: as does including paying transactions. Currently fees are insignificant compared to the block-reward.

Assuming 500mBTC of fees per block, we are talking a 2% bonus on 25BTC blocks.
If trying to fill a block pushes your Bitcoind latency up to 1 second, you can expect to get orphaned 0.2% of the time (when you otherwise wouldn't) assuming a 500 second block interval.
Edit: That seems to imply that paying transactions need to add up to at least 50mBTC to compensate for the CPU lag of including them. My network probably has at least 1 second of lag if I send the whole block at once (not sure how much gets "pre-sent".

[1714093761]

1714093761

[1714093761]

1714093761

[1714093761]

1714093761

[sconklin321]

Has anyone heard the outcome of the Petamine vote?  Just curious if we're going to see a 1 ph/s influx and a jump of share difficulty.

[oda.krell]

FYI, I just opened a thread on the speculation subforum about incentivizing p2pool, and how it is probably in the interest of investors and (most) traders to do so. Thought you might be interested, and would like to get your input as well:

https://bitcointalk.org/index.php?topic=667821.msg7545494

[Muhammed Zakir]

Between 12% hardware error rate, 20% penalty for no long polling, and another 10% for slow CPU lag (probably 3 seconds), I am getting 43% local DOA though.

Hard numbers?

Hardware error rate:
Blade thinks it is hashing: 10777 MHash/s
P2Pool sees an actual rate of: 9534MH/s
(1-(9534/10777))x100%=11.5%

No long polling penalty
Blade has 32 relatively independent hashing chips.
Blade handles 147 shares per minute.
147 shares/32chip.minute=4.59375shares/chip.minute
(60seconds/minute)/(4.59375shares/chip.minute)=13.1seconds.chip/share
Expected DOA: (13.1seconds.chip/share)/(2*30 second.share-interval)*100%=22%(chip/share-interval)

Slow CPU lag:
Probably the number you are most interested in, but the most shaky as well.
P2Pool reports the "Bitcoind GetBlockTemplate Latency" is an average of  0.454s
With a 10 minute sample period, P2Pool appears to use ( 17.19%/2.34%=7.35x) the CPU time bitcoind does.
7.35x0.454s=about 3.3seconds of CPU lag.
Penalty with 30 second round time: 3.3/30*100%=11.1%
The network probably has some lag as well.

Are the numbers sane?
The above percentages (neglecting hardware error rate) add up to an expected local DOA of 33.1%
P2Pool reports a local DOA rate (over last ~12 hours) of: (DOA mean 3.45GH/s)/(Total mean 9.48GH/s)x100%=36% DOA

My DOA number will vary by 12%; that is why I thought the hardware error rate was maybe counted.

Conclusion
I have a slow CPU, and including zero fee transactions does no generally push my bitcoind latency over 1 second (with 500000 byte blocks). Yes, including zero-fee transactions will increase your latency: as does including paying transactions. Currently fees are insignificant compared to the block-reward.

Assuming 500mBTC of fees per block, we are talking a 2% bonus on 25BTC blocks.
If trying to fill a block pushes your Bitcoind latency up to 1 second, you can expect to get orphaned 0.2% of the time (when you otherwise wouldn't) assuming a 500 second block interval.
Edit: That seems to imply that paying transactions need to add up to at least 50mBTC to compensate for the CPU lag of including them. My network probably has at least 1 second of lag if I send the whole block at once (not sure how much gets "pre-sent".

Thanks for that! Explained in simple words but understood a lot!
Kindly,
        Muhammed Zakhir

[sconklin321]

Well someone is definitely having their prayers answered by the hashing gods...whoever it is, don't stop praying...5 blocks in 3 days is nice.

[windpath]

Well someone is definitely having their prayers answered by the hashing gods...whoever it is, don't stop praying...5 blocks in 3 days is nice.

Agreed, what a great few days for p2pool!

[sconklin321]

Well someone is definitely having their prayers answered by the hashing gods...whoever it is, don't stop praying...5 blocks in 3 days is nice.

Agreed, what a great few days for p2pool!

I just wish I had my new miners a day or two earlier to have more shares.

[andmax]

There is an update for the p2pool software, not sure when it came out but I just applied the updates and now my S2's are hashing at the full 1ths.

[mdude77]

There is an update for the p2pool software, not sure when it came out but I just applied the updates and now my S2's are hashing at the full 1ths.

Didn't help here.

I just upgraded my node to the latest git.  Actually seems worse than before, barely above 900gh/s.  Before it was 930gh/s.

M

[windpath]

There is an update for the p2pool software, not sure when it came out but I just applied the updates and now my S2's are hashing at the full 1ths.

That is great news! Just took a look at GitHub though and looks like the only changes were for Terracoin?

Regardless, great to see Forrest active on p2pool GitHub again! Thanks.

