[ANN] profit switching auto-exchanging pool - www.middlecoin.com

[Lollaskates]

Needs 256 diff. I've got 5.8MH/s pointed at this and getting 60% and 70% rejects on my two rigs. I'm about to mod cgminer to forcibly split your work between my GPUs. Too many blocks go by with not a single share submission = 100% wasted time and electricity. When that happens it actually costs us money, seeing as how we still have to pay for the electricity. Mining on other pools until this is addressed.

Yes, this is a problem (there's a reason VARDIFF exists, for optimization for all combinations of hash strength and latency). After reading the thread it seems you are unwilling to accept this as a valid issue, which is sad.

[hinouchie]

If you aren't happy with the pool just leave ... geez so many ungrateful people always demand so fucking much when they contribute nothing.

@h2odysee

Awesome job with the pool dude, I love how I don't have to deal with the trading and such, I can simply just sit back and mine. Do whatever you think it's good for the pool, you can't please them all. Cheers~

[fredeq]

Hi, I was all in for the past few days to try 256 difficulty but something just occurred to me.

It doesnt matter if we go 32, 512 or 1024 for the global profit, if someone is about to find a block he will do it regardless of the share diff. Changing share diff wont increase the pool profit imo.

We can still speculate if the 512 is making old rigs fall behind and the 7xxx rigs may be indeed chewing income from those rigs, but the only way to know for sure would be to compare % of reject Mhash or daily payouts per acceppted average Mhash.

[CoinBuzz]

If you aren't happy with the pool just leave ... geez so many ungrateful people always demand so fucking much when they contribute nothing.

Are you leave your country for it's problem when you can help to make it better? We all love this pool, so we want to make it better by each day. everybody can leave! Those who stay and help, those are real men!

I've talked way too much about difficulty in this thread. It doesn't play a factor.

I think we can experiment it in other way. I suggest to increase the difficulty to 1024. We can see it's effect and also lower the server load

[juhakall]

I tried the pool for two days and my reject rate was about 6%. My average hashrate also dropped by a few percent, because of all the switching I presume. PPS ended up below 200% and I wasn't too satisfied. However, it's totally wrong to claim that high share difficulty could cause more rejects or loss of shares. Every diff1 share is evaluated and discarded if it doesn't meet the pool requirement. Observe the best share counter right after restarting. Or the API, and see how diff1 shares increases much more often than difficulty accepted. A high diff share does not take longer to calculate, every diff1 share just has a 1 in 512 chance of also being a diff512 share that gets submitted.

[WhPh]

First of all, this pool is great and I'm mining here since its very beginning.

And now I have yet another 2 cents about pool difficulty.
I understand that pool difficulty doesn't impact pool profitability. As long as coin difficulty is greater then pool difficulty, we still need much higher then 512 to find a block.
And unless the block is found there's nothing to divide between its members, even if the majority of shares are simply rejected or discarded by cgminer due to new block found.

However...
When mining some sort of fast-block coins, many 512-diff shares are simply not submitted. This leads to low hashrate in the table (cosmetic problem), but also to another, more serious issue. As soon as block is found, it is divided to share submitters as it was recorded by the pool. There are some lucky members who succeeded submitting shares during that round, but most of shares weren't recorded. So pool members' luck has serious impact on payments here.
This leads to somewhat unfair block fee payment for unlucky pool members in a short run. In long run it should be still fair though.

Do I miss or misunderstand something?

[Eastwind]

I really think we need two things in this pool.
256 difficulty.
2 payments per day.(i thought we agreed this was ok before?)
anyway, since were not constantly mining the same coin, the "512 diff is the same as any other diff" doesnt apply now if we were constantly mining one single coin, that is different. but when we switch to 512k diff coins, and only submit one share in an hour, then it changes to another coin that is 1m diff, then we wasted all that work on nothing.

2 payments a today will not pay us more than 1 payment a day. So I think it is OK.

I do agree with your suggestion of 256 difficulty. I moved all my 7950 rigs away from this pool and have only one 7970 rig (total hash 2.9 Mh/s) here. As it is the only one can cope with 512 difficulty for fast coins.

I have a question about share mining. Is one share distributed in all the GPUs in one rig, or one share is mined by a single GPU?

[Eastwind]

We can be "autoswitching into dud coins" because it is very possible that coins look very nice on paper (theoretically the most profitable) but in practice it could not be the most profitable, even by a long shot. The whole point of this pool is indeed to attempt to switch to the most profitable but perhaps some real-world conditions are coming into play that are not taken into account when calculating the most profitable coin on paper

There are some real world factors I don't take into account very well. One is that when we switch, we lose out on a few seconds of hashing. I've "tweaked" some specific coins to adjust for this, as to not switch as often to coins that are only going to be mined for 10 seconds then switch back.

Also, I'm sure it's the case that having that gap while switching is putting wear and tear on GPU fans, and creating system instability.

