.http://hhtt.1209k.com/user-details.php?user=1DPJHpt2hBNNumXqErSzw4r5SzFk9yL5X2
I got 0.00071569 per 32 difficulty share found instead of 0.00064245.
Any reason why?
Topic: [500 GH/s]HHTT -Selected Diff/Stratum/PPLNS/Paid Stales/High Availability/Tor - page 30. (Read 56546 times)
September 01, 2012, 07:41:55 AM
I was getting over paid for about 30 mins .
http://hhtt.1209k.com/user-details.php?user=1DPJHpt2hBNNumXqErSzw4r5SzFk9yL5X2
I got 0.00071569 per 32 difficulty share found instead of 0.00064245.
Any reason why?
.http://hhtt.1209k.com/user-details.php?user=1DPJHpt2hBNNumXqErSzw4r5SzFk9yL5X2
I got 0.00071569 per 32 difficulty share found instead of 0.00064245.
Any reason why?
August 31, 2012, 09:30:12 PM
Yeah, that was me kicking over the pool daemon. I changed the code to use difficulty 2 for anyone who wants to use 1 until I get that bug worked out. It should be fine now.
I figured maybe you restarted it or something. Okey. Thanks
August 31, 2012, 09:16:32 PM
Oh, and, uh... this
Out of nowhere it started throwing this a few minutes after mining. And it just stopped and my shares are being accepted again.
Quote
[2012-08-31 22:11:04] Rejected 3e934289.9dba2434 GPU 0 pool 0 (unknown-user)
[2012-08-31 22:11:37] Rejected 69e032bf.4c96ad30 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:50] Rejected 1cb4a275.52a67d59 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:54] Rejected 3ad22116.8e18450a GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:37] Rejected 69e032bf.4c96ad30 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:50] Rejected 1cb4a275.52a67d59 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:54] Rejected 3ad22116.8e18450a GPU 1 pool 0 (unknown-user)
Out of nowhere it started throwing this a few minutes after mining. And it just stopped and my shares are being accepted again.
Yeah, that was me kicking over the pool daemon. I changed the code to use difficulty 2 for anyone who wants to use 1 until I get that bug worked out. It should be fine now.
August 31, 2012, 09:15:11 PM
Difficulty does not need to be an integer (concept-wise), though I'm nearly sure an integer is the only value supported by mining software.
Yeah, I was talking about my pool specifically. My difficulty to target code is pretty stupid (as evidenced by my bug with difficulty 1) and certainly only works with integers.
August 31, 2012, 09:12:55 PM
Oh, and, uh... this
Out of nowhere it started throwing this a few minutes after mining. And it just stopped and my shares are being accepted again.
Quote
[2012-08-31 22:11:04] Rejected 3e934289.9dba2434 GPU 0 pool 0 (unknown-user)
[2012-08-31 22:11:37] Rejected 69e032bf.4c96ad30 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:50] Rejected 1cb4a275.52a67d59 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:54] Rejected 3ad22116.8e18450a GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:37] Rejected 69e032bf.4c96ad30 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:50] Rejected 1cb4a275.52a67d59 GPU 1 pool 0 (unknown-user)
[2012-08-31 22:11:54] Rejected 3ad22116.8e18450a GPU 1 pool 0 (unknown-user)
Out of nowhere it started throwing this a few minutes after mining. And it just stopped and my shares are being accepted again.
August 31, 2012, 09:09:59 PM
Difficulty 1 shares don't seem to be working. 2 worked fine, but work doesn't send when I use difficulty 1. 450Mh/s, but I waited well over a minute for a single share and saw none. I assume it defaulted back to 32 if there was a problem with 1.
I'll use 2 for now, its fine.
I'll use 2 for now, its fine.
Yeah, there is a bug with difficulty 1. I'll figure that out, thanks for letting me know.
August 31, 2012, 09:04:24 PM
So far I really like the pool, my only complaint is that the web site is kind of goofy looking. Makes it hard to look at while at work - people will think I'm f&*cking around. I may change my mind about running solo when the ASICs arrive. Currently running at difficulty 10 with 1.4Gh/s.
Ha. I thought other mining pools tried to hard to look all business and wanted to have a little fun.
August 31, 2012, 09:01:51 PM
Difficulty 1 shares don't seem to be working. 2 worked fine, but work doesn't send when I use difficulty 1. 450Mh/s, but I waited well over a minute for a single share and saw none. I assume it defaulted back to 32 if there was a problem with 1.
I'll use 2 for now, its fine.
I'll use 2 for now, its fine.
August 31, 2012, 08:59:31 PM
Bitcoin difficult still seems on the rise if anything, so I would just knock that up the variance in submitting a few shares quicker than expected in an hour, than you did last time.
Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
Variance at diff 32 is 32^2 times greater, i.e. 1024 times... so in my opinion, diff 32 is too high for current mining hardware in light of that (see my earlier post).Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
No - variance increase with mean not square of mean. And of more relevance to users is the Standard Deviation which is the Square Root of variance. (http://en.wikipedia.org/wiki/Bernoulli_distribution)
So 32 difficulty only increase standard deviation by factor of 5.6.
I would have thought that as long as you get at least a few hundred shares a day your PPS variance should be fine. I will work this out later with proof!
So anyone should be fine with 32 diff shares (even with 5 Mhash/sec you will find 100 shares a day).
No, you'd get 100 D1 shares a day, but only 3 D32 shares a day.
Calculating the variance induced hashrate variations is easy since this is a PPS pool. For a 95% confidence interval, just calculate the 0.025 and 0.975 quantiles for a poisson distributed variable with a mean of (in this case) 3. When we do this, the calculated variability in average daily hashrate is between 0 and 30 Mhps. This is a huge daily variation. If you're mining at 5 Mhps, D 32 is way too big.
Yes you are correct - I should have learnt that gut feelings and statistics never mix!
Assuming that the average miner would accept a variation in return such that 1 or 2 days a month they got less than 95% of their expected payout (obviously 1 or 2 days a month they would also get over 105% of their expected daily payout). Then a 50 Mhash/Sec miner should use difficulty 1 shares. And every time you double your mining power you should double the difficulty to keep the same variation.
So a 800Mhash/sec miner should be using difficulty 16 shares.
So as a rule of thumb you should take your Mhash/Sec and divide by 50 to find the difficulty you should use on a PPS pool - assuming you accept the above variation in returns and want to reduce the load on the mining pool servers.
See https://s3.amazonaws.com/bitcoinstatus.rowit.co.uk/other/share+difficulty.xls for details.
Make sense to everyone?
I think the CIs you calculated should show variation over a day rather than a month - people track their daily earnings closely. Or maybe weekly at most. Monthly is just too long a time frame for most people.
You will need to use Poisson probabilities instead of the CLT.
Otherwise, good idea.
August 31, 2012, 08:54:26 PM
Does the difficulty need to be a power of 2? It seems like I'm having just as many shares accepted as a normal difficulty 1 pool.
It needs to be an integer but doesn't need to be a power of 2. However, I wouldn't be surprised if some not very smart mining software didn't handle it correctly and did something stupid. I've only tested with cgminer and it works very well.
August 31, 2012, 08:48:12 PM
Does the difficulty need to be a power of 2? It seems like I'm having just as many shares accepted as a normal difficulty 1 pool.
It needs to be an integer but doesn't need to be a power of 2. However, I wouldn't be surprised if some not very smart mining software didn't handle it correctly and did something stupid. I've only tested with cgminer and it works very well.
August 31, 2012, 03:32:43 PM
Bitcoin difficult still seems on the rise if anything, so I would just knock that up the variance in submitting a few shares quicker than expected in an hour, than you did last time.
Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
Variance at diff 32 is 32^2 times greater, i.e. 1024 times... so in my opinion, diff 32 is too high for current mining hardware in light of that (see my earlier post).Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
No - variance increase with mean not square of mean. And of more relevance to users is the Standard Deviation which is the Square Root of variance. (http://en.wikipedia.org/wiki/Bernoulli_distribution)
So 32 difficulty only increase standard deviation by factor of 5.6.
I would have thought that as long as you get at least a few hundred shares a day your PPS variance should be fine. I will work this out later with proof!
So anyone should be fine with 32 diff shares (even with 5 Mhash/sec you will find 100 shares a day).
No, you'd get 100 D1 shares a day, but only 3 D32 shares a day.
Calculating the variance induced hashrate variations is easy since this is a PPS pool. For a 95% confidence interval, just calculate the 0.025 and 0.975 quantiles for a poisson distributed variable with a mean of (in this case) 3. When we do this, the calculated variability in average daily hashrate is between 0 and 30 Mhps. This is a huge daily variation. If you're mining at 5 Mhps, D 32 is way too big.
Yes you are correct - I should have learnt that gut feelings and statistics never mix!
Assuming that the average miner would accept a variation in return such that 1 or 2 days a month they got less than 95% of their expected payout (obviously 1 or 2 days a month they would also get over 105% of their expected daily payout). Then a 50 Mhash/Sec miner should use difficulty 1 shares. And every time you double your mining power you should double the difficulty to keep the same variation.
So a 800Mhash/sec miner should be using difficulty 16 shares.
So as a rule of thumb you should take your Mhash/Sec and divide by 50 to find the difficulty you should use on a PPS pool - assuming you accept the above variation in returns and want to reduce the load on the mining pool servers.
See https://s3.amazonaws.com/bitcoinstatus.rowit.co.uk/other/share+difficulty.xls for details.
Make sense to everyone?
August 31, 2012, 12:48:20 PM
Does the difficulty need to be a power of 2? It seems like I'm having just as many shares accepted as a normal difficulty 1 pool.