[windpath]

S2's and upcoming S3's are critical to p2pools success.

The bounty for Forrest to get them running smoothly is up to BTC0.75

"for a working solution for p2pool/anitminer S2 & S3 compatibility with the Ants achieving expected hashrate and shares."

Anyone is welcome to contribute to the donation address, bounty is paid to forrestv (or his designee) upon a working, distributed solution or returned in full in 30 days (July 25th), I will pay return tx fees for any donation over BTC0.001, if your donation is less, consider it donated.

Bounty Address: 1Ft1DZ4EcpH1jFttDoPwsFbe11a6diyZ1H

https://blockchain.info/address/1Ft1DZ4EcpH1jFttDoPwsFbe11a6diyZ1H

Signed message:

Code:

forrestv p2pool antminer s2 and s3 compatibility fix.

Message signature:

Code:

G7zuvJ7/LPv7LpZveAsQWet1gAR42bnwFRVPB7NsnGqPnWkN8bqSSN1UwR+NnfDs/8LkruqZcVsTTdfcy5pcNj4=

Please contribute.

[IYFTech]

Great. More alt-coins that can't be mined.......

Really, wtf is he doing faffing around with alt-coin additions when there are far more urgent things to be done? Concentrate on the btc side of p2pool - there are plenty of forks for alt-coins already, ffs.

[sconklin321]

S2's and upcoming S3's are critical to p2pools success.

The bounty for Forrest to get them running smoothly is up to BTC0.75

"for a working solution for p2pool/anitminer S2 & S3 compatibility with the Ants achieving expected hashrate and shares."

Anyone is welcome to contribute to the donation address, bounty is paid to forrestv (or his designee) upon a working, distributed solution or returned in full in 30 days (July 25th), I will pay return tx fees for any donation over BTC0.001, if your donation is less, consider it donated.

Bounty Address: 1Ft1DZ4EcpH1jFttDoPwsFbe11a6diyZ1H

https://blockchain.info/address/1Ft1DZ4EcpH1jFttDoPwsFbe11a6diyZ1H

Signed message:

Code:

forrestv p2pool antminer s2 and s3 compatibility fix.

Message signature:

Code:

G7zuvJ7/LPv7LpZveAsQWet1gAR42bnwFRVPB7NsnGqPnWkN8bqSSN1UwR+NnfDs/8LkruqZcVsTTdfcy5pcNj4=

Please contribute.

I happen to really agree with this, just wish I had some spare BTC to add, maybe'll I'll send 0.01 or 0.02, just emptied my wallet as I'm in the process of going from U1's and U2's to Rockminer R-Boxes.  Currently debating whether to go S3 when we get details, or go up to the next level of rockminer when they are released.  If I want to stay with P2Pool, I don't think I have much choice, well at least not until we know whether or not the new rockminer will work well with P2Pool, luckily the R-Boxes do so I'm hopeful.  I'd really like to see this fixed though, because I think the S3 might be the better way to go.

[nreal]

Reservoir Dogs

Nice Guy Eddie: C'mon, throw in a buck!
Mr. Pink: Uh-uh, I don't tip.
Nice Guy Eddie: You don't tip?
Mr. Pink: Nah, I don't believe in it.
Nice Guy Eddie: You don't believe in tipping?
Mr. Blue: You know what these chicks make? They make shit.
Mr. Pink: Don't give me that. She don't make enough money that she can quit.
Nice Guy Eddie: I don't even know a fucking Jew who'd have the balls to say that. Let me get this straight: you don't ever tip?
Mr. Pink: I don't tip because society says I have to. All right, if someone deserves a tip, if they really put forth an effort, I'll give them something a little something extra. But this tipping automatically, it's for the birds. As far as I'm concerned, they're just doing their job.
Mr. Blue: Hey, our girl was nice.
Mr. Pink: She was okay. She wasn't anything special.
Mr. Blue: What's special? Take you in the back and suck your dick?
Nice Guy Eddie: I'd go over twelve percent for that.

[sconklin321]

Wow, there's another one.  Please...no one change anything they are doing!!! I like a 2 a day average!!!  That's 6 in 70.5 hrs.!!!

[jedimstr]

Wow, there's another one.  Please...no one change anything they are doing!!! I like a 2 a day average!!!  That's 6 in 70.5 hrs.!!!

Boom... and another.... make that 7 in 72 hours

In honor of our good fortune lately, it's time to break out some Animal House...

[Muhammed Zakir]

It is rapidly growing. Very happy to know it.

Kindly,
        Muhammed Zakhir

[sconklin321]

Wow, there's another one.  Please...no one change anything they are doing!!! I like a 2 a day average!!!  That's 6 in 70.5 hrs.!!!

Boom... and another.... make that 7 in 72 hours

In honor of our good fortune lately, it's time to break out some Animal House...

I for one have to second this sentiment.

