Topic: [1050 TH] BitMinter.com [1% PPLNS,Pays TxFees +MergedMining,Stratum,GBT,vardiff] - page 226. (Read 837122 times)

no worries, I just had not noticed that until today.

Thanks Dr. Haribo!!

New blocks are showing as stale, I'll get this changed to "processing" or something like that soon.

Why are they like that for more than some seconds? Because while the mining server has a lighter load than usual, the payment processing system is lagging badly. I'll improve this soon.

Apologies for the confusion it causes to show so many "stale" blocks, especially with all the fast blocks we had today it looked crazy.

Why are all of the latest namecoin blocks stale?

The funny thing is that the load on the mining server hasn't been lower in months. I guess some slow miners gave up and were replaced with ASIC miners. With var diff it's the number of workers and connections that matter, not the hashrate. Whether you have 1 Ghps or 1 Thps doesn't matter to the server - the load is the same.

Wow look at realasicminer go!  Is there an end in sight, as far as how high he can go? 

Probability theory and statistics are not my fields of expertise. The smart thing for me then is to rely on people who are good with this.

If people like organofcorti or Meni Rosenfeld, people who understand numbers, tell me that the numbers look strange, then I worry.

When miners see bad luck and let their emotions run away with them, I ask organofcorti to perform a calm and thorough analysis of the data and present his findings so everyone can see what the situation really is.

If you want to make the most coins, read his report when it's out. When it comes to money, math beats emotions every time.

By the way, we just got 5 blocks in 22 minutes.


I'm sorry you feel disappointed with the 1% fee. I was a bit worried about introducing a fee, but surprisingly I got mostly positive feedback.

Last december a string of orphan blocks killed all the profits of the pool, back to the beginning. So the pool had zero profit. Server costs were not covered. What about lost income from the 1.5 man years of work I had put into this. As a skilled software developer and web security specialist, that's .. a few coins. BitMinter was a disaster. I couldn't keep working my ass off to help others make money, while I only had losses (server costs) myself. A minimum wage job would be a dream in comparison.

So now we are the second cheapest pool, after P2pool. That's a much smaller pool with much more variance. If you think the luck swings on BitMinter are bad, then you'll be running away screaming. But if you set p2pool donation to zero, set up merged mining with automatic restart of namecoind when it locks up daily, and ensure you have good and fast connectivity to the most important nodes on the bitcoin peer-to-peer network to avoid orphans, and pray that other p2pool users have the same, then you should come out a little bit ahead on average. If you're lucky. I'm hoping that what BitMinter has to offer is worth more than that.


Thanks, that is much appreciated!

"Set and forget" doesn't mean you should expect something for nothing. If you use p2Pool, are you going to turn off the donation to the developers?

 thanks for  the thanks.  I really should do it perfectly correct but frankly it is a shit load of work.  I am still satisfied the pool is running honestly. I am mining my 10Gh/s here and always mined with the doctor's pool since i found him last August 2012.  As for the 1% fee causing a problem for a miner pleaseee .

I have paid the Doctor 2 to 4 percent since August of 2012 why because he bothered to write an easy interface that to this day no one else has.

I don't understand what you imply here (given your quote). If it's Re: the 1% fee, I think it was long time due, given the added value of BitMinter, the excellent uptime and reliability of the pool, etc. If it's Re: the notification method, I see nothing wrong in using the website to communicate (it has been pinned as a very visible banner for weeks), e-mail would only have been a plus IMHO.

We just want some luck now

PS @organofcorti @philipma1957 thanks for you comments and analysis.

I guess it would be very useful if there was some way to separate analysis for getwork, stratum and GBT. A statistical anomaly for one of the 3 protocols would be a great hint in case there was indeed a problem.

I'm one of the "set and forget" miners and I'm kind of disappointed. I knew this day would come... I guess it's time for me to run p2pool

dude that is just wrong I can pick 5 good days out of the last 14.  edit only 4 good days.  maybe the server needs more balls/power.. but this is 14 days  and I ballparked some math as I am tired



  april 25 17 blocks   13.6  would have been normal for 6.1Th hash and 8.97 mill diff.    very good luck
  apr 26th 12 blocks   13.7  would have been normal for 6.2 Th hash and 8.97 mill diff .. slightly bad luck
   apr 27th  8 blocks   13  would have been normal    for 5.8Th   and 8.97  mill diff      bad luck   for this 3 day period we mined 37 blocks should have mined 40.3      about 91.8 percent  or normal add in the nmc  you push against most any other pool.  

 apr 28th 10 blocks  hash is 5.5th        12.32 normal   below
apr  29th  15 blocks  hash is 5.5th        12.32 normal     above
apr 30th     6 blocks  hash is 5.7th        11.36 normal     way below   difficulty changed to 10.07 mill

31 total for 3                                          36.08 normal  

for 6 days 68 mined                        76.38 normal


may 1 13 blocks hash is 5.9          11.76 is normal   above
may 2 10 blocks   hash is 6.1        12.16 is normal  below
may3   7  blocks    hash is 6.2       12.36 is normal below

for 9 days 98        113.08 is normal

may 4 10 blocks      15.5   under
 may 5 20 blocks     16.5    over
may 6 11 blocks      16.6        under
 may 7 21 blocks     20.96    just over
 may 8 18 blocks    22          under      

4 out of 14 good 2 close 7 bad

178 block mined  out of  204.56    

 this is roughly 87 percent of normal   .

 if the trend continues I would guess that the server struggles to keep up with the load.   remember there is down time and this number ignores all down time.   it also does not reflect the factor of 1.08 for name coins. so 1.08 x 178 = 192.24 divide by 204.56 =  93.9 percent and it still does not reflect any down time.   so it is not as bad as it seems.   but maybe more power is needed to handle higher volumes.

The "luck" is best thought of as number of shares per round / difficulty. The average number of shares per round / difficulty for a number of rounds is Erlang distributed. So you can work this out for your self.

1. Get the following data from either the API or website: Accepted shares and difficulty
2. copy paste, and divide shares per round by difficulty for example (using last 20 rounds):

3. Once you have the average, you can use wolfram alpha to figure out how likely that result is. The command is:

So for example, the data above would be:

CDF[ErlangDistribution[20, 20], 1.22444]


4. If the result is less than 0.975 or greater than 0.025, then you can consider the result within a 95% confidence interval and not very unusual. The result for the last two days is about 0.844, so the "luck" has not been unusual at all.


If you want to find out the same result for any number of rounds, then just follow the steps above and you can work it out for yourself.

Enjoy!

notice how nobody complains when bitminter mines three blocks in a row...

Will

DrHaribo, may I ask this question without sounding accusing or implying anything because I'm honestly wondering:

How unlucky and for how long should it last before you would start to suspect anything? I don't have many months of experience but for the last 2-3 weeks I've only seen bad luck. Weeks are not a long time you will say, and bad luck can be just that, granted. But if the block/hashrate stayed like the past 2 weeks for, say, 2 more years, should we still not worry at all?

So basically because I'm biased and the pool has lived long before I joined, I was wondering when someone like you would start to worry.

 how about the 2 blocks right before it  6m and 33m   and the block after it took 11m.

 btw my 3 large samples dating from april 5 to now show no oddness.

 1 ratio of btc blocks to nmc blocks earned was high one was low and one was close to perfect.

Why is bad luck strange? Wouldn't it be more strange if we only had good luck all the time?

Wow, 6 and half hours without block ?

Imho
Looks strange with pool speed more than 11 000 ghs !

Ok, true, more users in the pool affects your work reject ratio, but the difference will be tiny.

The BitMinter server orders long pollers by hashrate for getwork and getblocktemplate (GBT). For Stratum there is no ordering when pushing work (at least not yet).

Currently at each block change work is pushed to 7000-8000 Stratum connections in 0.01-0.03 seconds. 8000-9000 long pollers (getwork/GBT) are notified in 1.5-1.8 seconds. A few more ASICs running Stratum won't really make any difference.


Yes, the server software is closed source. And it wouldn't make much difference if it was open source, because you would never know if the source code I gave you is what I was running on the server.

Not sure what you mean having to pay for the miner. There's no extra fee for using BitMinter client over other mining clients.

However, that depends on the order you send LP messages.
If you order them higher hash rate first (like slush does) then those with a lower hash rate will get their LP that tiny bit later, if there are more higher hashing people on the pool. Of course this only means a small drop if it is the case.

Also of course without the source code one can't verify how the pool prioritises responses to miners.
But since you have to pay for the miner (it isn't free), I'd guess no one else has the source code

There is some confusion lately that having big miners in the pool means the smaller miners get paid less. That's not how it works. Your average payout is determined by your hashrate, the difficulty, and fees/donations. Other people's hashrate does not influence this at all.

If someone joins the pool with 10% of the hashrate, then the pool will be pulling in 10% more bitcoins, the new user will be doing 10% of the work, and the new user will get 10% of the payouts. You don't make more coins by mining in a big/small pool, and you don't make less money mining in a big/small pool.

But a higher pool hashrate reduces variance for everyone in the pool. That means you will have smaller variations than before in your daily payouts.