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?
Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 764. (Read 2591916 times)
February 28, 2012, 11:09:11 AM
February 28, 2012, 08:31:15 AM
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.
February 28, 2012, 08:21:59 AM
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.
February 28, 2012, 07:39:06 AM
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! February 28, 2012, 07:22:11 AM
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
Here is the list of transactions:
2/28/12 - http://blockchain.info/tx/7f5658c8cd2868cfc87e1ed921be74b80c06b0cb8c60b98fa709b0a23d971c31
2/29/12 - http://blockchain.info/tx/02cfe0ee7caa9342127b6c2255e22033c386bc9863da9e837e45c0e3bb7ddbca
February 28, 2012, 04:50:32 AM
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 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.
February 28, 2012, 12:50:19 AM
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.
February 28, 2012, 12:39:08 AM
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 ...
February 27, 2012, 11:46:07 PM
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
February 27, 2012, 11:01:08 PM
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
February 27, 2012, 09:14:27 PM
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?
February 27, 2012, 08:51:21 PM
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.
February 27, 2012, 08:44:19 PM
I'm dumb. How do you find and submit p2pool difficulty x>1 share chain shares if you set the difficulty to 1
February 27, 2012, 07:48:39 PM
My pool works a little different from a public p2pool server. My pool tracks the difficulty 1 shares submitted and then distributes the p2pool reward proportionally. This is so your miner doesn't actually need to get a block on the p2pool share chain to get the reward (Great for low hashrate miners). Current fee is 0.5% with auto payouts after a balance greater than 0.5 BTC. See the thread https://bitcointalksearch.org/topic/offlinep2pminingcom-hybrid-p2pool-no-fee-btcnmcixci0cdevltc-66202
Ah, ok. 
February 27, 2012, 07:06:40 PM
My pool works a little different from a public p2pool server. My pool tracks the difficulty 1 shares submitted and then distributes the p2pool reward proportionally. This is so your miner doesn't actually need to get a block on the p2pool share chain to get the reward (Great for low hashrate miners). Current fee is 0.5% with auto payouts after a balance greater than 0.5 BTC. See the thread https://bitcointalksearch.org/topic/offlinep2pminingcom-hybrid-p2pool-no-fee-btcnmcixci0cdevltc-66202
February 27, 2012, 06:00:42 PM
On my system
Memory:
bitcoind uses about 120MB, p2pool uses about 70MB.
CPU:
CPU time is essentially 0% (except when downloading new chain until caught up).
Disk:
bitcoind ~1.3GB p2pool ~<100MB (truncating log would make it <30MB).
Memory:
bitcoind uses about 120MB, p2pool uses about 70MB.
CPU:
CPU time is essentially 0% (except when downloading new chain until caught up).
Disk:
bitcoind ~1.3GB p2pool ~<100MB (truncating log would make it <30MB).
Thanks
Probably not going to run very well on my router
February 27, 2012, 05:34:19 PM
Or you can run your miner at http://p2pmining.com
http://p2pmining.com:9332/fee = umm what? February 27, 2012, 05:18:40 PM
So between running Bitcoind + p2pool and whatever else is needed, what's the footprint in terms of processor/memory/disk space?
On my system
Memory:
bitcoind uses about 120MB, p2pool uses about 70MB.
CPU:
CPU time is essentially 0% (except when downloading new chain until caught up).
Disk:
bitcoind ~1.3GB p2pool ~<100MB (truncating log would make it <30MB).
February 27, 2012, 05:12:02 PM
So between running Bitcoind + p2pool and whatever else is needed, what's the footprint in terms of processor/memory/disk space?
February 27, 2012, 04:37:50 PM
If you comment out line 533-538 in the main.py file (newest version) it will serve only diffficulty 1 shares to your miner
Thanks I may have to do that. My p2pool rig is my windows workstation so I use windows binary version temporarily I could use python version but really I think this should be a user defined option.
Jump to: