[20 gh] NMCBit pool 3% fee prop 6.6% PPS

[DeathAndTaxes]

How much variance one should see in 24 hours I am not sure.  I think there is something beyond standard variance just be aware variance does exist. 

I admit, I am definately aware of it....but just not +/-20% worth using PPS (or so I would hope....lol)

Yeah I am not sure what the standard deviation would be over say 1 hour, 6 hours, 24 hours, 7 days, 30 days, etc.  If someone is interested they could write up a simple simulator.  I am sure there is a way to prove it mathematically but for me writing up a command line simulator seems easier.

[1714001643]

1714001643

[1714001643]

1714001643

[1714001643]

1714001643

[1714001643]

1714001643

[bitlane]

I am sure there is a way to prove it mathematically but for me writing up a command line simulator seems easier.

Then I will bow out and give way to the skillz  ....because anything more than a simple bat file or spreadsheet and I will be in WAY over my head...LOL

[DavinciJ15]

Mining for a bitcoin you a variance has high as 20% in ether direction.  Thus your test needs to be carried out over a long period of time to be valid as one pool you can have many shares found fast and another be unlucky and find shares slower.

Doesn't that kinda defeat the sole purpose of PPS mining ? I don't see how a variance of up to 20% would ever come into play, especially with PPS and the fact that current mining software has built-in work-unit timeouts (60secs by default in CGMiner). I just don't see it.....

I will run my test if I find nothing I will assume it's a latency issue. Can you give me the ping time for my pool and the comparison pool?

Unfortunately I am unable to ping anything on/inside your domain. No reply, all timeouts....obviously security on your end.

However, quick ping stats to the 'other' server (with a bit of anonymity to keep this from looking like a You vs. Them competition):

Pinging xxx.xxx [1xx.x0.xx.1xx] with 32 bytes of data:
Reply from 1xx.x0.xx.1xx: bytes=32 time=94ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=91ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=95ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=89ms TTL=55

Ping statistics for 1xx.x0.xx.1xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 89ms, Maximum = 95ms, Average = 92ms

Hope that helps, atleast a bit.....as again, your servers do not respond to my pings.

I have enabled it try it now.  My time out is time=23.661 ms from my house.

[DavinciJ15]

How much variance one should see in 24 hours I am not sure.  I think there is something beyond standard variance just be aware variance does exist. 

I admit, I am definately aware of it....but just not +/-20% worth using PPS (or so I would hope....lol)

Yeah I am not sure what the standard deviation would be over say 1 hour, 6 hours, 24 hours, 7 days, 30 days, etc.  If someone is interested they could write up a simple simulator.  I am sure there is a way to prove it mathematically but for me writing up a command line simulator seems easier.

The 20% deviation is over 2016 shares.

[DavinciJ15]

I am sure there is a way to prove it mathematically but for me writing up a command line simulator seems easier.

Then I will bow out and give way to the skillz  ....because anything more than a simple bat file or spreadsheet and I will be in WAY over my head...LOL

You are not over your head we all have to start at the beginning.  You learn every day someone is kind enough to spend the time and teach you.  Most people don't.

I appreciate everything you have done to point out this issue to me, most people would have said nothing and left the pool.  You took the time to inform me of a problem and it means a lot to me.

Thank you.

Davinci

[johnj]

Davincij,

I currently use your pool as a backup pool. Although the end fee is kinda high (~10%), you've put a lot of hard work in and I don't mind showing my support for good folk like you.  I understand the 7%pps is for your protection from variance, and the 3% is because shit ain't free.

That being said, what are the odds of you switching to a smpps type payout?

[DavinciJ15]

Davincij,

I currently use your pool as a backup pool. Although the end fee is kinda high (~10%), you've put a lot of hard work in and I don't mind showing my support for good folk like you.  I understand the 7%pps is for your protection from variance, and the 3% is because shit ain't free.

That being said, what are the odds of you switching to a smpps type payout?

SMPPS gives you the same thing as proportional since the your payment per share moves up and down with just not as much as proportional.  All it does is smooths out proportional payments and thus what's the point go prop?

I'm asking because I don't see the attraction.

[DeathAndTaxes]

Davincij,

I currently use your pool as a backup pool. Although the end fee is kinda high (~10%), you've put a lot of hard work in and I don't mind showing my support for good folk like you.  I understand the 7%pps is for your protection from variance, and the 3% is because shit ain't free.

That being said, what are the odds of you switching to a smpps type payout?

SMPPS gives you the same thing as proportional since the your payment per share moves up and down with just not as much as proportional.  All it does is smooths out proportional payments and thus what's the point go prop?

I'm asking because I don't see the attraction.

