Author Topic: Important: p2pool LTC miner recommendation  (Read 9668 times)

Offline warren

  • Litecoin Dev Team
  • Sr. Member
  • **
  • Posts: 827
  • Linux Software Engineer
Important: p2pool LTC miner recommendation
« on: May 28, 2013, 12:40:24 PM »
Update: p2pool-13 has made it possible for miners as small as 1MH to mine with p2pool.  If you are a smaller miner you are still recommended to mine elsewhere as you will be frustrated by the high difficulty and variance of shares.

Quote
Hi folks,

TLDR - if you are a small miner below 3MH/s you probably want to switch to a normal pool.

If you are a "big" p2pool miner with ~2MH/s or more you probably don't need to concern yourself with this.  Otherwise small p2pool miners need to read this important message.  As you can see from the official fundraising thread, I am the lead developer on Litecoin-0.8.2.  I am also a p2pool developer.  Two months ago I fixed a serious problem in p2pool's scrypt stratum that was previously causing p2pool LTC to lose some work.  So it suffices to say I know a lot about Litecoin and p2pool.

Problem with p2pool for small LTC Miners
http://ltc.block-explorer.com/block/f969ae9e8e5eaf5563c0a07ca3fc5d7ea3b49192179136e2b9793a58836e157b
Here is a typical p2pool LTC block where there are often 500-600 payouts.  This is fine for the bigger miners who receive like 0.1 LTC per block, but it becomes increasingly problematic for the smaller miners as payout amounts per block become tinier and tinier.  As many of you know, spending the tiny p2pool dust can be very expensive in fees, or annoyingly time consuming if you use the free tx combining method.

Bigger chunks, paid less frequently?
I talked with forrestv about this dust issue, as the quantity of dust and the increasing difficulty of spending it is growing far worse.  We agreed in principle today that future versions of p2pool must enforce a minimum dust payout size.  What this means is if you are a small hashrate miner, you will have less frequent payouts in larger sized chunks.  This can be good for you, as larger chunks are easier to spend with lower fees.  This can also be bad because it will seem frustrating to have many hours of zero payouts and other hours of large chunks.  Even if you logically know the expected payout is the same, humans typically don't psychologically deal with variance all that well.

So what are your recommended options today?

Option 1: Increase your minimum share difficulty.
Currently p2pool's variable share difficulty is floating around 1.47.  If you multiply that by 2^16 you get 96337.92, the equivalent share difficulty that you see in cgminer.  If you hit that share, you collect ~0.01 LTC per block for a few hours.  Small miners will frequently get annoyingly small dust in tiny increments like 0.01-0.05 which are expensive and inconvenient to deal with.

If you edit your miner config to append /<number> to the end of your miner username, you can force the required share difficulty to be higher.  If you set it to /2, then 2*2^16 = 131072 is the share difficulty.  This means you will get shares less frequently, but when you do have shares, subsequent payouts will be in larger chunks.  /4 would mean shares half as often as /2, with double the minimum size chunks.  The /<number> can be a maximum of 10 times the floating share difficulty, so if 1.47 is the current floating share difficulty, the highest you can set it is /14.7.

The trouble is ... if you are a small miner, you probably need something like /6 or higher to have minimum dust sizes that approaches the level where they are easier to spend.  If you use a mining calculator plugging in your hash rate of 427KH/s and a difficulty of 6, it might tell you an expected average of 16 hours to find a share of difficulty 6.  But that is only an average.  In reality shares will happen within a bell curve of probability centered at 16 hours.  You might have two days with no shares.  You might have lucky days with three shares.  These are all normal and expected, and quite frustrating.

A future version of p2pool will enforce higher share difficulty and/or less frequent payouts for small miners, basically making this recommendation automatic.  But you can see from the above that very high variance is scary in general to small miners.

Option 2: Switch to a Normal Pool
As most humans don't understand probability and cannot psychologically handle extreme variance, most small miners are better off using traditional pools.

Unfortunately, the cost to the network of hundreds of dust payouts and the cost to miners in high fees makes p2pool a very bad option for small miners.  It hurts me to admit this, as decentralized trustless mining with p2pool is supposed to be good for the security of the network.  Fortunately, today's Litecoin mining pools are spread out with no pool exceeding 20% of the total network hash rate.  Leading pools like give-me-ltc.com or wemineltc.com have committed to discouraging more miners if they are in danger of approaching 50%.

So it is my recommendation that most smaller miners use the traditional pools.  It is better for the network as long as those pools don't grow too large.
« Last Edit: August 22, 2013, 10:18:34 AM by warren »
If you appreciate our work, please support the Litecoin Development Team with a small contribution.
Follow the latest dev news on Litecoin Project Google+
GPG: AEC1884398647C47413C1C3FB1179EB7347DC10D

Offline Chuck Finley

  • Sr. Member
  • ****
  • Posts: 612
Re: Important: p2pool LTC miner recommendation
« Reply #1 on: May 28, 2013, 01:00:10 PM »
Thank you once again for an informative post!

