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

[HellDiverUK]

the ks-2g could run one adequately, but that, the ks-4g and the ks-16g are sold out

You're dreaming if you think a single core Atom will run p2pool with any efficiency.  A Celeron 847, which is WAY faster than an Atom really struggles, even running a text-only Ubuntu server install on SSD. 

The slowest machine I've had it running properly on is a Pentium G620 (dual core 2.6GHz Sandy Bridge).

Currently my p2pool node runs on my mining/workstation machine, which is an i3-3220T with 8GB RAM and SSD drive.

[daemondazz]

Changing the topic slightly, my node is now showing peaks of 14TH/s and a 24hr mean of 10.7TH/s, so it might be worth updating the thread topic

[Krellan]

Nice, I just noticed P2Pool has appeared in this list for the first time:

http://organofcorti.blogspot.com/2013/09/september-1st-2013-weekly-pool-and.html

Any thoughts on perhaps spending 7-or-so bytes per block, to put "P2Pool" in the coinbase signature, to make it easier for P2Pool mined blocks to be identified as such, for purposes of making these charts?  It's an exciting time for P2Pool, as it is now over 1% of the entire Bitcoin network, and this will draw more attention (and thus hopefully more people joining the pool and adding hashpower).

Josh

[organofcorti]

Nice, I just noticed P2Pool has appeared in this list for the first time:

http://organofcorti.blogspot.com/2013/09/september-1st-2013-weekly-pool-and.html

While I appreciate the props, p2Pool has been in the weekly pool stats on the blog since the start:

http://organofcorti.blogspot.com.au/2012/07/weekly-pool-statistics-29072012.html


This is the first time p2Pool was in my weekly stats.

Any thoughts on perhaps spending 7-or-so bytes per block, to put "P2Pool" in the coinbase signature, to make it easier for P2Pool mined blocks to be identified as such, for purposes of making these charts?  It's an exciting time for P2Pool, as it is now over 1% of the entire Bitcoin network, and this will draw more attention (and thus hopefully more people joining the pool and adding hashpower).

Josh

Or why not sell coinbase space to the general public as a vanity signature? I'll pay a couple of btc to have my name as a coinbase sig for a block.

[kjj]

p2pool already has several unique features that can be used to identify the blocks.

And if you want your signature in a block, the patch to bitcoind is trivial. 

[organofcorti]

p2pool already has several unique features that can be used to identify the blocks.

Are they the same every block or do they vary? Also, could you share what some of them are?

And if you want your signature in a block, the patch to bitcoind is trivial. 

Except I'm not a miner anymore

[ShadesOfMarble]

And if you want your signature in a block, the patch to bitcoind is trivial.  

How is it done?

[kjj]

And if you want your signature in a block, the patch to bitcoind is trivial.  

How is it done?

I use this in the binary for p2pcoin's bitcoind:

Code: (init.cpp)

    const char* pszP2PCOIN = "[P2PCOINv0]";
    COINBASE_FLAGS << std::vector<unsigned char>(pszP2PCOIN, pszP2PCOIN+strlen(pszP2PCOIN));

I also have command line options for -addtag and -addhextag that I use to add my personal tags.  They are done pretty much the same way.

[ShadesOfMarble]

Thanks!

[gmaxwell]

Are they the same every block or do they vary? Also, could you share what some of them are?

The last output in the coinbase is the special P2P hashchain commitment. It has zero value and it now looks like OP_RETURN PUSH <data>.

[organofcorti]

Are they the same every block or do they vary? Also, could you share what some of them are?

The last output in the coinbase is the special P2P hashchain commitment. It has zero value and it now looks like OP_RETURN PUSH <data>.

Thanks gmaxwell. Is there anywhere a script could pull the special P2P hashchain commitment from if and when it changes (in order to do a regex search of the coinbase)? Or is that something you'd have to run a p2Pool node in order to do?

[forrestv]

Are they the same every block or do they vary? Also, could you share what some of them are?

The last output in the coinbase is the special P2P hashchain commitment. It has zero value and it now looks like OP_RETURN PUSH <data>.

Thanks gmaxwell. Is there anywhere a script could pull the special P2P hashchain commitment from if and when it changes (in order to do a regex search of the coinbase)? Or is that something you'd have to run a p2Pool node in order to do?

The <data> will always be different and depends entirely on P2Pool's sharechain, so I don't think it's too useful. Instead, the second to last txout will always be to the donation address (1Kz5QaUPDtKrj5SqW5tFkn7WZh8LmQaQi4) and then you can match the template of the last txout (0 value to OP_RETURN PUSH <data>). Either of those alone is probably good enough, but combined should work very well.