PROP = I have too many Bitcoins I enjoy being robbed blind by poolhoppers (speaking as an ex-poolhopper).
SMPPS = I want to keep fees low so I accept slightly higher volatility but still want my fair share.

[bitlane]

I have enabled it try it now.  My time out is time=23.661 ms from my house.

I actually have a lower PING to you, than the other Pool I used......

-----------------------------------------------------------------------
coinserver4.nmcbit.com
Reply from 107.20.142.218: bytes=32 time=79ms TTL=51
Reply from 107.20.142.218: bytes=32 time=86ms TTL=51
Reply from 107.20.142.218: bytes=32 time=81ms TTL=51
Reply from 107.20.142.218: bytes=32 time=76ms TTL=51

Ping statistics for 107.20.142.218:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 76ms, Maximum = 86ms, Average = 80ms
-----------------------------------------------------------------------
My 'other' Pool (repost from previous page)
Reply from 1xx.x0.xx.1xx: bytes=32 time=94ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=91ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=95ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=89ms TTL=55

Ping statistics for 1xx.x0.xx.1xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 89ms, Maximum = 95ms, Average = 92ms

[johnj]

Davincij,

I currently use your pool as a backup pool. Although the end fee is kinda high (~10%), you've put a lot of hard work in and I don't mind showing my support for good folk like you.  I understand the 7%pps is for your protection from variance, and the 3% is because shit ain't free.

That being said, what are the odds of you switching to a smpps type payout?

SMPPS gives you the same thing as proportional since the your payment per share moves up and down with just not as much as proportional.  All it does is smooths out proportional payments and thus what's the point go prop?

I'm asking because I don't see the attraction.

Ahh, I may be using the wrong term - I was under the impression that SMPSS all but removed variation.  Such as Luke-Jr's pool.

The attraction I was aiming for was if somehow we could reduce that 7% from the pure pps system.

But perhaps I'm misunderstanding how those other payout methods work.  In either case I'll support your pool.

[bitlane]

PROP = I have too many Bitcoins I enjoy being robbed blind by poolhoppers (speaking as an ex-poolhopper).

LOL.... Nice one.

It's like the jelous boyfriend who is actually the one cheating on his girlfriend....hehe....not the other way around.

[DeathAndTaxes]

Ahh, I may be using the wrong term - I was under the impression that SMPSS all but removed variation.  Such as Luke-Jr's pool.

No it is just different meanings for the word variance.

You can have value variance.  Different shares have different values.  PROP is the worst.  Short block = high value per share.  Long block = low value per share.  That variance is what makes pool hopping profitable (at the expense of 24/7 miners).  Remember mining is a zero sum game.  If someone is coming out ahead then someone else is coming out behind.

You can also have payout variance.  The timing of payouts (not value varies).  With SMPPS each share is worth the same but it payouts may have variance is size and timing.

        Value Variance     Payout Variance
PROP    HIGH                    HIGH
SMPPS  LOW                    LOW
PPS      LOW                    LOW

[DavinciJ15]

NMCBIT has upgraded to Merged Mine Version of PoolServerJ.

Shadders has identified an issue with cgminer
https://bitcointalk.org/index.php?topic=28402.msg597203#msg597203
I would like to let everyone know how I have chosen to handle the partial-stales created by cgminer.
NMCBIT.COM will not count the partial-stales mostly caused by cgminer for now, however, once the issue gets fix if it's not fixed already I will then be counting the partial stales as valid for the NMC shares only and showing up as a stale on BTC shares.

This will motivate miners to upgrade if they wish to continue Merged Mining.

[DavinciJ15]

Ahh, I may be using the wrong term - I was under the impression that SMPSS all but removed variation.  Such as Luke-Jr's pool.

No it is just different meanings for the word variance.

You can have value variance.  Different shares have different values.  PROP is the worst.  Short block = high value per share.  Long block = low value per share.  That variance is what makes pool hopping profitable (at the expense of 24/7 miners).  Remember mining is a zero sum game.  If someone is coming out ahead then someone else is coming out behind.

You can also have payout variance.  The timing of payouts (not value varies).  With SMPPS each share is worth the same but it payouts may have variance is size and timing.

        Value Variance     Payout Variance
PROP    HIGH                    HIGH
SMPPS  LOW                    LOW
PPS      LOW                    LOW

Well then maybe I will implement smpps as well but it does not save you from pool hoppers as they can jump on when the smpps is high and jump away when it's low.  To be honest I am not sure how it's implemented but if I where to guess I would start at even money PPS and zero reserves if there was good block like 50% dif then there would be 50% in the smpps reserves.  If there where string of bad blocks and the reserves are over a set limit per smpps worker it would increase the pps by some percentage point.  If there where string of bad blocks cleaning out the reserves and making it go negative smpps would go down until it was not lower than current PPS.

