Pages:
Author

Topic: bit pit - (LP, ESMPPS, 8-decimal payout, SSL, API, 0% fee, Almost 0% Stales!) - page 11. (Read 80568 times)

member
Activity: 112
Merit: 10
i just figured out the defence against pool hopers!
once the pool shares reaches difficulty/2, then the shares are reset to zero and start over Wink
do you think it's possible?

Actually, if we went with PPLNS as our payment system, it would have a similar behavior.
full member
Activity: 226
Merit: 100
i just figured out the defence against pool hopers!
once the pool shares reaches difficulty/2, then the shares are reset to zero and start over Wink
do you think it's possible?
member
Activity: 112
Merit: 10

Ok, the stats should be catching back up now. Sorry for the inconvenience.

Ok, the stats are taking too long to catch up on their own... I'm going to restart the stats process manually and get it caught up.

EDIT: And now we're back!
member
Activity: 112
Merit: 10

and btw, the current shares are displayed wrong, I think they are over 1 million shares till now, but there are only 438000 displayed.


The shares and scoring system are currently correct, and working fine. Those API and stats in the top bar are driven via static files. Perhaps a cron task has failed to update them, I will look into it.

Ok, the stats should be catching back up now. Sorry for the inconvenience.
member
Activity: 112
Merit: 10

and btw, the current shares are displayed wrong, I think they are over 1 million shares till now, but there are only 438000 displayed.


The shares and scoring system are currently correct, and working fine. Those API and stats in the top bar are driven via static files. Perhaps a cron task has failed to update them, I will look into it.
member
Activity: 112
Merit: 10
Don't forget that the hoppers aren't the problem. They just aren't willing to work for the pool when it's paying less than market value for the shares. The problem is that the pool is rewarding predictably uneven (including above and below the market value).

Agreed,

We want everyone to get paid fairly. That is why we are currently looking into the various other reward systems.
legendary
Activity: 2576
Merit: 1186
Don't forget that the hoppers aren't the problem. They just aren't willing to work for the pool when it's paying less than market value for the shares. The problem is that the pool is rewarding predictably uneven (including above and below the market value).
newbie
Activity: 28
Merit: 0
wow! 140GH! with 100GH we have 1 block/day. then prop is working just fine!
i would say to monitor the pool behavior for a while before any changes are made Smiley

I agree with zoro. unfortunately my ISP had problems again - I was offline since saturday morning till a few hours ago. that is why I missed the short round, too Undecided
I think prop. is fine, until we really know how much pool hoopers there are. is there any way to meassure that?

and btw, the current shares are displayed wrong, I think they are over 1 million shares till now, but there are only 438000 displayed.

cheers iro
member
Activity: 112
Merit: 10

please do something against hopping like smpps. Together with the low stales this would make a formidable pool.

We are going to, and hopefully very soon. However, SMPPS has some fatal flaws and we are currently trying to model a few variants of SMPPS that should make it better.

I think the opinion of everybody has been that we should change from proportional to something else, and that something else should reduce variance (e.g., one of the many PPS variants).

So, at this point, it's simply a matter of us figuring out which PPS variant we all feel comfortable with, and switching to it. Switching to another reward system is going to be the easy part. For most of the PPS variants the code is pretty simple, and we'd switch over at the start of a new round.

In fact, the code for a variant we're calling xPPS is already done. Though, we're not entirely sure if this one is our best option.
legendary
Activity: 1708
Merit: 1019
Thanks everyone for your input. We are currently working on running some simulations to determine the behavior of the various PPS methods described above.

We hope to have some feedback shortly, but in the mean-time we are still very much interested in everyone's opinions.

please do something against hopping like smpps. Together with the low stales this would make a formidable pool.
full member
Activity: 226
Merit: 100
according to statistics this is usual for this pool Cheesy
i would expect the same thing in one or 2 days from now!
member
Activity: 112
Merit: 10
Great job everyone, we had two quick blocks in a row yesterday!
member
Activity: 112
Merit: 10
Thanks everyone for your input. We are currently working on running some simulations to determine the behavior of the various PPS methods described above.

We hope to have some feedback shortly, but in the mean-time we are still very much interested in everyone's opinions.
hero member
Activity: 560
Merit: 517
Quote
Tracking every single share is basically absurd.
Would you mind elaborating? I was under the impression pushpoold stored all submissions to its database anyway, and SQL servers will handle hundreds of millions of rows, if not more. If performance begins to degrade, you can just dump all the paid shares to a separate table or a backup database and remove them from the live database.

legendary
Activity: 2576
Merit: 1186
Quote
It remembers the difference, but it isn't a debt as nothing is considered due or owed.
Isn't that just arguing semantics?
It's important semantics. If people think it's debt, they will demand to be paid it. Since it's not debt, per system design, it's important that miners understand it so they're not surprised when/if they never get it.

SMPPS and its variations suffer from long-term block withholding attacks, and will never recover from them unless the pool-op uses their fees to re-balance the coffers, or uses Generation Fees if they are enough.
All reward systems suffer from block withholding in different ways. With SMPPS, it affects all the miners more or less equally.

Here's a variation that is at least partially immune to block withholding, and is simple to implement. I'm sure it was already mentioned in this thread:
Tracking every single share is basically absurd.
hero member
Activity: 560
Merit: 517
Quote
It remembers the difference, but it isn't a debt as nothing is considered due or owed.
Isn't that just arguing semantics?

Quote
Nowhere in there does it mention "remembering" the difference between the ideal payout and the realized payout.
It mentions it with "TotalMinerWork" and "TotalPaid" which are long-term values, lasting during the existence of your pool. After paying proportionally, TotalPaid increases but not enough to match TotalMinerWork, so the next round it will attempt again to make all the miners whole.

SMPPS and its variations suffer from long-term block withholding attacks, and will never recover from them unless the pool-op uses their fees to re-balance the coffers, or uses Generation Fees if they are enough.

Here's a variation that is at least partially immune to block withholding, and is simple to implement. I'm sure it was already mentioned in this thread:

On Valid Share: Add share to database, with values like (user_id, datetime, round, difficulty, value, paid). Value is the current PPS value (50 / Difficulty if the pool takes no fees). paid is False initially.

round and difficulty are for future-proofing, like handling Generation Fees. If the exact value cannot be determined at time of submittal, it can later (at end of round) be calculated from round and difficulty.

On Round End: Select all shares from the database where paid is False, sorted from oldest to newest. Using available pool funds, pay each share in full from oldest to newest, marking paid as you go, and stop when no more funds are available.

OR: Select all shares from the current round, pay in full from oldest to newest with half of the available funds. Repeat with all unpaid shares with the remaining funds.


It's simple from a coding perspective, and would result in less chance of making a mistake because you are keeping a log of all paid and unpaid shares (which you could even publish). You could even switch schemes at a later date, and still pay people fairly because you have a record of all their shares.

From a miner's perspective, it's simple: "You earn X amount of BTC per share, and get paid when funds are available."

And for the nosy miners who want more details: "Shares are paid out from oldest to newest, so you eventually get paid no matter what."

The alternative I mentioned above (in OR) might be preferred if people fear that they won't be paid soon enough for the work they are doing during the current round. The downside is that it does not discourage block withholding attacks, because everyone, including the attackers, get paid eventually. Block withholding will make payments take longer, so it's still affected by them but not in a way that starves the miners of income.
legendary
Activity: 2576
Merit: 1186
While it doesn't carry debt (because it never rewards more than it has), it does remember the difference, and try to make it up in the future.
Can you clarify? How can it try to make up the difference in the future if it doesn't remember the debt? Thanks.
It remembers the difference, but it isn't a debt as nothing is considered due or owed.
member
Activity: 84
Merit: 10
While it doesn't carry debt (because it never rewards more than it has), it does remember the difference, and try to make it up in the future.
Can you clarify? How can it try to make up the difference in the future if it doesn't remember the debt? Thanks.
newbie
Activity: 27
Merit: 0
Alternate to proportional need not be non-proportional

Maybe leave things just as they are  ...BUT...  report the Round....   .Shares.....Difficulty.....Duration.....Hashrate info from 24 hours ago

Alternatively

report total shares so far this week and make the distribution at week end when all the round info could be published too

Just thinking aloud here and wondering if all pool's reports are similar just because it started that way and became the norm? sort of making everyone Hashing lemmings
newbie
Activity: 43
Merit: 0
yes, my wifi-through.concrete.floor connection broke broke about one our before we found no. 10 and i even missed no.11 Sad

how about let the user choose between prop and smpps, like on deepbit.
Pages:
Jump to: