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

[gnar1ta$]

I'm dumb. How do you find and submit p2pool difficulty x>1 share chain shares if you set the difficulty to 1 

[DeathAndTaxes]

I'm dumb. How do you find and submit p2pool difficulty x>1 share chain shares if you set the difficulty to 1 

Because >1 is larger than 1 so if miners return all diff 1+ shares by definition that includes p2pool difficulty shares. 

The same way a conventional pool sets difficulty to 1 yet when you solve a block they get that too.

[gnar1ta$]

I see so your just setting the level the pool accepts shares not the level the miner is mining shares. So it won't affect my payout to set p2pool difficulty to 1 to run some comparisons?

[kano]

I see so your just setting the level the pool accepts shares not the level the miner is mining shares. So it won't affect my payout to set p2pool difficulty to 1 to run some comparisons?

Except that I wonder what the point of the comparison would be?

Do 1-difficulty shares on P2Pool get found as often as 1-difficulty shares on normal pools?
There's only 2 possible answers:
1) Yes
2) No coz the 10 second LP reduces the count and p2pool loses that comparison
That magical: everyone should be happy with losing 9% because everyone else does ... except everyone else doesn't.
You actually shouldn't be happy with 9% coz if you use a good miner it isn't necessary

[gnar1ta$]

Kano it's not to compare to other pools but to compare with different GPU settings.  Not useful to compare stats on 2 48 hr runs when the share chain difficultly is changing during the 48 hrs (I think that's what DAT was saying). It should still be useful to compare U of several of the same cards at different settings though - assuming auto-gpu isn't throttling them...someone please turn down the sun in FL so I can get some good numbers

[kano]

Kano it's not to compare to other pools but to compare with different GPU settings.  Not useful to compare stats on 2 48 hr runs when the share chain difficultly is changing during the 48 hrs (I think that's what DAT was saying). It should still be useful to compare U of several of the same cards at different settings though - assuming auto-gpu isn't throttling them...someone please turn down the sun in FL so I can get some good numbers

As long as your miner is reporting MH/s accurately, the U: value isn't worth taking note of.
U: is random.
Same card same time same temperature same cgminer settings but: different block headers will give different values - the shorter the test time, the bigger the variance ... just like block mining

Of course rejects are worth noticing but that's a different issue ...

[DeathAndTaxes]

Kano it's not to compare to other pools but to compare with different GPU settings.  Not useful to compare stats on 2 48 hr runs when the share chain difficultly is changing during the 48 hrs (I think that's what DAT was saying). It should still be useful to compare U of several of the same cards at different settings though - assuming auto-gpu isn't throttling them...someone please turn down the sun in FL so I can get some good numbers

As long as your miner is reporting MH/s accurately, the U: value isn't worth taking note of.
U: is random.
Same card same time same temperature same cgminer settings but: different block headers will give different values - the shorter the test time, the bigger the variance ... just like block mining

Of course rejects are worth noticing but that's a different issue ...

I would just like to double check.  Got some strange results and it likely is due to the variable share difficulty but I just want to nail it down.  Had two rigs started roughly the same time (within 30 sec).  Involved over 11200 difficulty 9 shares so ~100K difficulty 1 shares.  First rig had higher MH/S but second rig had higher A.  They had different worksize. 

Now 99% sure it is due to variable share difficulty but they should have been adjusting to new share difficulty at the same time and thus it shouldn't affect stats.

The larger view is I see absolutely no value in variable share difficulty at local p2pool level.  It simply adds more variance into the miner level stats making it take 10x longer to get a good sample size.  It accomplishes nothing and makes apples to apples comparison tough.

[kano]

Kano it's not to compare to other pools but to compare with different GPU settings.  Not useful to compare stats on 2 48 hr runs when the share chain difficultly is changing during the 48 hrs (I think that's what DAT was saying). It should still be useful to compare U of several of the same cards at different settings though - assuming auto-gpu isn't throttling them...someone please turn down the sun in FL so I can get some good numbers

As long as your miner is reporting MH/s accurately, the U: value isn't worth taking note of.
U: is random.
Same card same time same temperature same cgminer settings but: different block headers will give different values - the shorter the test time, the bigger the variance ... just like block mining

Of course rejects are worth noticing but that's a different issue ...

I would just like to double check.  Got some strange results and it likely is due to the variable share difficulty but I just want to nail it down.  Had two rigs started roughly the same time (within 30 sec).  Involved over 11200 difficulty 9 shares so ~100K difficulty 1 shares.  First rig had higher MH/S but second rig had higher A.  They had different worksize. 

Now 99% sure it is due to variable share difficulty but they should have been adjusting to new share difficulty at the same time and thus it shouldn't affect stats.

...

Regarding A: - well they are still hashing different block headers even at the same time - so A: will of course differ between the two.

The time field will vary, the coinbase may vary (not sure though since p2pool creates the coinbase, not bitcoind) when a new txn comes in even the merkle root will vary between 2 getworks one before and one after, and same for the coinbase txn each time a new share comes in (every 10 seconds)

Think of it this way - if they were actually hashing the same block header, they'd find the same pseudo shares and thus you'd be wasting double the power on the same result.