August 31, 2012, 11:17:00 AM
Bitcoin difficult still seems on the rise if anything, so I would just knock that up the variance in submitting a few shares quicker than expected in an hour, than you did last time.
Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
Variance at diff 32 is 32^2 times greater, i.e. 1024 times... so in my opinion, diff 32 is too high for current mining hardware in light of that (see my earlier post).Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
No - variance increase with mean not square of mean. And of more relevance to users is the Standard Deviation which is the Square Root of variance. (http://en.wikipedia.org/wiki/Bernoulli_distribution)
So 32 difficulty only increase standard deviation by factor of 5.6.
I would have thought that as long as you get at least a few hundred shares a day your PPS variance should be fine. I will work this out later with proof!
So anyone should be fine with 32 diff shares (even with 5 Mhash/sec you will find 100 shares a day).
No, you'd get 100 D1 shares a day, but only 3 D32 shares a day.
Calculating the variance induced hashrate variations is easy since this is a PPS pool. For a 95% confidence interval, just calculate the 0.025 and 0.975 quantiles for a poisson distributed variable with a mean of (in this case) 3. When we do this, the calculated variability in average daily hashrate is between 0 and 30 Mhps. This is a huge daily variation. If you're mining at 5 Mhps, D 32 is way too big.
August 31, 2012, 11:10:12 AM
Bitcoin difficult still seems on the rise if anything, so I would just knock that up the variance in submitting a few shares quicker than expected in an hour, than you did last time.
Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
Variance at diff 32 is 32^2 times greater, i.e. 1024 times... so in my opinion, diff 32 is too high for current mining hardware in light of that (see my earlier post).Variance can be quiet abit higher than predictable diff 1 shares, when doing diff 32 shares.
No - variance increase with mean not square of mean. And of more relevance to users is the Standard Deviation which is the Square Root of variance. (http://en.wikipedia.org/wiki/Bernoulli_distribution)
So 32 difficulty only increase standard deviation by factor of 5.6.
I would have thought that as long as you get at least a few hundred shares a day your PPS variance should be fine. I will work this out later with proof!
So anyone should be fine with 32 diff shares (even with 5 Mhash/sec you will find 100 shares a day).
August 31, 2012, 11:05:03 AM
Just put red cellophane over the screen.
August 31, 2012, 11:02:29 AM
So far I really like the pool, my only complaint is that the web site is kind of goofy looking. Makes it hard to look at while at work - people will think I'm f&*cking around. I may change my mind about running solo when the ASICs arrive.
August 31, 2012, 10:46:53 AM
I think you're right. Even if you could put it in cgminer, the person using cgminer it shouldn't have to dynamically change D with the pool's hashrate, or make sure the correct variable are entered to make it happen.
The pool knows its hashrate and reward method, so it should be possible to set a D | hashrate relationship that will not affect daily earnings for any miner by more than a certain percentage. Someone will need to figure out some sort of simplified table of variance for pools though.
The pool knows its hashrate and reward method, so it should be possible to set a D | hashrate relationship that will not affect daily earnings for any miner by more than a certain percentage. Someone will need to figure out some sort of simplified table of variance for pools though.
August 31, 2012, 10:13:27 AM
Sorry, I was replying to the part I bolded above. You could mine here just as well without asics just by reduce D to below 32.
I've been experimenting with d2, 4, 8, 10, 16, 20 and 32.
All work fine, the variance is abit wild with 32 on my CM1, it all works out over 24 hours of course, it's no bother. It's just funny to see it say I'm doing 2.1Gh/s sometimes. I believe it be more stable on for higher performance system like ASIC, that is all.
I've been mining quiet happily on diff4 recently. Might keep it there or go back to 8. 8 matches the number of chips, seemed like a nice thing to do.
*waits for someone to point out that makes no mathematical sense*
I just realised that being a PPS pool, the variance will bother you much less than at a pool that has pool based and reward methos based variance.
ckolivas, that's something might need to be taken into account with cgminer - the variance of the reward method. It starts to get complicated very quickly though, and I'm not sure there'd be an easy way to solve it.
August 31, 2012, 10:09:39 AM
Sorry, I was replying to the part I bolded above. You could mine here just as well without asics just by reduce D to below 32.
I've been experimenting with d2, 4, 8, 10, 16, 20 and 32.
All work fine, the variance is abit wild with 32 on my CM1, it all works out over 24 hours of course, it's no bother. It's just funny to see it say I'm doing 2.1Gh/s sometimes. I believe it be more stable on for higher performance system like ASIC, that is all.
I've been mining quiet happily on diff4 recently. Might keep it there or go back to 8. 8 matches the number of chips, seemed like a nice thing to do.
*waits for someone to point out that makes no mathematical sense*
I just realised that being a PPS pool, the variance will bother you much less than at a pool that has pool based and reward methos based variance.
ckolivas, that's something might need to be taken into account with cgminer - the variance of the reward method. It starts to get complicated very quickly though, and I'm not sure there'd be an easy way to solve it.
Jump to: