Topic: [4+ EH] Slush Pool (slushpool.com); Overt AsicBoost; World First Mining Pool - page 946. (Read 4382653 times)

With current pool hash rate Early shares submitted by a pool hopper will decline very quickly
every increase in total hash rate of the pool increases that rate of decline
The score system esentially makes what you submited at the beginning of a round worthless
even if you mine for the whole round

OK, it's only >estimated< reward; it's calculated from slightly outdated info (becuase of caching). I'll watch how rewards from this block will be finally distributed and if there will be something weird, I'll fix it. Thanks for pointing me to that.

Estimated rewards etc seem totally off though atm 

Everything seems to be ok. It's "only" 9.5mil shares, so nothing impossible. For example, deepbit had 10+ mil shares in one round today, too.

It looks like something is up.

My miners are all running properly though, so I suspect that it's Slush poking around with stats etc.

Hello everybody,

Is the pool ok?
As for me the current 9+ hour block looks a bit unlucky.
And the previous 4 hour too.

Another limitation of the function I give above is that it won't calculate any imaginary shares (math joke for the geeks out there)

For that you would need the full derivation:


Now you can check the score of non-real shares! I wonder what my imaginary gpu's hashrate is? Something pretty damn devastating, i expect.

I've spent way too much time analysing Slush's pool, and it is definitely my favourite with lots of easter eggs for the mathematically inclined. More importantly, the payouts over a week or so are exactly as I'd expect, the statistics are fantastic, and the website quick and intuitive. I have to agree with Slush that you really don't need to know your submitted shares and unless you have the exact time each was submitted you won't end up with the correct score anyway.


If you know your average hashrate and score for a round, then you can approximate the number of shares submitted using: where c=300, z is your score and c is your miner speed in shares per second. It will probably be +/- 5% out, but it should be close.



You probably won't trust me on my say-so, but I've analysed my scores and the pool scores on and off for a few months and my earnings match simulated earnings quite well. That said, the longer a round lasts the more the variance in payout you will observe - you're more likely to get much more or less than you'd expect for longer rounds.

One guy reported me that he lost some coins, because he didn't confirm new wallet fast enough while his reward already crossed over threshold. To prevent this stupid issue, pool won't process payouts for users who have pending wallet change in progress.

To enable payouts again, you need to cancel request or confirm new wallet.

Thanks, I didn't notice that it actually changed the code that was pasted in...

You have to go back and put the image back and when i hit 50 posts it took a few hours for the system to change my permissions.

hmm, still got "image removed"...

maybe I need 51 posts?

edit:  still there, hopefully there's just a delay...

YES!  50 posts at the same time User Bars are available.   

Yes, of course. Everything you need is to be the God :-).

I'm kidding. "Reverse pool hopping" does not work, because nobody knows when round will finish (in the oposite everybody knows when round begin, which is the problem with standard proportional method).

Does the score system mean that the pool is vulnerable to reverse pool hopping? Like only joining in on a block towards the end?

30 seconds on google would explain the scoring system and all the other payout systems

https://en.bitcoin.it/wiki/Comparison_of_mining_pools

Score - Score based system: a proportional reward, but weighed by time submitted. Each submitted share is worth more in the function of time t since start of current round. For each share score is updated by: score += exp(t/C). This makes later shares worth much more than earlier shares, thus the miner's score quickly diminishes when they stop mining on the pool. Rewards are calculated proportionally to scores (and not to shares). (at slush's pool C=300 seconds, and every hour scores are normalized)


so when you see it drop back to 200 or whatever its because the scores were normalized...

Actually I can do it. However people then start asking "when I get a round payout 0.03123 when I should get 0.03122?". I can show submitted shares, but it isn't used for calculating rewards, so it's mostly pointless. I'm always saying "one round does not matter". There is proof that it will average out after few thousands of shares (link taken from News on website). Everything what you need to follow is 7-day average on Graphs page.


Now you understand what I'm talking two lines above, right? :-).


If you're mining all time and don't have any connection issues which is lowering your round reward, then it's only tax for score method. Actually - nobody who understand how it works care about single round reward.


Everything is described in link above. This 'weird stuff' with score is calling 'renormalization'. Renormalization is running every hour to avoid calculations with huge numbers (you know that score is rising exponentially in time). But everytime you calculate (your score)/(pool total score), you'll get the same ratio reflecting your part on final round reward. No black magic here.


Profile will display stale share ratio soon, it's in ongoing pool update. However only your miner shows you stale ratio and connection issues with 100% accuracy.


About nmc stats - I simply don't have such precisious stats to display, as I explained before.


... which double database writes... World is pretty hard place to live :-). I understand your questions, but I have some limited resources (mostly time) and I'm choosing what can I done with them. However I agree that sometimes my choice don't fit preferences of pool users. I'll try to make stats better soon.

You've explained a lot, and I appreciate that.  I still don't see why you can't show a user's submitted shares for a found block on the stats page.  That part doesn't seem like it'd be that hard to do.  Somewhere you have to have that data because you use that date to calculate the score to determine the payout.  So why can't you add a column to the stats next to total shares and call it your shares so you can see how many shares you submitted for that block. 

The only reason I bring this up and question the scoring system btw is because I should have a steady payout when mining.  For the most part I should mine at around 900Mhash/s, and currently the pool's rate has been around 1100Ghash/s - 1200Ghash/s, so for each found block my payout should be roughly around the same amount.  Sometimes I'm making 0.030 BTC per block found, and sometimes I'm making as low as 0.006 per block found.  That's a large variance in payout.  And I can't help but look at the scoring system and question it.  Also, sometimes I view my account page and see my score will say something like 100000.00 + of a score.  And then a few hours later it'll be 200.00 or some number in the hundreds.  Which is why I really do not understand your scoring system.  And I am unaware of anywhere on the site where I can see my hash rate from the site's side, which could explain some of this.  Or even a stale share ratio, so I know how many stale shares I'm submitting. 

I'm not trying to criticize your work or your site.  You've done something I could never do on my own.  I use your site because it is the largest merged mining pool I have found.  But I see so many other sites that have merged mining all having the same basic stats available to the users.  I just don't see how some of these basic stats would be so difficult to display.  Just in my thinking, it would seem that you would need to have these stats anyway to calculate proper payout, so displaying these stats to the users would be as simple as storing them in another table to on database and re-writing some of the stat pages to display them. 

User bars - finally! Thanks twmz for hell fast support. Please visit btcstats.net to pick yours. I'll integrate userbars to website soon.

You answered this by yourself. It's not "the same", but it is proportional to number of found NMC blocks in the meantime of BTC block.


I explained this when I was introducing Merged mining; My pool was first one programmed back in December 2010. Back at this time, I didn't have an idea about alternative chains, so pool natively support only one blockchain. It is all about performance and scalability, I simply calculate shares and other stats once instead of twice (separate for BTC and NMC). Because of this, I decided to pick solution described above; Pool is collecting Namecoins and split collected NMCs between users using score from bitcoin rounds. It is perfectly fair mechanism, however I'm unable to provide fancy stats and detailed graphs about mining of namecoins. I'm planning to rewrite pool core for full support of more blockchains, but (as you will see in few days) there're more important stuff coming, so please be patient.


No. This is just side effect of merged stats - it *looks like* your NMC reward is also "invalid", but this state is related only for bitcoin block. As far as Namecoin blocks are valid, you'll get it.


It's same as with bitcoin mining - you need to trust me. Pools aren't cheat proof from side of operators and every operator can cheat pool members easily.

You cannot recalculate if your NMC reward is correct, because you don't know who subscribed for mining namecoins and who does not. However if I'll show on site "60.3% of hashpower are NMC miners" (or something like this), you still need to trust this number. Also in pool in completely open stats (where people can list who submitted how many shares when), cheating is still possible, because you need to trust the operator that those members are real.


This is a little technical, but: every time single share is submitted, it is compared not only against Bitcoin target, but also against Namecoin target. When submitted share has difficulty above Bitcoin difficulty, it leads to valid Bitcoin block. When share has difficulty above Namecoin difficulty, it leads to valid Namecoin block. And because NMC difficulty is currently lower than Bitcoin difficulty, everytime some miner find a valid Bitcoin block, it is automatically also Namecoin block.




No.


Yes, current Namecoin difficulty is around 156000, Bitcoin difficulty is 1460000. So there should be (1460k / 156k) ~ 9 Namecoins block inside one Bitcoin block in average.


Rome wasn't built in a day. I introduced merged mining as a first large pool, so I don't think it's so bad . As I explained above, collecting detailed stats need major rewrite of pool accounting internals which wasn't in a scope of recent merged mining project. And now I'm working on another important stuff for the pool.


Your hashrate is still the same, merged mining didn't affect it at all.


Actually score system is still pretty easy, I'm thinking about geometric system, which is hell more complicated. Score system is basically normal proportional method, only weight of every share is rising in time. For basic math you still can use share count instead of score and it will +/- fit expectations.


Actually you can. Those calculations are done on all pool history and it fits expectations. All calculations based on earnings of a single person (if you're comparing 'expected earnings' with 'real earnings') are subject of local pool luck, but that math done in linked thread shows that pool is as lucky as it should be.

P.S. Sorry for my crappy english, I'm really tired and this stuff is hard to explain for me . I hope that everything is much clearer for you now.