The real solution is to eliminate the gap.

One rig keeps on crashing mining here, so I move it away. It is fine mining LTC directly.

[Eastwind]

Hi, I was all in for the past few days to try 256 difficulty but something just occurred to me.

It doesnt matter if we go 32, 512 or 1024 for the global profit, if someone is about to find a block he will do it regardless of the share diff. Changing share diff wont increase the pool profit imo.

We can still speculate if the 512 is making old rigs fall behind and the 7xxx rigs may be indeed chewing income from those rigs, but the only way to know for sure would be to compare % of reject Mhash or daily payouts per acceppted average Mhash.

We are not just competing within our pool to solve blocks. We are competing with all the computers in the world. So if other pool's share difficulty is low, their 5xxx and 6xxx rigs can work as well as our 79xx.

[Cryptos2go]

Yes lower Diff Could help maintain productivity on the fast coins.

It's also just as likely low diff could result in Less productivity on fast coins.

You have to realize the server will have to make 2x the work allocations at 256 difficulty. We should assume this server is already doing significantly more work than a typical single coin pool server.

When this pool alone drops 600+mh on a 20 second coin that currently has a 1-200 mh difficulty equivalent, (not to mention multipool and others running cryptoswitcher, cgwatcher, etc or going off coinchoose) the blocks will and do fly so fast you would do well to get even 1 share @ 32 or less difficulty for each block. Sometimes maybe more than 1 x 32 diff share but not many more if even that on average.

Remember the 5 second rule and LITTLE ORPHAN FUKIN ANNIE? When I say 5 second rule I am not talking about dropping your dam hot pocket on the floor either.    

I'm not making this up folks. I have watched it play out time and time again running Load Balance between Middlecoin and a dedicated pool with fairly well optimized VARDIFF on the same coin at the same time.

To be fair H20 is well aware of this "overloading" that occurs on fast coins with low diff and too many blocks between diff adjustments scenario and the orphans that result from it.
 
503guy Replied to my last post by flatly saying "the (coin) diff changes"  Hes right, in the sense that it's both part of the problem and the solution, a solution that creates another problem, for switching pools anyway.


 H20, Lollaskates is right about VARDIFF IMO and some pools implement it better than others IME. Optimized VARDIFF keeps the work moving like no other though I would assume it comes at some cost in terms of server resources. Please don't give up on the VARDIFF.

H20 Earlier you made an open offer that in exchange for assistance with setting up a certain coin you would waive pool fees. Obviously you have similar options for the pool in general. This could range from technical advice to hardware funding.

@ hinouchie  Nothing is exactly what you will see if people just left the pool without offering their respective suggestions for a pool they obviously care enough about to post issues,suggestions, concerns etc.
I suggest you take a break from trolling and take a look at the payouts from this pool to those who are no longer mining at it, wrap your head around that, do the math on what the pool could be hashing at then google symbiotic relationship and connect the dots.

[fredeq]

Hi, I was all in for the past few days to try 256 difficulty but something just occurred to me.

It doesnt matter if we go 32, 512 or 1024 for the global profit, if someone is about to find a block he will do it regardless of the share diff. Changing share diff wont increase the pool profit imo.

We can still speculate if the 512 is making old rigs fall behind and the 7xxx rigs may be indeed chewing income from those rigs, but the only way to know for sure would be to compare % of reject Mhash or daily payouts per acceppted average Mhash.

We are not just competing within our pool to solve blocks. We are competing with all the computers in the world. So if other pool's share difficulty is low, their 5xxx and 6xxx rigs can work as well as our 79xx.

No. Lower share difficulty does not make you find blocks easier and we are not competing with the rest of the world about share numbers.

[btcdrak]

First of all, this pool is great and I'm mining here since its very beginning.

And now I have yet another 2 cents about pool difficulty.
I understand that pool difficulty doesn't impact pool profitability. As long as coin difficulty is greater then pool difficulty, we still need much higher then 512 to find a block.
And unless the block is found there's nothing to divide between its members, even if the majority of shares are simply rejected or discarded by cgminer due to new block found.

However...
When mining some sort of fast-block coins, many 512-diff shares are simply not submitted. This leads to low hashrate in the table (cosmetic problem), but also to another, more serious issue. As soon as block is found, it is divided to share submitters as it was recorded by the pool. There are some lucky members who succeeded submitting shares during that round, but most of shares weren't recorded. So pool members' luck has serious impact on payments here.
This leads to somewhat unfair block fee payment for unlucky pool members in a short run. In long run it should be still fair though.

Do I miss or misunderstand something?

This is correct. Pooled mining works by everyone proving they are working on the problem. They don't submit all their shares, the difficulty is set so that you only submit some of your shares. That way you can prove you are working on the puzzle and you can proportion the effort by working out the percentage of everyone's submissions. The problem with a high difficulty for a fast coin is that some people might be working on the puzzle, but never get to submit any shares, therefor they never get to prove they were working on the puzzles - so for some blocks, they will actually miss out on a piece of the pie for that block.

