Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 759. (Read 2591916 times)

Well p2pool may not grow any larger so it may not matter.  Still at 200 MH/s p2pool growing larger doesn't help your variance.  In the long run you expected return remains the same however in short run higher hashing power results in more variance.

If/when p2pool grows larger it becomes more of a concern.  If you want to use a conventional pool try p2pool hybrid pool (it provides a 1 difficulty shares as a front end to p2pool so you can still support p2pool indirectly).

So in other words little guys like me doing 200 MH/s should probably just stick with a traditional pool until these p2pool issues are settled.

I have indicated that a backbone & subpool model is likely the best and the most lasting however it is the most complex.  In the short term p2pool front ends or proxys (conventional pools using p2pool as backend) which accept diff 1 work buys us some time.  forrest also indicated he will consider splitting p2pool at around 500 GH/s into two instances.


As far as I can tell three major things need to change (and be extensively tested) to allow a multilayered p2pool network:

1) Increase LP interval from 10 sec to something like 30 sec or 60 sec.  This is necessary otherwise sub pools LP interval will be too small. Remember a sub pool is dependent on the main network thus it's effective LP it whenever an LP occurs on the subpool OR main network.  Yes increasing LP target interval will increase difficulty so it shouldn't be done until sub pools are ready.

2) Allow p2pool shares to be assigned to multiple addresses.  Currently when you find a share the network records a single address (your payment address) in the reward split.  Sub pools will need to split 1 share across multiple miners (based on their current share chain work distribution).

3) Modify p2pool code to work as a sub pool. It means building a "subpool" level share chain, and reward splits and being aware of backbone p2pool status (blocks, LPs, share split, etc). The subpool is "invisible" to backbone p2pool accept when subpool submits shares.

Why can't p2pool stay the way it is with small miners migrating to p2pool subpools that send out 1 difficulty shares so small miners can still get their cut?

I would be willing to setup / run a small p2pool pool (i coin this a p2p pool, you heard it hear first) without a fee but would need some open source software to run since my time is limited.

Should we have a "CPU And Crappy GPU Miners" p2pool, and a "Big Ass Hard Core Rig Monster" p2pool?

Anyone else notice the trend of rising hashing power and falling users?  The avg hashing power per user is rising.

Higher network hashing power = lower block variance.
Higher network hashing power = higher share variance.

The above combo is a win-win for large miners but a win-lose for smaller ones.  Taken to the extreme at 600 GH/s avg block time is roughly 3 hour (current difficulty).  However at 600 GH/s share difficulty is ~1400 and for a 500 MH/s miner that means a 3.3 hour share time.  Further increases in network hashing power mean increased not reduced effective variance.

Of course there are potential solutions I am just pontificating.  

Ah, I figured it may have been something like that.
I missed the part where this thread is about P2Pool! lol
That would be awesome, running P2Pool on a Pi, controlling a cluster of FPGAs or even full GPU machines, like D&T said.
I can't wait to see where this goes!

On another note, I'm trying to get my 7970 hashing in the pool, but I can't seem to get it configured correctly with diablominer.
It works fine with cgminer, but I get better hashrates with diablo.
Anyone got this combo (7970+diablominer+P2Pool) working?
I'd love a few pointers in the right direction.

Hell I don't have an FPGA but if someone can get p2pool and bitcoind running on one I would buy one (and donate the developer).  Pretty cool to have p2pool server for a 10GH/s farm running on 5 watts.

It isn't hashing, it is simply acting as a server to hold p2pool and FPGA control software. It is awesome because it removes the need for a host computer to do the same task, and instead runs on a low power embedded device.

That's awesome....but why?
I don't mean that in a sarcastic way, I'm actually asking what makes it awesome?
Without a GPU, wouldn't hashing on a Rasberry Pi be in the 1-2 Mh/s range?
Please tell me I'm missing something, those Pi's look SWEET and I need an excuse to get one!

Short note:

Bitcoin-qt, p2pool and Modular python bitcoin miner sucessfully running on 600 mhz ARM board.

I expect to get it running on rasberry pi as soon as i get my hands on one

Bit of an unlucky day today, boys. I thought we were starting out hot with two blocks early on, but the last block is just dragging.

That wasn't found by the main p2pool but was found by a small fork of miners that didn't update.  Blocks found by the un-updated splinter group are now intentionally excluded.  Those miners are not part of the reported pool hashrate and so their blocks shouldn't be credited against the main pool's hashrate for the purposes of things like round length and luck statistics.  I suppose I could still display "fork blocks" in some distintive way, though.

fork that.

That block was found by a fork of people who didn't upgrade to the latest version/protocol change, so p2pool.info is not "missing" anything.

Beware The Gambler's Fallacy. 

It seems p2pool.info miss block 169625 http://blockexplorer.com/block/00000000000006ca7bb5be0dbd0e2b81ea3b179f17218d6a2ca3116c4678ff78 (2012-03-04 16:03:22)

I think it is time for some lucky blocks! 

You can't do what you are asking.

I asked because if there were a good reason for wanting this, then it might be worth exploring further and possibly doing the development to make it possible. 

Thanks D&T, that make sense to me.  I believe I know why p2pool.info was so low yesterday...I was updating p2pool software (3/4 deadline) as well as the bitcoind server to rc2.  Sounds like all that downtime skewed my numbers.  Understanding p2pol.info is not near real-time was what got me.  BTW, I was just checking back into p2pool.info and it is reporting numbers a bit higher than my current miner hashrate.