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

[jonnybravo0311]

And

Code:

SPREAD = 3 # blocks

is not tell that max 3 blocks can be paid?
As far I understand, every share is valid 8640 shares of time but up to 3 blocks found in mean time.
So if we will found more than 1 block per day, and your power is not enough to get more than one share per 3 days you will not get payout from every block.

No, it is not a hard limit of 3 blocks, it is 3 times the average work required to solve a block.  In other words, if the average time to block is 8 hours, then your shares will fall off the payout after 24 hours.  So, if p2pool happens to get lucky and solve 10 blocks in that 24 hour period, your share will be paid for all 10 blocks.

[mdude77]

And

Code:

SPREAD = 3 # blocks

is not tell that max 3 blocks can be paid?
As far I understand, every share is valid 8640 shares of time but up to 3 blocks found in mean time.
So if we will found more than 1 block per day, and your power is not enough to get more than one share per 3 days you will not get payout from every block.

No, it is not a hard limit of 3 blocks, it is 3 times the average work required to solve a block.  In other words, if the average time to block is 8 hours, then your shares will fall off the payout after 24 hours.  So, if p2pool happens to get lucky and solve 10 blocks in that 24 hour period, your share will be paid for all 10 blocks.

That explains why it dropped from 3 days to 1 to 1.5 days.  I know very well I saw my sole share fall off within 36 hours.

It also means as the pool hashrate increases, this value gets smaller and smaller.  Say we get to 2hrs per block.  That means 6 hours, it rolls off.  Ouch.

Thanks.

M

[CartmanSPC]

It also means as the pool hashrate increases, this value gets smaller and smaller.  Say we get to 2hrs per block.  That means 6 hours, it rolls off.  Ouch.

We could always fork to increase the spread if necessary 

[nreal]

I'm having some questions about P2Pool mining, I'm mining on http://elizium.name node with 2 Antminer S3+'s, at about 920GH/s.
But I've been mining for 12+ hours now, and I still have nothing. I added after my payout address "+518" but still nothing.
Is this normal or...?

Because P2pool is PPLNS so it takes around 3 days to build up enough shares to start seeing payment. conversely when you stop mining you'll still receive payments for a few days. have a little patience, it'll come and when it does it'll be sweet. p2pool has been finding quite a few blocks lately.

I don't think the 3 day thing is true anymore.  I've watched my 1 share roll off in 36 hours or less.

Also, at 920 GH/s, your variance is going to be bad.  Be patient, and expect lots of time without shares.

M

This is covered in the "READ THIS FIRST" article here: http://p2pool.org/#learn-p2pool

But want to put it to bed once and for all...

The variable for determining how many shares to pay can be found here in the p2pool source code:
https://github.com/forrestv/p2pool/blob/master/p2pool/networks/bitcoin.py

This is an excerpt:

Code:

# CHAIN_LENGTH = number of shares back client keeps
# REAL_CHAIN_LENGTH = maximum number of shares back client uses to compute payout
# REAL_CHAIN_LENGTH must always be <= CHAIN_LENGTH
# REAL_CHAIN_LENGTH must be changed in sync with all other clients
# changes can be done by changing one, then the other

PARENT = networks.nets['bitcoin']
SHARE_PERIOD = 30 # seconds
CHAIN_LENGTH = 24*60*60//10 # shares
REAL_CHAIN_LENGTH = 24*60*60//10 # shares

REAL_CHAIN_LENGTH sets the total number of previously found shares to include in the payout when a block is found.

It is a fixed number, it is constant and does not change without the consensus of the p2pool network (i.e. the majority of miners on p2pool all change it on their nodes at the same time). It has not been changed since I have been around.

Currently it is 24*60*60//10...

(24*60*60)/10 = 8,640 shares are paid when a block is found, this is a mathematical fact and not subject to interpretation.

SHARE_PERIOD is the length in seconds expected before a share is found.

Unlike Bitcoin which adjusts difficulty about every 2 weeks, p2pool adjusts difficulty dynamically so a share is expected to be found about every SHARE_PERIOD seconds, currently set to 30 seconds.