[Tigggger]

This is correct. Pooled mining works by everyone proving they are working on the problem. They don't submit all their shares, the difficulty is set so that you only submit some of your shares. That way you can prove you are working on the puzzle and you can proportion the effort by working out the percentage of everyone's submissions. The problem with a high difficulty for a fast coin is that some people might be working on the puzzle, but never get to submit any shares, therefor they never get to prove they were working on the puzzles - so for some blocks, they will actually miss out on a piece of the pie for that block.

But then the reverse is true because when you do have submitted shares, they will represent a much greater % of the pie than they should be, so in the end although it may appear that high diff is hurting it may not be.

My gut further up this thread was that low diff for the fast coins would help, but now I'm not so sure.

[Eastwind]

Needs 256 diff. I've got 5.8MH/s pointed at this and getting 60% and 70% rejects on my two rigs. I'm about to mod cgminer to forcibly split your work between my GPUs. Too many blocks go by with not a single share submission = 100% wasted time and electricity. When that happens it actually costs us money, seeing as how we still have to pay for the electricity. Mining on other pools until this is addressed.

Yes, this is a problem (there's a reason VARDIFF exists, for optimization for all combinations of hash strength and latency). After reading the thread it seems you are unwilling to accept this as a valid issue, which is sad.

What are the GPU's in your two rigs?

[TXSteve]

We had this exact same issue on a small private pool. We tried lower share diff and it made no difference re profitability. I know it's counter intuitive, but it's true.

You need to reduce your rejects with cgminer settings.

On a 7950, with 2 gpu threads with an intensity of 13. You'll get a lot lower rejects but also a lower hash, of about 560.

On a 7970, with 2 gpu threads, and I=13 you'll get about 720 hash and very few rejects

on a 7950, with 1 gpu thread, an I=20 will give a lot more rejects (than I=18 or I=17) on low diff coins

low scan time & expiry also seem to help too
if you use a low expiry you MUST submit your stale shares !! -- most of them will be accepted

Everything is a trade off, if you cut the rejects down on low diff coins, you'll lose hash on high diff coins

Here's the 7950 settings I use for 2 gpu threads, and I=13 -- hash aprox 560

{

"intensity" : "13",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "2",
"thread-concurrency" : "8192",
"gpu-fan" : "20-85",
"gpu-engine" : "800-1050",
"gpu-memclock" : "1500",
"gpu-memdiff" : "450",
"gpu-powertune" : "11",
"gpu-vddc" : "0.000",
"temp-cutoff" : "85",
"temp-overheat" : "80",
"temp-target" : "75",
"auto-fan" : true,
"auto-gpu" : true,

"queue" : "5",
"scan-time" : "1",
"expiry" : "1",
"scrypt" : true,
"gpu-threads" : "2"
}

Here's the 7950 settings I use for 1 gpu thread, and I=(20,19,18,17)-- hash aprox 625
if you have 4 gpus, set each GPU to a different Intensity then run cgminer for awhile, look at the onscreen individual GPU stats for accepts vs rejects and see which intensity gives the best overall performance, then set all gpus to the best intensity

{
"intensity" : "20,19,18,17",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "2",
"thread-concurrency" : "22336",
"gpu-fan" : "20-85",
"gpu-engine" : "900-1045",
"gpu-memclock" : "1500",
"gpu-memdiff" : "455",
"gpu-powertune" : "0",
"gpu-vddc" : "0.000",
"temp-cutoff" : "85",
"temp-overheat" : "80",
"temp-target" : "75",
"auto-fan" : true,
"auto-gpu" : true,

"queue" : "5",
"scan-time" : "1",
"expiry" : "1",
"scrypt" : true,
"gpu-threads" : "1"
}

[Kuroth]

"On a 7970, with 2 gpu threads, and I=13 you'll get about 720 hash and very few rejects"

This is true and exactly what I get..  I run 2 7970's at these settings..


One thing, not sure if its a clue, but the last 2 days my cgwatcher has restarted my miner several times as the "Accept" has not taken place for 5 min..  I bumped this to 10 min(Which I dont think it good)..  could this be an issue that it has happened several times in the last 36 hrs?  (I have had this setting on since day one and this is the first time I have seen it triggered)

[San1ty]

I'm running on -I 20 with two 7950's running at 620 each with barely any rejects. Less then 1 percent.
I don't understand that so many people are having issues, am I missing something?

[Kuroth]

But I will say the last 2 days payouts have been very low and so far this morning, looks on track to be a another very low payout day..

[h2odysee]

We've been mining FTC for the past hour. We just switched to DGC, so your rejected rate is expected to go from 1% to 15%.

[SullyTex]

I see that a payout "processed" but yet to see anything in my cryptsy wallet.   It has usually not taken near this long.  Was there issues??