[organofcorti]

Are they the same every block or do they vary? Also, could you share what some of them are?

The last output in the coinbase is the special P2P hashchain commitment. It has zero value and it now looks like OP_RETURN PUSH <data>.

Thanks gmaxwell. Is there anywhere a script could pull the special P2P hashchain commitment from if and when it changes (in order to do a regex search of the coinbase)? Or is that something you'd have to run a p2Pool node in order to do?

The <data> will always be different and depends entirely on P2Pool's sharechain, so I don't think it's too useful. Instead, the second to last txout will always be to the donation address (1Kz5QaUPDtKrj5SqW5tFkn7WZh8LmQaQi4) and then you can match the template of the last txout (0 value to OP_RETURN PUSH <data>). Either of those alone is probably good enough, but combined should work very well.

Thanks forrestv. I didn't think of looking for the donation address to confirm.

[spiccioli]

Can I say I'm disappointed?

I went from an AMD Sempron (with the second core unlocked by BIOS) to a six core Phenom II and GBT time is the same!

http://imgur.com/St1vbAO

WTF?

spiccioli

[HellDiverUK]

p2pool really needs a fast clockspeed.  More cores won't make any difference. 

Only time I got decent performance from it was running it on an i3-3220.  3.3GHz, running bitcoin-qt with high priority on Windows.  It floated around 0.2s latency then.  On a Celeron 847 (dual 1.1GHz) it ran around 0.6-1.0s, and a Sempron X2 190 it was about 0.5s on average.

[spiccioli]

p2pool really needs a fast clockspeed.  More cores won't make any difference. 

Only time I got decent performance from it was running it on an i3-3220.  3.3GHz, running bitcoin-qt with high priority on Windows.  It floated around 0.2s latency then.  On a Celeron 847 (dual 1.1GHz) it ran around 0.6-1.0s, and a Sempron X2 190 it was about 0.5s on average.

I thought (hoped) that the bigger caches of the Phenom and the ability to run p2pool code and bitcoind (which has a lot of threads) on more cores would have ended helping up GBT times... I was wrong

spiccioli

[gyverlb]

forrestv: maybe it's time to measure the Bitcoin p2pool hashrate in TH/s in the thread's title?

I suppose the current title reflects the weekly average (we are at 11.6TH/s for the last 24h).
Looking at the monthly graph we have a sustained rise over the month with a sudden spike 3 days ago. We are tailing Eligius/Bitminter/Ozcoin, not there yet but not so much to add to catch up, it's 10TH/s+ time !

Good job of making p2pool ASIC friendly, seems many miners have found out p2pool is working beautifully for them.

[Krak]

forrestv: maybe it's time to measure the Bitcoin p2pool hashrate in TH/s in the thread's title?

I suppose the current title reflects the weekly average (we are at 11.6TH/s for the last 24h).
Looking at the monthly graph we have a sustained rise over the month with a sudden spike 3 days ago. We are tailing Eligius/Bitminter/Ozcoin, not there yet but not so much to add to catch up, it's 10TH/s+ time !

Good job of making p2pool ASIC friendly, seems many miners have found out p2pool is working beautifully for them.

Pyramining.

[gmaxwell]

I suppose the current title reflects the weekly average (we are at 11.6TH/s for the last 24h).

I've been updating it to reflect the weekly average... daily is kinda noisy and I'd rather understate it during growth than have it bounce all over or be accused of sandbagging.

[twmz]

Are they the same every block or do they vary? Also, could you share what some of them are?

The last output in the coinbase is the special P2P hashchain commitment. It has zero value and it now looks like OP_RETURN PUSH <data>.

Thanks gmaxwell. Is there anywhere a script could pull the special P2P hashchain commitment from if and when it changes (in order to do a regex search of the coinbase)? Or is that something you'd have to run a p2Pool node in order to do?

The <data> will always be different and depends entirely on P2Pool's sharechain, so I don't think it's too useful. Instead, the second to last txout will always be to the donation address (1Kz5QaUPDtKrj5SqW5tFkn7WZh8LmQaQi4) and then you can match the template of the last txout (0 value to OP_RETURN PUSH <data>). Either of those alone is probably good enough, but combined should work very well.

Thanks forrestv. I didn't think of looking for the donation address to confirm.

That's what p2pool.info does, essentially.  Note, I would also suggest excluding any blocks that do not have at least 25 outputs.  There are some non-main p2pool installations running out there (either forks, or someone running their own mini-pool with the p2pool codebase).  They all have only a few people mining on them, as far as I know, so their blocks only have a handful of outputs (besides the donation address and the special last transaction).