[wlz2011]

As jedimstr said, I did receive a sizable donation from the Litecoin Project (EDIT: credited wrong organization), and I have since been brainstorming and prototyping ways to fix the core problems with P2Pool - scalability and excessive variance for small miners. I prototyped a way to do trustless verification of blocks with O(1) storage so that we could free most P2Pool nodes from needing a local *coin node. I've spent a ton of time thinking of ways to implement parallel sharechains so that we can have shares more often, decreasing variance for small miners, while decreasing the strict latency requirements for hardware, sadly without any concrete results yet. Because this is a hard problem. I spent hundreds of hours in the summer of 2011 thinking about how P2Pool could possibly work, in a time before merged mining or pools with coinbase payouts even existed. Any further improvements that get us out of this local maxima that we're stuck in will require comparable amounts of thought. Some relief though - a few days ago, when people prodding renewed my eagerness, I may have made a breakthrough in ideas for how to get parallel sharechains to work (which I discussed in #p2pool).

Despite that, it is hard to work on this project when most people insist on cutting off the pool donations to me. When I see guides telling people to use --give-author 0 without any mention made of what it does, people blindly following them, or people asking me for help, providing screenshots of them running P2Pool with donations disabled, it hurts a bit. Actually, more than a bit. Right now, I get 0.09% of the revenue from P2Pool due to node donations (which default to 1%), which likely means that about 90% of people have completely disabled donations. That doesn't result in much revenue. Just today I spent about 10 hours getting http://p2pool.info/ working, which is the equivalent of two weeks of pool donations - if I were being paid minimum wage! I don't have a job (I am a student) and I do have other side-ventures that I'm working towards that look a bit more optimistic, so pardon me for rationally allocating my free time.

I don't believe anybody disabled donations because of a lack of development - pool donations have been comparably low for a very long time, since before this development hiatus - so I don't think I'm being "punished" for not working hard enough or obviously enough. I don't think that most miners ever think about their donation amount again after setting it or pay attention to development. (I don't blame them for not dedicating their lives to tweaking P2Pool. ) But this is vicious downward cycle and anyone running P2Pool with donations lower than the default while demanding improvements (or even planning to use P2Pool for an extended amount of time) is a hypocrite. It's simply a lack of foresightedness when miners decide to prioritize an amount of income that is invisible in the noise of variance of P2Pool payouts over the sustainability of P2Pool. Perhaps we'd be better off if I hadn't changed the mandatory 0.5% fee to an optional 0.5% donation (does anyone here even remember that?).

I do plan to continue working on P2Pool, and eventually great changes will happen. Unless something changes though, don't expect it to happen too quickly, and don't be surprised if people who disable donations get bugged a bit more when they start P2Pool (or with any other similar change to increase donations).

Hello, Mr Forrestv
      I represent P2POOL Chinese users thank you, thank you always hard to pay, we have been using a 1% donation to express our gratitude.
But in December 2013, P2POOL encountered a problem, bitcoin difficulty of continued growth in, small miners can only choose to leave.
I have tried to contact you, try to communicate. I hope you can solve some problems, but you disappeared.
P2POOL income has been declining, we have suspended the donation. 2014 we did not donate, because you do not continue to work for P2POOL, up to eight months.
P2POOL 2013 in December, creating a very good income. April 23, 2014, I took my Chinese friends, try to P2POOL impetus,
We have assembled a force of about 250T count P2POOL, until June 20, 2014. We are their own servers, non-profit pools, revenue is not good, you're better than that
We know, we have no reason to donate to you. We continuous losses, still insist on P2POOL mining BTC.
We are very sad, so June 20 we left P2POOL. You are very lucky to have a count of about 300T of unknown force mining in P2POOL, they also have a lot of 0 donations
Did you say they are hypocrites? ? ? If they leave, you will lose your time and effort spent countless while building P2POOL, P2POOL will collapse.
Your time is limited, miner's time is limited. We should understand each other.
I saw the June 22 and 27, have updated and P2POOL again， shown remarkable performance gains, if these updates are really effective to improve the income,In July I will return to the set count force P2POOL, and consider donating. But if the performance is still so poor, we do not need to return to P2POOL.

[nreal]

"These are the results of the survey. 68% of all unit holders have cast their vote:

- Yes, PetaMine should switch over to P2Pool – 96%
- No, PetaMine should stay with ghash.io – 3%
- No, Petamine should switch to another pool like Eligius or BTCGuild – 1%

The result is clear that a big majority of unit holders prefer a switch to p2pool. As we were performing additional tests the past week, we have to inform you that unfortunately the Bitfury systems are not compatible with p2pool. We are currently working on a solution for this and we will switch to p2pool as soon as we are able to fix this problem."

Sharedifficulty rising..  

I cant wait to see if Peta makes more coins with p2pool then before...