The SHARE_PERIOD is a target that p2pool attempts to hit for every share by dynamically adjusting the difficulty.

Actual SHARE_PERIOD is subject to variance/luck, so it is possible a share could be found in 1 second, or in 120 seconds, however the difficulty is constantly adjusted so the expected time is always 30 seconds based on the current pool hashrate.

So if we know 8,640 shares will be paid, and that a share is expected to be found every 30 seconds...

8,640 * 30 = 259,200 seconds
259,200 / 60 = 4,320 minutes
4,320 / 60 = 72 hours
72 / 24 = 3 days

So....

An easy way to explain it so anyone can understand is:
A share is valid, and paid if a block is found, for about 3 days.

The accurate way to explain it is:
The last 8,640 shares are paid when a block is found, the expected time to share is 30 seconds.

Hope this helps...

Maybe you can try a miner, let it find one share then stop it. I can guarantee its not valid not much more than 24 hours!!!

Or if thats too difficult see with your eyes - http://p2pool.servebeer.com:9332/static/sharetime.jpg  the last share in graps wasnt valid for a single block....

[mdude77]

Maybe you can try a miner, let it find one share then stop it. I can guarantee its not valid not much more than 24 hours!!!

Or if thats too difficult see with your eyes - http://p2pool.servebeer.com:9332/static/sharetime.jpg  the last share in graps wasnt valid for a single block....

That's why I left p2pool.  I was squeezed out.  The one share I got rolled off within 36 hours before a block was found (it was one of those long streaks).

M

[IYFTech]

Hey guys,

I was thinking of renting some hash power for my node, so opened an account at miningrigrentals. I asked them via their ticket system if it was possible to make adjustments to the miner on the rented rig either via ssh or a web interface so as to adjust the -queue & diff settings, but they said this was not possible. Shame, as I don't want to rent a rig then watch 10-20% of the hash get wasted with DOA's etc.

I also inquired at Genesis Mining, but they don't allow for choosing which pool to point your hash at.

Does anyone know of a company that allows the renter to make these kind of adjustments and choices?

Thanks 

[windpath]

So....

An easy way to explain it so anyone can understand is:
A share is valid, and paid if a block is found, for about 3 days.

The accurate way to explain it is:
The last 8,640 shares are paid when a block is found, the expected time to share is 30 seconds.

Hope this helps...

And

Code:

SPREAD = 3 # blocks

is not tell that max 3 blocks can be paid?
As far I understand, every share is valid 8640 shares of time but up to 3 blocks found in mean time.
So if we will found more than 1 block per day, and your power is not enough to get more than one share per 3 days you will not get payout from every block.

Just when you think you have got it figured out...

Thanks Rav3n, I'll dig into it this week and see if I can figure out an easy way to explain what SPREAD is doing.

I know there has been some talk about rewriting p2pool in C++, any one have any thoughts on a Ruby version?

Have not given it a lot of thought, but after seeing Toshi thought there might be some advantages...

http://blog.coinbase.com/post/97671295752/introducing-toshi-an-open-source-bitcoin-node-for

[RoadStress]

Code:

0d 1h 6m 17s	2014-09-20 16:48:58	0 	0000000000000000095574410c8b9d067829bba651e3a8941fa60ee5b78bc20d	25.14698169	314	b78bc20d	0.01%	

Is it me or there is weird stuff going on? First the 2 orphaned blocks, now this...

[jonnybravo0311]

So....

An easy way to explain it so anyone can understand is:
A share is valid, and paid if a block is found, for about 3 days.

The accurate way to explain it is:
The last 8,640 shares are paid when a block is found, the expected time to share is 30 seconds.

Hope this helps...

And

Code:

SPREAD = 3 # blocks

is not tell that max 3 blocks can be paid?
As far I understand, every share is valid 8640 shares of time but up to 3 blocks found in mean time.
So if we will found more than 1 block per day, and your power is not enough to get more than one share per 3 days you will not get payout from every block.

Just when you think you have got it figured out...

Thanks Rav3n, I'll dig into it this week and see if I can figure out an easy way to explain what SPREAD is doing.

I know there has been some talk about rewriting p2pool in C++, any one have any thoughts on a Ruby version?

