[1200 TH] EMC: 0 Fee DGM. Anonymous PPS. US & EU servers. No Registration!

[Inaba]

That quick block made up for the last one, almost! 

I have 5 cards down at the moment, bah... So painful on short blocks.

[1713918931]

1713918931

[1713918931]

1713918931

[Reuef]

Looks like that block has collided with one from deepbit....which one will be invalid?

[1984]

Nope, no reward glitch as far as I can tell.  I had some miners go down on my rigs in the past 18 hours or so, so several percentage of the payout went back into the pool that I would normally get at my hashrate.  Scoring in action

1984 - how much do you think you should have gotten?  As in, how low is it compared to what you were expecting?

Our hashrate was down to 50 GH/s or so from 80+, so those people that dropped out should have boosted everyone elses reward compared to prop.

Should of added this -758.16% Proportional Difference ouch. I guess my main miner stopped working just before the block was found, not logging anymore so no way of telling, poos.

[Inaba]

Well, it looks like we got the crap end of the stick on that.  At least it wasn't a monster block.

It also looks like my orphan code doesn't work properly!  Doh! 

I will look at that tonight.

[FreeJAC]

Damn that sucked~! Invalid block! I've never seen that before, how often does that happen?

[Inaba]

Well, traditionally about 3% are invalid blocks.  I've tried to minimize that and get it lower than that.  So far, it seems to be good, since we would be running a little under 2%.

Invalid/orphan code is fixed, though.

[bacon strips]

Since we found this new block (block # 56) there was a rather quite jump in harsh rate from about 50 to 80 GH/s.  Which means there are a lot of pool hoppers.  Inaba I hope you keep any proportional or PPS method separate from the current pool because of pool hopes. 

[Inaba]

The scored pool is separate from anything else, so it's not buying the pool hoppers anything.  I'm not sure what they are hoping for, honestly.

[organofcorti]

The scored pool is separate from anything else, so it's not buying the pool hoppers anything.  I'm not sure what they are hoping for, honestly.

I'm not hopping here, for old time's sake, but if you have a proportional part of the pool, people *will* hop it. I noticed that the option is scored/prop and read on your forums that to get the prop option you need to use a different port, so I'm going to go ahead and assume you're being hopped. This means the people who begged you for proportional are back to where they were without scoring.

People, if you want something fair for part time or inconsistent miners, use Inaba's PPS. If you're a constant miner, use geometric scoring. If you use prop and you don't hop, you will earn less if the round is hopped by others. I can't put it more plainly than that.

[kripz]

People, if you want something fair for part time or inconsistent miners, use Inaba's PPS

No, it's PPLNS (unless he decided to change it), you effectively cannot leave so you might as well use the scoring pool.

[organofcorti]

People, if you want something fair for part time or inconsistent miners, use Inaba's PPS

No, it's PPLNS (unless he decided to change it), you effectively cannot leave so you might as well use the scoring pool.

I thought I read something about plans for PPS somewhere? No? Fair enough. PPLNS and score both have an expectation of 1.0 efficiency long term, but I don't know which has less variance. Do you have any extra info on that point? I'd be interested to see what you have.

[kripz]

I thought I read something about plans for PPS somewhere? No? Fair enough.

https://bitcointalk.org/index.php?topic=16385.msg480477#msg480477

PPLNS and score both have an expectation of 1.0 efficiency long term, but I don't know which has less variance. Do you have any extra info on that point? I'd be interested to see what you have.

No i dont have any extra info, but there was a graph or pdf study i read/saw before, and im pretty sure i saw you posting some stuff already so you probably already seen it. I think they will have 1 efficiency as well.

I'm not sure if you understood me but my point was that with PPLNS you cannot "inconsistenly mine" and effectively have to treat it like the scoring method so in the end, you may as well use the scoring method (and inaba should just scrap the pplns pool entirely imo).

[organofcorti]

I'm not sure if you understood me but my point was that with PPLNS you cannot "inconsistenly mine" and effectively have to treat it like the scoring method so in the end, you may as well use the scoring method (and inaba should just scrap the pplns pool entirely imo).

Thanks for the link! That where I read it. Thought I was going mad. Teach me not to read posts properly.

As far as score v PPLNS goes I understood what you meant, but it's the variance difference that would make you decide one way or another. Part time miners are always hit more by variance than full timers, and if one method averages out sooner than another then that's the one you'll want to use. It might be as you suggested (score better than PPLNS in terms of variance) but then there's lots about mining that isn't intuitive.

[Meni Rosenfeld]

Geometric has higher variance than PPLNS. This is its well-known weakness, and is why I now recommend that people use either double geometric or PPLNS.

I'm not sure if you understood me but my point was that with PPLNS you cannot "inconsistenly mine" and effectively have to treat it like the scoring method so in the end, you may as well use the scoring method (and inaba should just scrap the pplns pool entirely imo).

This is a myth. You most certainly can mine inconsistently, and your payout will be on average just like with PPS/solo. Your variance will be higher, but PPLNS doesn't have a lot of variance so you won't have a problem.

[ciuciu]

Hey,
I test drive your pool again, and I contributed 269435 shares out of 2824488 for the last block. My payout was only 0.00735799. I don't know what method you use, but for sure you are not making any friends with it.
Basically, you stole me 4.76 bitcoins and I'm not happy at all.

[Meni Rosenfeld]

Hey,
I test drive your pool again, and I contributed 269435 shares out of 2824488 for the last block. My payout was only 0.00735799. I don't know what method you use, but for sure you are not making any friends with it.
Basically, you stole me 4.76 bitcoins and I'm not happy at all.

If you mined solo, found 269435 difficulty-1 hashes but no blocks, who would you blame for your lost 8 BTC?

Pools exist to alleviate the high variance of solo, but except for PPS they do not eliminate it completely. You just had bad luck.

Your calculation is only valid for the proportional method, but this method is known to be broken and is not the default for this pool.

(Of course, an implementation bug is always a possibility, even though we've checked it thoroughly. If you supply more details we may be able to double-check this.)

[organofcorti]

Hey,
I test drive your pool again, and I contributed 269435 shares out of 2824488 for the last block. My payout was only 0.00735799. I don't know what method you use, but for sure you are not making any friends with it.
Basically, you stole me 4.76 bitcoins and I'm not happy at all.

Were you mining on score or PPLNS? Also what's the hashrate of your miner? Finally, did you mine constantly or intermittently? TIA

[Inaba]

Ok, looks like I need to clear up some confusion.

The pool is Meni's CPS geometric scoring system.  All the stats and everything you see displayed are for the scored pool.  I constantly check the block distributions when they are solved and they have been accurate, both for constant miners and for non-constant miners (I have miners go down regularly and their share of the block reward reduces accordingly). 

There is a proportional pool, on a different port, that is under testing right now (though it is on the like block chain).  Due to time constraints, I have not advanced the proportional pool beyond that stage, and the only "stats" display there is for it is a tab display on the Block Stats page.  Other than that, there is no web feedback for the proportional pool.  The proportional pool has yet to solve a block (the hash rate is reallllly low, which is fine and expected.)  The proportional pool is completely separate from the scored pool, and thus the scored pool is not affected by what happens on the prop pool.

As stated in previous posts, the setting to change your miner type is not active at the moment.  So changing it wont' do anything.  Hopping the pool won't do anything, though I don't mind if it's hopped, since it doesn't really benefit anyone in particular. 


PPS, LNPPS, Some other method, has not been decided upon (if any) yet.  Mostly due to a lack of good alternative and integration issues.  So there is no option for PPS or LNPPS.

Ciuciu - Your user name is not Ciuciu on the pool, so I can't look up your shares until I know what it is to verify they are correct.  If you want to PM me your name or post it here, I can look them up and see when your last share was submitted.

[ciuciu]

The user name is ursu.

[Inaba]

I show you submitted exactly 59 shares in the 5 hours preceding 2011-09-05 02:03:45, when the block was found:

select count(id) from shares_history where username like "ursu%" and blocknumber = 56 and autotime > "2011-09-04 21:00:00"
count(id) - 59

If I expand that to twelve hours: 

select count(id) from shares_history where username like "ursu%" and blocknumber = 56 and autotime > "2011-09-04 14:00:00"
count(id) - 135

12 hours is well outside the window of decayed shares and there was only 135 submitted in that time frame, 59 is barely in the window (and some will fall outside the window).