I definitely do not mean for this to sound rude but if you guys are addressing the "dust/fee" issue in the upcoming 0.8 release why is this now such an important matter?

I'm just guessing here but I would think these guys making the smallest "dust" wouldn't be looking to turn around and sell the 0.01 LTC they spent a few days mining but would instead save for a lot longer (so hopefully by the time they have a couple of LTC the new version will be out!)

I'd just hate to see people scared away from p2pool since I think the fact that payments go right to your wallet (so you don't get pools like notroll etc. where all of a sudden you can't get your coins anymore) as well as the decentralized nature (no 51% attack and the ability to fall back to another node in the same pool in case of DDOS, HW failure, etc.) are very important features that no  other pool can really match.

Thank you again for all your hard work for the LTC community :)
LTC: LchuckfK3g2tHrdAv8CDTUSJpZjbBSk66m

Offline warren

  • Litecoin Dev Team
  • Sr. Member
  • **
  • Posts: 827
  • Linux Software Engineer
Re: Important: p2pool LTC miner recommendation
« Reply #2 on: May 28, 2013, 01:43:00 PM »
Fees will go down in 0.8.2 but it will still be high compared to this dust.  It will be months before we figure out the design for Smart Fee that will allow lower fees without compromising Litecoin's tough anti-spam stance.
If you appreciate our work, please support the Litecoin Development Team with a small contribution.
Follow the latest dev news on Litecoin Project Google+
GPG: AEC1884398647C47413C1C3FB1179EB7347DC10D

Offline FoBoT

  • Newbie
  • *
  • Posts: 15
    • FoBoT
Re: Important: p2pool LTC miner recommendation
« Reply #3 on: May 28, 2013, 05:20:46 PM »
thank you for this concise answer, i have been reading many posts over the past couple weeks on small miners, ltc dust and fees and this summary is very helpful

rock on

Offline warren

  • Litecoin Dev Team
  • Sr. Member
  • **
  • Posts: 827
  • Linux Software Engineer
Re: Important: p2pool LTC miner recommendation
« Reply #4 on: June 17, 2013, 07:05:30 AM »
forrestv decided on the following details for the next hardfork version of p2pool LTC:

  • 20 second interval instead of 10 seconds. This will allow the p2pool network to be more competitive against traditional pools because currently people decrease Intensity due to fear of DOA.  Miners will be able to use higher intensity and the pool as a whole would have more hashrate with the same hardware.  20 seconds also reduces the frequency of orphaned shares which are also hated by p2pool miners.
  • Payouts will be averaged across two blocks instead of the current 12 blocks.  This means valid shares will be at least six times larger in payout size, but you will have fewer payouts.  Larger and fewer payouts is healthier for the network and better for miners because they have lower fees to spend the dust.
  • Dynamic minimum share difficulty to enforce minimum dust size of 0.05 LTC.  This means if you don't already have a current share, p2pool will increase the minimum difficulty for your first share to ensure that you will never get painfully small dust.

All of this makes p2pool healthier for the network and less costly for miners.  Small miners are strongly advised to either accept extremely high variance (infrequent but large chunks of coins) or just switch to a traditional pool.

forrestv will need veteran p2pool testers soon to test the new code.  Please watch this thread.
If you appreciate our work, please support the Litecoin Development Team with a small contribution.
Follow the latest dev news on Litecoin Project Google+
GPG: AEC1884398647C47413C1C3FB1179EB7347DC10D

Offline General-Beck

  • Litecoin Association Member
  • Jr. Member
  • ***
  • Posts: 159
  • LTC p2pool node
    • LTC p2pool
Re: Important: p2pool LTC miner recommendation
« Reply #5 on: June 17, 2013, 11:37:23 AM »
good news
UPD if I correctly understood, in 12 versions it is possible not to use a divider after the address of the user?

Offline debianlinux

  • Full Member
  • ***
  • Posts: 298
  • Freow
Re: Important: p2pool LTC miner recommendation
« Reply #6 on: June 17, 2013, 12:50:27 PM »
Small miners are strongly advised to either accept extremely high variance (infrequent but large chunks of coins) or just switch to a traditional pool.

Define small?

Offline Fallout

  • Full Member
  • ***
  • Posts: 483
Re: Important: p2pool LTC miner recommendation
« Reply #7 on: June 17, 2013, 12:58:49 PM »
Read the first lines of the post ... Small = less than 2MH/s
BTC: 18b4FuDzMJnaaZK63tm5vLULUGd3YwSLu
LTC: LRr9CTjQxUo4nvW5UMWz2jMwBHJAKLvreR
FTC: 6hMioPvuzexjcNWJHuhqEySTzF3TiWLUBp
BBQ: bQRPD59v1aqvQAM7knuo7CG9QtiamnUjZi

Offline gomeler

  • Jr. Member
  • **
  • Posts: 107
Re: Important: p2pool LTC miner recommendation
« Reply #8 on: June 17, 2013, 09:40:54 PM »
I wrote a script to form raw transactions and I've managed to pay 0% fees when consolidating my p2pool wallet. That's a third option for small users who are willing to wait 1-2 week for coins to mature and are ok with maintaining a balance of aged coins in their p2pool wallet. Inconvenient but valid.