That would be my guess but looking at it from that point of view it's no different than proportional for a miner staying at one pool.  Thus it would be advantagous to jump away once smpps is low.

Unless I am missing something.

[DavinciJ15]

I have enabled it try it now.  My time out is time=23.661 ms from my house.

I actually have a lower PING to you, than the other Pool I used......

-----------------------------------------------------------------------
coinserver4.nmcbit.com
Reply from 107.20.142.218: bytes=32 time=79ms TTL=51
Reply from 107.20.142.218: bytes=32 time=86ms TTL=51
Reply from 107.20.142.218: bytes=32 time=81ms TTL=51
Reply from 107.20.142.218: bytes=32 time=76ms TTL=51

Ping statistics for 107.20.142.218:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 76ms, Maximum = 86ms, Average = 80ms
-----------------------------------------------------------------------
My 'other' Pool (repost from previous page)
Reply from 1xx.x0.xx.1xx: bytes=32 time=94ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=91ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=95ms TTL=55
Reply from 1xx.x0.xx.1xx: bytes=32 time=89ms TTL=55

Ping statistics for 1xx.x0.xx.1xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 89ms, Maximum = 95ms, Average = 92ms

Rats! I was hoping that would explain it.

[johnj]

Well then maybe I will implement smpps as well but it does not save you from pool hoppers as they can jump on when the smpps is high and jump away when it's low.  To be honest I am not sure how it's implemented but if I where to guess I would start at even money PPS and zero reserves if there was good block like 50% dif then there would be 50% in the smpps reserves.  If there where string of bad blocks and the reserves are over a set limit per smpps worker it would increase the pps by some percentage point.  If there where string of bad blocks cleaning out the reserves and making it go negative smpps would go down until it was not lower than current PPS.

That would be my guess but looking at it from that point of view it's no different than proportional for a miner staying at one pool.  Thus it would be advantagous to jump away once smpps is low.

Unless I am missing something.

If the poolhoppers are going to hop you, they're gonna hop the prop side.

So any changes to the pps side would be irrelevent to hoppers.

But i'm not an expert in these kinds of things, just want to brainstorm on how to reduce that 7% pps fee, that's all.

You've done a lot of work recently so don't let me put more on your plate.

[DutchBrat]

Hey Davinci,

any problems with rewarding the BTC-pps on the last round ?

My balance hasn't gone up even though there's no (*) behind the last BTC block and it's been well over 5 hours ago since the round was concluded

Thanks

Brat

Never mind, it just showed up in my balance   edit: it appears this was for the last  block, so still no payment for this block:

2011-11-02 02:23:54   1,922,701   38   20:28:11 hours

[DavinciJ15]

Hey Davinci,

any problems with rewarding the BTC-pps on the last round ?

My balance hasn't gone up even though there's no (*) behind the last BTC block and it's been well over 5 hours ago since the round was concluded

Thanks

Brat

Never mind, it just showed up in my balance   edit: it appears this was for the last  block, so still no payment for this block:

2011-11-02 02:23:54   1,922,701   38   20:28:11 hours

What is your username I will check.

If the poolhoppers are going to hop you, they're gonna hop the prop side.

So any changes to the pps side would be irrelevent to hoppers.

But i'm not an expert in these kinds of things, just want to brainstorm on how to reduce that 7% pps fee, that's all.

You've done a lot of work recently so don't let me put more on your plate.

I can implement it but it will add a new dimension to hopping as hoppers can hop the highest smpps.  More options means more miners and I will exclude transaction fees from smpps that will be more revenue.  You will then get that 7% back because there is no risk to me your take on all the risk.  Personally PPS is the best way to go because it insures the pool is working to maximize it's block find rate at all times as it's like I am paying you for your hashing power and I must use it the most efficient way possible.

[DavinciJ15]

Hey Davinci,

any problems with rewarding the BTC-pps on the last round ?

My balance hasn't gone up even though there's no (*) behind the last BTC block and it's been well over 5 hours ago since the round was concluded

Thanks

Brat

Never mind, it just showed up in my balance   edit: it appears this was for the last  block, so still no payment for this block:

2011-11-02 02:23:54   1,922,701   38   20:28:11 hours

So I checked it and my jaw dropped you are correct you where not paid!  Your shares are in the db you did 20979 shares and 44 rejects.  Not sure how this failed as there are 2 tests to insure that everyone gets paid and they failed.
I have to stop payouts until this is debugged.   You will get paid for those shares thanks for pointing it out.

[DavinciJ15]

Still working on it but turns out no one got paid for that block.