MrTeal
Legendary
 Offline
Activity: 1274
Merit: 1004
|
 |
March 02, 2015, 03:10:59 PM |
|
I believe there is an error in the pool statistics table on the mining pool page for Eligius' hash rate.
I have a question about the CDF vs luck included in that table. Luck is a straightforward to calculate, but it seems counter-intuitive that the CDF for GHash would be higher than for Kano.is, even though the luck was worse for Kano. I'd guess there is a factor in there that accounts for the increased hashrate of GHash leading to less variance and a tendency to stick closer to 1.0 luck, but I'd be interested to see how you calculated the CDF for that.
|
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
March 03, 2015, 04:07:43 AM
Last edit: March 03, 2015, 09:24:34 AM by organofcorti |
|
I believe there is an error in the pool statistics table on the mining pool page for Eligius' hash rate.
I'll check it out. What do you think it should be? Edit: You're right, it's wrong, but it's the data I got from the API - it might have been down for a while. The data should be ok by next week. I have a question about the CDF vs luck included in that table. Luck is a straightforward to calculate, but it seems counter-intuitive that the CDF for GHash would be higher than for Kano.is, even though the luck was worse for Kano. I'd guess there is a factor in there that accounts for the increased hashrate of GHash leading to less variance and a tendency to stick closer to 1.0 luck, but I'd be interested to see how you calculated the CDF for that.
Good question, and it comes up a lot. The CDF is a function of mean(submitted shares)/(expected shares) and of number of blocks solved. The more blocks solved, the narrower the CDF should be. You can play with it here: In this example, the number of blocks solved is 100, and the mean(submitted shares/expected shares) is 1.1: http://www.wolframalpha.com/input/?i=CDF%5BErlangDistribution%5B100%2C+100%5D%2C++1.1%5DGenerally, use CDF[ErlangDistribution[nblocks, nblocks], mean(submitted shares/expected shares)] |
|
|
|
MrTeal
Legendary
Offline
Activity: 1274
Merit: 1004
|
|
March 03, 2015, 02:52:06 PM |
|
I believe there is an error in the pool statistics table on the mining pool page for Eligius' hash rate.
I'll check it out. What do you think it should be? Edit: You're right, it's wrong, but it's the data I got from the API - it might have been down for a while. The data should be ok by next week. I have a question about the CDF vs luck included in that table. Luck is a straightforward to calculate, but it seems counter-intuitive that the CDF for GHash would be higher than for Kano.is, even though the luck was worse for Kano. I'd guess there is a factor in there that accounts for the increased hashrate of GHash leading to less variance and a tendency to stick closer to 1.0 luck, but I'd be interested to see how you calculated the CDF for that.
Good question, and it comes up a lot. The CDF is a function of mean(submitted shares)/(expected shares) and of number of blocks solved. The more blocks solved, the narrower the CDF should be. You can play with it here: In this example, the number of blocks solved is 100, and the mean(submitted shares/expected shares) is 1.1: http://www.wolframalpha.com/input/?i=CDF%5BErlangDistribution%5B100%2C+100%5D%2C++1.1%5DGenerally, use CDF[ErlangDistribution[nblocks, nblocks], mean(submitted shares/expected shares)] Thanks, I knew I was missing something. It's not the average CDF of the block which makes a lot of sense, since that essentially is the same information of the average luck and wouldn't actually tell you how likely that level of luck is. With Bitcoin mining, the generation of a block is an exponential distribution; so from a quick Wikipediaing of the Erlang distribution, for a stochastic process like Bitcoin mining the expected luck (inclusive of orphaned blocks) is 1.0. Then for 100 blocks, your rate is 100, and k is chosen to give you the mean value of 1 (so 100 as well). |
|
|
|
|
|
|
Sine(X)
Member
Offline
Activity: 118
Merit: 10
|
|
March 09, 2015, 05:18:14 AM |
|
Still more bad luck for GHash.IO, plus a DDOS this week, which has caused a precipitous drop in hashrate
Ghash lost 12Phs, BW Pool added 12 Phs. Looks like transit Phs, doesn't it? |
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
March 09, 2015, 05:29:18 AM |
|
Still more bad luck for GHash.IO, plus a DDOS this week, which has caused a precipitous drop in hashrate
Ghash lost 12Phs, BW Pool added 12 Phs. Looks like transit Phs, doesn't it? I still don't know anyone mining at BW Pool, only cloud mining. |
|
|
|
MrTeal
Legendary
Offline
Activity: 1274
Merit: 1004
|
|
March 09, 2015, 02:49:05 PM
Last edit: March 09, 2015, 03:15:08 PM by MrTeal |
|
As an electrical engineer, the graph of Eligius' hashrateprofitability makes me happy on the inside.
|
|
|
|
|
kano
Legendary
Offline
Activity: 4620
Merit: 1851
Linux since 1997 RedHat 4
|
|
March 09, 2015, 03:10:43 PM |
|
As a bitcoin enthusiast, the graph of kano.is is so random it could only be a graph related to block finding  |
|
|
|
Sine(X)
Member
Offline
Activity: 118
Merit: 10
|
|
March 15, 2015, 12:34:05 PM |
|
As an electrical engineer, the graph of Eligius' hashrateprofitability makes me happy on the inside.
As a Sine(X) I also like this graph  |
|
|
|
|
|
|
|
|
|
abracadabra
|
|
April 01, 2015, 12:51:42 AM |
|
Eclipse and Ghash haven't hit a "lucky streak" since October.  |
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
April 01, 2015, 01:01:49 AM |
|
Eclipse and Ghash haven't hit a "lucky streak" since October.
Eclipse is not getting many blocks so their unluckiness in that period of time is not unlikely. I've been in talks with GHash, trying to figure out what's going on there. |
|
|
|
|
abracadabra
|
|
April 01, 2015, 01:03:31 AM |
|
Eclipse and Ghash haven't hit a "lucky streak" since October.
Eclipse is not getting many blocks so their unluckiness in that period of time is not unlikely. I've been in talks with GHash, trying to figure out what's going on there. Gotcha.. what's your take on Ghash?  |
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
April 01, 2015, 01:05:22 AM |
|
Eclipse and Ghash haven't hit a "lucky streak" since October.
Eclipse is not getting many blocks so their unluckiness in that period of time is not unlikely. I've been in talks with GHash, trying to figure out what's going on there. Gotcha.. what's your take on Ghash? Don't know yet. As soon as I'm sure I know what's going on, I'll post. |
|
|
|
Sine(X)
Member
Offline
Activity: 118
Merit: 10
|
|
April 07, 2015, 12:32:39 PM |
|
I think BitClub doesn't have 34Phs.. They resolved 2 blocks Apr02 and then only 1 block Apr07
|
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
April 07, 2015, 12:47:19 PM |
|
I think BitClub doesn't have 34Phs.. They resolved 2 blocks Apr02 and then only 1 block Apr07
So it was probably just very good luck, or they might have hired some machines so they could begin with a bang. |
|
|
|
|
|
organofcorti (OP)
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
|
April 26, 2015, 02:41:39 PM |
|
|
|
|
|
|
|
|
