[ANN][AUTO-SWITCH] Profit-switch auto-exchange pool: CleverMining.com

[zneww]

The rejects caused by "switching coins" that Terk keeps is bullshit IMO.

I have tested a few switching pools and all rejects are under 3% without a doubt. What do they do? Not switch coins? I would like to know really why everyone has a ton of rejects.

[KC82]

I also wonder why rejects are higher here than with other coin switching pools. Maybe CM switches coins more frequently than others?

[tyj8tim]

Your thought process regarding lower diffs for a miner has been argued to death over at MC, the consensus is no, diff does not have any long term effect on profitability or otherwise.

Anywho, I see a lot of posts whining about high rejects and swapping pools because of it.  But no mention of profits had during said period.  If you swapped away from a pool just because of rejects without considering your btc/mhs, you're stupid. 

True that most people think it doesn't have long term effect. But IMO I think those conclusions are wrong, as pointed out in my description.

And again, I tallied the profit average between Wafflepool and CM (basically the same, CM 0.89% higher). However, CM has +10% reject rates (the pool average given at the stats page), and WP has ~<1% rejects (reported by many users, and I checked as well).

To do the math I'll use 3.33Mh/s for example

WP@ 3.33Mh/s --> 3.2967Mh/s after rejects
CM@ 3.33Mh/S --> 2.997Mh/s after rejects

Using average profitability from each pool

WP@0.010807 BTC/Mh/Day @3.2967Mh/s --> 0.035626BTC
CM@0.010903 BTC/Mh/Day @2.997Mh/s --> 0.032626BTC

Difference --> 0.00295 BTC

Not a completely negligible difference.

[Terk]

I checked the profit average from 2/5 to 2/17 from both sites (biggest window provided by wafflepool) and found

Wafflepool range average = 0.010807 btc/Mh/24hour
Clevermining range average = 0.010903 btc/Mh/24hour

While clevermining has +10% rejects, and wafflepool has <1% rejects.....Wafflepool is the clear winner atm.

CleverMining had some very rough days just after launch (Feb 4th) because of technical problems. Please compare either 30-day averages or last 3-4 days because we've been suffering from tech issues which were affecting our profitability between Feb 4th and Feb 13th).

[boondock66]

i started mining this pool with my 6gpu rig but after a few hours cards 5 & 6 go dead, also these 2 cards sometimes hang at a low hashrate on startup

anyone else had this problem?

[tyj8tim]

I checked the profit average from 2/5 to 2/17 from both sites (biggest window provided by wafflepool) and found

Wafflepool range average = 0.010807 btc/Mh/24hour
Clevermining range average = 0.010903 btc/Mh/24hour

While clevermining has +10% rejects, and wafflepool has <1% rejects.....Wafflepool is the clear winner atm.

CleverMining had some very rough days just after launch (Feb 4th) because of technical problems. Please compare either 30-day averages or last 3-4 days because we've been suffering from tech issues which were affecting our profitability between Feb 4th and Feb 13th).

No problem

2/14 to 2/17 stats from WP and CM
CM average 0.012403 BTC/Mh/Day
WP average 0.011455 BTC/MH/Day

WP@ 3.33Mh/s --> 3.2967Mh/s after rejects (1% rejects)
CM@ 3.33Mh/S --> 2.997Mh/s after rejects (10% rejects)

Earning Per day average
WP = 0.037763 BTC/Day
CM = 0.03717   BTC/Day

Different = 0.00059 --> WP still on top for now.

I'm not trying to to say CM is a bad pool. I love the designs and easy access to quick stats. If the reject rates can be lowered to a comparable level like WP, you can expect pool's hash power to increase exponentially.

[Prima Primat]

@tyj8tim: Share difficulty has nothing to do with reject percentage. Or with overall profits, or with anything that matters.

I actually have my doubts on that with a switching pool.

I agree that IF we mine at a pool that only mines one coin, then the variance will average out and you will see a steady payout after a few hours or 24 hours.

However, at CM we get a lot of stale rejects. And I'm assuming these are not all counting towards our hashrate (correct me if I'm wrong). From my understanding, stale rejects are the ones that's too old because the pool already pointed to the new block. So if someone has a ~10% reject on average, that means ~10% of the time when he finished his share, the pool already moved on and his calculated shares are useless. This is especially a problem for a pool that's hopping around and getting into block times that's less than a minute or even 30 seconds.

So IMO, by lowering the share difficulties, you can shorten individual miner's share round time and minimize the probability of submitting stale shares.

Share difficulty isn't really a round time. It's more of a threshold. You're calculating shares all the time, but most of them have a value below the difficulty threshold, so you're not submitting them.
A share difficulty of 500k doesn't mean you're mining away on a huge chunk for ages (which would be bad for fast block times), it just means that you only submit a share when it randomly happens to have a value over 500k. But that single share wouldn't be any more likely to be rejected than any diff 16 share.

Also, empirically speaking, Middlecoin has a difficulty of 1024 but I never had particularly high reject rates there, always around 2-3%.

[becks619]

Hi everyone,

I have just received my payout from altcoinplex.

Its about 0.024 BTC/MH for two days.

They didnt payout yesterday due to some issue with dogecoin blockchain. However, they payout to me for today and yesterday.

http://stats.altcoinplex.com/users/bitcoin/1MjNG9pGippwTSHraPnB1MBKwjr4q2ykiY.html

[tyj8tim]

Share difficulty isn't really a round time. It's more of a threshold. You're calculating shares all the time, but most of them have a value below the difficulty threshold, so you're not submitting them.
A share difficulty of 500k doesn't mean you're mining away on a huge chunk for ages (which would be bad for fast block times), it just means that you only submit a share when it randomly happens to have a value over 500k. But that single share wouldn't be any more likely to be rejected than any diff 16 share.

Also, empirically speaking, Middlecoin has a difficulty of 1024 but I never had particularly high reject rates there, always around 2-3%.

Wouldn't a 16 share take less time to compute compare to a 1024 share?

Middlecoin also have multiple servers, which I think is one of the biggest problem with CM right now.

[Prima Primat]

Share difficulty isn't really a round time. It's more of a threshold. You're calculating shares all the time, but most of them have a value below the difficulty threshold, so you're not submitting them.
A share difficulty of 500k doesn't mean you're mining away on a huge chunk for ages (which would be bad for fast block times), it just means that you only submit a share when it randomly happens to have a value over 500k. But that single share wouldn't be any more likely to be rejected than any diff 16 share.

Also, empirically speaking, Middlecoin has a difficulty of 1024 but I never had particularly high reject rates there, always around 2-3%.

Wouldn't a 16 share take less time to compute compare to a 1024 share?

Middlecoin also have multiple servers, which I think is one of the biggest problem with CM right now.

No, it wouldn't. Watch your cgminer (or whatever you use) closely. You sometimes get huge shares, in the hundreds of thousands or even millions, out of nowhere. These don't take long to compute, they're just unlikely to happen.
And yeah... more servers would be neat indeed.

[_r2h]

Difference --> 0.00295 BTC
Not a completely negligible difference.

I would call that negligible.

[tyj8tim]

Difference --> 0.00295 BTC
Not a completely negligible difference.

I would call that negligible.

Really? Maybe because you don't actually get that much?

Let's calculate it with BTC@1000 for simplicity.

That's $2.95 a day. Round that up to $3.

30 Days a month, that's $90.

If someone has a 10Mh/s setup, that's $270 difference.

Far from negligible.

[zneww]

Difference --> 0.00295 BTC
Not a completely negligible difference.

I would call that negligible.

Really? Maybe because you don't actually get that much?

Let's calculate it with BTC@1000 for simplicity.

That's $2.95 a day. Round that up to $3.

30 Days a month, that's $90.

If someone has a 10Mh/s setup, that's $270 difference.

Far from negligible.

muney

[pjv]

Wouldn't a 16 share take less time to compute compare to a 1024 share?

that's a total misunderstanding of how the mining process works. every share you calculate takes (roughly) the same amount of time to calculate. your GPU's are churning through hashes as fast as they can and only submitting the ones that are "better" than the current difficulty. each one takes just as long as the one before and the one after to calculate, but only some of them are better than the current difficulty. it doesn't take any longer to calculate them - the difficulty just makes it less likely that any given calculated share will be good enough to submit.

the reason we are all so keen to have our GPU's have higher kHash rates is not because that makes them calculate a given share faster, it's because it measures how many shares they can spew out every second.

[Terk]

Regarding shares rejected percentage

Coin-switch pool is always a balance between profitability and wasted hashpower. Different pools can have different rejected percentages because of different coin switching strategies and/or aggressiveness and/or mining different coins with different difficulty/speed. Our strategy is quite aggressive as I found it provides good results.

I tweaked our strategy a little bit to further discriminate high-reject coins and will see how it changes both reject ratio and profitability in the next hours. I hope we should expect lower rejected ratio from now on. It was deployed an hour ago and in the last hour pool-wide rejected percentage was 8.8% compared to 14-15% in the previous three hours. One hour is a small sample but I hope the trend will continue. It might take couple of hours with gradually lowered rejected ratio before it will settle, because of the nature of this tweak. Please let me know what your experience will be.

If you compare pools, you should compare your net profitability. Wafflepool publishes their today's (partial day) profitability as 0.0083 BTC/MHs. CleverMining profitability for the same period is 0.0108 BTC/MHs (I don't publish partial day results on the website, but I took this from database to compare exactly the same period of time). I didn't find Waffle's published pool-wide rejected ratio, but even if it's 1% (Middlecoin for example has 6.2% today), then you get 0.99 * 0.0083 = 0.0082 BTC/MHs there. Our average rejects today is 11%, so here you get 0.89 * 0.0108 = 0.0096 BTC/MHs, which is still 17% more even after wasting 11% hashpower on rejects.

We could achieve as low as 2-3% rejects but it would hurt profitability. Lowering rejects is important, but net profitability is much more important.

That said, with changes that I just made, I hope we'll be able to lower rejected ratio without compromising profitability too much. I'm tweaking and monitoring changes continuously to find the right balance.

[pjv]

Difference --> 0.00295 BTC
Not a completely negligible difference.

I would call that negligible.

Really? Maybe because you don't actually get that much?

Let's calculate it with BTC@1000 for simplicity.

That's $2.95 a day. Round that up to $3.

30 Days a month, that's $90.

If someone has a 10Mh/s setup, that's $270 difference.

Far from negligible.

muney

It's negligible compared to the statistical variance of mining luck over the time period measured. If that difference persisted for months, then for sure not negligible. But over a period of three days it is pure statistical noise and doesn't say anything conclusive one way or another.

[Terk]

2/14 to 2/17 stats from WP and CM
CM average 0.012403 BTC/Mh/Day
WP average 0.011455 BTC/MH/Day

WP@ 3.33Mh/s --> 3.2967Mh/s after rejects (1% rejects)
CM@ 3.33Mh/S --> 2.997Mh/s after rejects (10% rejects)

Earning Per day average
WP = 0.037763 BTC/Day
CM = 0.03717   BTC/Day

Different = 0.00059 --> WP still on top for now.

I'm not trying to to say CM is a bad pool. I love the designs and easy access to quick stats. If the reject rates can be lowered to a comparable level like WP, you can expect pool's hash power to increase exponentially.

Our rejected ratio wasn't that high during this period. It was increasing daily because of strategy changes that I made and reached its peak today. Daily values for 2/14-2/17 were 5.31%, 6.30%, 7.47% and 8.90%, which gives 7% average (today it reached 11% partial-day average before I made changes to reduce this).

That should give 3.33 MH/s => 3.0969 MH/s after 7% rejects and 0.038410851 BTC/day.

I'm not trying to argue over pennies/satoshis here, just want to provide you the correct data to get your math straight.

I'm fully aware that high reject ratio is a problem at CleverMining and I'm actively investigating and working on this issue as I wrote in the post above. Last hour rejected ratio was 8.8%, compared to 14-15% in three previous hours. Current hour is 6.7%. I think the changes are working well.

I won't ever aim to get as low as 2% rejected ratio. It's not that hard to do, but it would need to be done by leaving some money on the table and net profitability is my main objective.

[Prima Primat]

... and net profitability is my main objective.

Amen to that!

[sammy007]

Mining for 1 hour, stats -- 829Kh instead of ~2.4Mh.

[Prima Primat]

@Terk: You should put the most important information regarding rejects in the 1st post, or even in the FAQ on the website. The topic keeps coming up, and I think it does put a lot of people off.