Offline Carnth

  • Administrator
  • Elite Member
  • *****
  • Litecoin Association Member
  • ***
  • Posts: 2307
  • Litecoin Association
Re: Important: p2pool LTC miner recommendation
« Reply #9 on: June 18, 2013, 02:00:37 PM »
I would be willing to test on one of my public nodes.

Offline General-Beck

  • Litecoin Association Member
  • Jr. Member
  • ***
  • Posts: 159
  • LTC p2pool node
    • LTC p2pool
Re: Important: p2pool LTC miner recommendation
« Reply #10 on: June 18, 2013, 06:49:01 PM »
forrestv decided on the following details for the next hardfork version of p2pool LTC:

  • 20 second interval instead of 10 seconds. This will allow the p2pool network to be more competitive against traditional pools because currently people decrease Intensity due to fear of DOA.  Miners will be able to use higher intensity and the pool as a whole would have more hashrate with the same hardware.  20 seconds also reduces the frequency of orphaned shares which are also hated by p2pool miners.
  • Payouts will be averaged across two blocks instead of the current 12 blocks.  This means valid shares will be at least six times larger in payout size, but you will have fewer payouts.  Larger and fewer payouts is healthier for the network and better for miners because they have lower fees to spend the dust.
  • Dynamic minimum share difficulty to enforce minimum dust size of 0.05 LTC.  This means if you don't already have a current share, p2pool will increase the minimum difficulty for your first share to ensure that you will never get painfully small dust.

All of this makes p2pool healthier for the network and less costly for miners.  Small miners are strongly advised to either accept extremely high variance (infrequent but large chunks of coins) or just switch to a traditional pool.

forrestv will need veteran p2pool testers soon to test the new code.  Please watch this thread.
I started on my node litecoind 0.6.4rc1 with p2pool patched 'litecoindust', welcomme testers!

Offline warren

  • Litecoin Dev Team
  • Sr. Member
  • **
  • Posts: 827
  • Linux Software Engineer
Re: Important: p2pool LTC miner recommendation
« Reply #11 on: June 18, 2013, 08:28:23 PM »
I started on my node litecoind 0.6.4rc1 with p2pool patched 'litecoindust', welcomme testers!

NOOOO!! DO NOT RUN THAT!  That is unfinished code from weeks ago.
If you appreciate our work, please support the Litecoin Development Team with a small contribution.
Follow the latest dev news on Litecoin Project Google+
GPG: AEC1884398647C47413C1C3FB1179EB7347DC10D

Offline DapperDon

  • Jr. Member
  • **
  • Posts: 66
Re: Important: p2pool LTC miner recommendation
« Reply #12 on: June 19, 2013, 12:56:29 AM »
Any miner should do the trick

Offline General-Beck

  • Litecoin Association Member
  • Jr. Member
  • ***
  • Posts: 159
  • LTC p2pool node
    • LTC p2pool
Re: Important: p2pool LTC miner recommendation
« Reply #13 on: June 19, 2013, 04:17:16 AM »
I started on my node litecoind 0.6.4rc1 with p2pool patched 'litecoindust', welcomme testers!

NOOOO!! DO NOT RUN THAT!  That is unfinished code from weeks ago.
Asked to test, we test) Someone shall test a code on real loading.  I delivered warnings. And i return to 0.6.3 & p2pool master
DOA index magnificent.

Offline freeb

  • Full Member
  • ***
  • Posts: 272
Re: Important: p2pool LTC miner recommendation
« Reply #14 on: December 19, 2013, 12:00:34 AM »
forrestv decided on the following details for the next hardfork version of p2pool LTC:

  • 20 second interval instead of 10 seconds. This will allow the p2pool network to be more competitive against traditional pools because currently people decrease Intensity due to fear of DOA.  Miners will be able to use higher intensity and the pool as a whole would have more hashrate with the same hardware.  20 seconds also reduces the frequency of orphaned shares which are also hated by p2pool miners.
  • Payouts will be averaged across two blocks instead of the current 12 blocks.  This means valid shares will be at least six times larger in payout size, but you will have fewer payouts.  Larger and fewer payouts is healthier for the network and better for miners because they have lower fees to spend the dust.
  • Dynamic minimum share difficulty to enforce minimum dust size of 0.05 LTC.  This means if you don't already have a current share, p2pool will increase the minimum difficulty for your first share to ensure that you will never get painfully small dust.

All of this makes p2pool healthier for the network and less costly for miners.  Small miners are strongly advised to either accept extremely high variance (infrequent but large chunks of coins) or just switch to a traditional pool.

forrestv will need veteran p2pool testers soon to test the new code.  Please watch this thread.

Sorry to revive an old thread but is any of this still in the works? Especially bullet points one and two.