Though after 100K diff 1 shares you'd expect them to be close ... if the hash rates were the same ... but get someone like Meni to do the actual stats calculation of the variance
(and of course two cards exactly the same can get different hash rates as is obvious)

As for pseudo-share difficulty - I'm not sure how and when that changes - though that should be that python code above that JayCoin pointed out.

[twmz]

Someone made a donation to the address I list on p2pool.info.  Over the next few days, I will be sending it to the p2pool miners using the patron_sendmany algorithm.  

Here is the list of transactions:

2/28/12 - http://blockchain.info/tx/7f5658c8cd2868cfc87e1ed921be74b80c06b0cb8c60b98fa709b0a23d971c31
2/29/12 - http://blockchain.info/tx/02cfe0ee7caa9342127b6c2255e22033c386bc9863da9e837e45c0e3bb7ddbca

[thirdlight]

Someone made a donation to the address I list on p2pool.info.  Over the next few days, I will be sending it to the p2pool miners using the patron_sendmany algorithm.

That's very generous of you. p2pool.info is a fantastic resource, you deserve the donations received!

[twmz]

Someone made a donation to the address I list on p2pool.info.  Over the next few days, I will be sending it to the p2pool miners using the patron_sendmany algorithm.

That's very generous of you. p2pool.info is a fantastic resource, you deserve the donations received!

Well, the p2pool.info page pretty clearly said that any donations to that address would be redistributed in this way, so it's not that generous (since the person who made the original donation expected this to happen).

In case people want to donate to p2pool.info directly, I'll add a separate address for that on the page.

[DeathAndTaxes]

In case people want to donate to p2pool.info directly, I'll add a separate address for that on the page.

That is a good idea.  I like the idea of donating to the trifecta
forrestv
p2pool.info
p2pool miners

It's as easy as 1-2-3.  All 3 play an important part in keeping Bitcoin decentralized and open.

[martychubbs]

Yet, I am not sold on hopping with P2P cause I have little hashes (2 GH) and I hop out before completed P2P share and thus get no credit?  I might not understand how it all works.

Hoppers like control and cutting edge tech, P2pool is all that but I don't know how to compare P2p shared against 1/difficulty that I use for non-p2pools.  And if I stop mining for 30-90 minutes, do I pickup on the P2p share where I left off or do I have to start a new share?  Anyone clear up my confusion?

[rjk]

Yet, I am not sold on hopping with P2P cause I have little hashes (2 GH) and I hop out before completed P2P share and thus get no credit?  I might not understand how it all works.

Hoppers like control and cutting edge tech, P2pool is all that but I don't know how to compare P2p shared against 1/difficulty that I use for non-p2pools.  And if I stop mining for 30-90 minutes, do I pickup on the P2p share where I left off or do I have to start a new share?  Anyone clear up my confusion?

It's in the title. Hop-proof pool.
You can't hop it.

[DeathAndTaxes]

Yet, I am not sold on hopping with P2P cause I have little hashes (2 GH) and I hop out before completed P2P share and thus get no credit?  I might not understand how it all works.

Hoppers like control and cutting edge tech, P2pool is all that but I don't know how to compare P2p shared against 1/difficulty that I use for non-p2pools.  And if I stop mining for 30-90 minutes, do I pickup on the P2p share where I left off or do I have to start a new share?  Anyone clear up my confusion?

There is no such thing as "progress" toward any share or block.  In the long run you will earn shares at the same rate.  You might hop in provide half the work required ON AVERAGE for a share and get a share (or two, or ten), and you may also hop in and get no shares.

In the LONG RUN yuor # of shares thus your compensation will be the same regardless of if you mine intermitently, intentionally hop in/out, or mine continually 24/7/365.

You can't work toward a share (any share in any pool) because there is no progress.  You either find a share or you don't.  Each hash is completely independent and the prior "chances" have no effect on the odds of finding a hash. 
There is no penalty to hopping PPLNS pools and p2p is no exception. 

It is "hop proof" in that you can never get greater than 100% PPS expected value.  You can't steal revenue from other miners (like you do in the Prop pools you hop) as each miner is paid fairly for the work they complete.

[Diapolo]

This project rocks, I switched from a normal pool and use this now for everyday mining (even if I only can add ~600 MH/s). Keep up the great work!

Dia

[echris1]

Yeah, we seem to be having a bit of trouble topping that 300 mark for any amount of time.  I just added another card, but 350MH more doesn't move the needle too much =)

I'm with you Holliday, we should be much bigger, keep getting the word out!

[twmz]

Several people asked, so I enabled p2pool sig and avatar image generation for p2pool miners at btcstats.net.  Just go to http://btcstats.net?p2pool and enter your payment address to get your sig or avatar image URL.

Please note, that the hashrate shown is highly imprecise (just like the one at p2pool.info -- its really the same data source). The hashrate show is an estimation based on the number of high-difficulty shares you have found in the past 24 hours.  For most miners, variance will cause the estimated hashrate to vary significantly.

Also note that you can't use the sig image here because bitcointalk.org has disabled adding images to sigs.  You can, however, use the avatar image here.

[Jointops420]

Great idea and loving it. Added my 700mh theres really no need for big pools anymore, getting small shares but not running all of the time just through the day. When I got up this morn and connected some shares from yesterday came through to my wallet.

[bitcool]

Just found this gem, added 9GH, it will be fun