Have not given it a lot of thought, but after seeing Toshi thought there might be some advantages...

http://blog.coinbase.com/post/97671295752/introducing-toshi-an-open-source-bitcoin-node-for

Check out my reply to both you and rav3n... I explained it there.

[MissouriMiner]

请使用正确的难度。
P2POOL，这15天效率很低，
原因：中国矿工约有1.5PH/S，使用了超大的难度（+5120或者更多），
我们及时的更改了错误，目前全部使用大约是难度比例1:1。
P2POOL这些天低效率，我们表示抱歉。
但是我们认识到了最好的难度选择是：选择和矿机算力相等的难度。
当前我们的蚂蚁S3+518（+512，S3不显示BestShare），
1TH/S+1028，

S3+518，平均5秒一个Accepted。

S3+5120，大部分计算都超时，被服务器拒绝，能挖到的Share，但是很难产生块。

希望这些数据能帮助到更多的P2POOL矿工，

使用正确的难度选择。

来自中国的P2POOL爱好者。

百度在线翻译

Please use the correct difficulty.
P2POOL, this 15 days the efficiency is very low,
Reason: the Chinese miners about 1.5PH/S, using the large difficulty (+5120 or more),
We amend the wrong, all currently used is about difficulty ratio 1:1.
P2POOL these days and low efficiency, we apologize.
But we recognize the difficulty of selecting the best: the selection and calculation of force is equal to the difficulty of mining machine.
Our current ant S3+518 (+512, S3 does not display the BestShare),
1TH/S+1028,
S3+518, an average of 5 seconds for a Accepted.
S3+5120, most computing timeout, was rejected by the server, can dig into Share, but it is difficult to produce block.
Hope that these data will help to more P2POOL miners,
The use of the difficulty of choosing the right.
From Chinese P2POOL lovers.
Baidu online translation

Thank you for the valuable information.

[-ck]

Does anyone know if I change the COINBASE_NONCE_LENGTH (larger to make nonce2 larger) on a local p2pool whether shares will be rejected by the p2pool network? I'm trying to make it more suitable to using a ckproxy front end.

[windpath]

Does anyone know if I change the COINBASE_NONCE_LENGTH (larger to make nonce2 larger) on a local p2pool whether shares will be rejected by the p2pool network? I'm trying to make it more suitable to using a ckproxy front end.

Con, I do not know the answer to your question, however can you expand a little on ckproxy?

Is it directly integrated into ckpool, or can it stand it stand alone? Might it be part of a solution to variance for smaller miners?

Thx!

[-ck]

Does anyone know if I change the COINBASE_NONCE_LENGTH (larger to make nonce2 larger) on a local p2pool whether shares will be rejected by the p2pool network? I'm trying to make it more suitable to using a ckproxy front end.

Con, I do not know the answer to your question, however can you expand a little on ckproxy?

Is it directly integrated into ckpool, or can it stand it stand alone? Might it be part of a solution to variance for smaller miners?

Thx!

Yes.

Yes and yes and no.

It's a massive project. Please read the announcement:
https://bitcointalk.org/index.php?topic=790323.0

[windpath]

Does anyone know if I change the COINBASE_NONCE_LENGTH (larger to make nonce2 larger) on a local p2pool whether shares will be rejected by the p2pool network? I'm trying to make it more suitable to using a ckproxy front end.

Con, I do not know the answer to your question, however can you expand a little on ckproxy?

Is it directly integrated into ckpool, or can it stand it stand alone? Might it be part of a solution to variance for smaller miners?

Thx!

Yes.

Yes and yes and no.

It's a massive project. Please read the announcement:
https://bitcointalk.org/index.php?topic=790323.0

Looks amazing! Exciting times...

[kgb2mining]

Wow, was just reading that thread... first, congrats CK on the launch, it looks very nice.

Hmm... now if only there was a way to combine a properly written pool implementation that is written in a language that understands performance and multihreading, with a concept that peers together a number of nodes to form a de-centralized network of stand-alone miners large and small.... now that would be something......

[norgan]

Maybe you can try a miner, let it find one share then stop it. I can guarantee its not valid not much more than 24 hours!!!

Or if thats too difficult see with your eyes - http://p2pool.servebeer.com:9332/static/sharetime.jpg  the last share in graps wasnt valid for a single block....

That's why I left p2pool.  I was squeezed out.  The one share I got rolled off within 36 hours before a block was found (it was one of those long streaks).

M

I am starting to see what you mean. even 400gh/s is now considered small?! I'm hanging in there but really starting to loose patience with lack of block payouts. What to do?!

[mdude77]

Maybe you can try a miner, let it find one share then stop it. I can guarantee its not valid not much more than 24 hours!!!

Or if thats too difficult see with your eyes - http://p2pool.servebeer.com:9332/static/sharetime.jpg  the last share in graps wasnt valid for a single block....

That's why I left p2pool.  I was squeezed out.  The one share I got rolled off within 36 hours before a block was found (it was one of those long streaks).

M

I am starting to see what you mean. even 400gh/s is now considered small?! I'm hanging in there but really starting to loose patience with lack of block payouts. What to do?!

1 TH/s is small.

M

[-ck]

Hmm... now if only there was a way to combine a properly written pool implementation that is written in a language that understands performance and multihreading, with a concept that peers together a number of nodes to form a de-centralized network of stand-alone miners large and small.... now that would be something......

Any scalability issues with the python client are not as important as the marginalising problem so that's not really going to help. You can see the problems I described a long time ago as users are now discovering on this thread.

That's why I left p2pool.  I was squeezed out.  The one share I got rolled off within 36 hours before a block was found (it was one of those long streaks).

I am starting to see what you mean. even 400gh/s is now considered small?! I'm hanging in there but really starting to loose patience with lack of block payouts. What to do?!

Ironically the only current solution I can see is... putting regular pools in front of the distributed pool which of course adds centralisation detracting from the distributed nature, so it's hard to see how this is actually better than just using a full regular pool. I've yet to come up with a better solution to the problem, but that doesn't mean I've given up on looking for solutions entirely.

[norgan]

Hmm... now if only there was a way to combine a properly written pool implementation that is written in a language that understands performance and multihreading, with a concept that peers together a number of nodes to form a de-centralized network of stand-alone miners large and small.... now that would be something......

Any scalability issues with the python client are not as important as the marginalising problem so that's not really going to help. You can see the problems I described a long time ago as users are now discovering on this thread.

That's why I left p2pool.  I was squeezed out.  The one share I got rolled off within 36 hours before a block was found (it was one of those long streaks).

I am starting to see what you mean. even 400gh/s is now considered small?! I'm hanging in there but really starting to loose patience with lack of block payouts. What to do?!

Ironically the only current solution I can see is... putting regular pools in front of the distributed pool which of course adds centralisation detracting from the distributed nature, so it's hard to see how this is actually better than just using a full regular pool. I've yet to come up with a better solution to the problem, but that doesn't mean I've given up on looking for solutions entirely.

Well full respect to you Con. What do you think of Bitmains comments on p2pool, what do we expect to see from them? I wish I had the spare cash (and power prices) to go buy a few chunky miners but it's just a hobby for me. Your pool is private atm correct? some of the work you are doing looks great, does it help smaller miners?

[-ck]

Well full respect to you Con. What do you think of Bitmains comments on p2pool, what do we expect to see from them? I wish I had the spare cash (and power prices) to go buy a few chunky miners but it's just a hobby for me. Your pool is private atm correct? some of the work you are doing looks great, does it help smaller miners?

Bitmain's contribution to p2pool will not help regular miners. They see that being proponents of p2pool will raise their status in the eyes of potential customers. However it appears they simply want to run their own solo mining operation using their own p2pool instance which is really of no relevance to miners. Even if they dump their hashrate directly into regular p2pool, it just increases share difficulty even further, making it even harder for regular miners. The only way they may be contributing is by adding code/improvements to the p2pool network, but in my experience every hardware manufacturer directs the absolute minimum amount of effort towards software/drivers as possible to just expedite delivery of their hardware.

Offtopic about ckpool: There are public pools using the ckpool code - regular, solo and testnet. Just check the announce thread for the listed pools.