Pages

▼

Sunday, 15 September 2013

September 15th 2013 weekly pool and network statistics


Welcome, miners.

Changelog:
  • Nil
Usual pools missing from results:
  • Nil
Errors:
  • "Unknown" in pie chart. The total number of blocks solved by the network this week was 1352, yet the total number of blocks solved by the constituent pools was 1362. I think this is because one of the pools is not using UTC to publish their stats and so their published blocks solved is offset compared to the network. I'll see if I can figure out which one by next week.

Pools with coinbase signature:


Pool hopping:
  • Nil.

1. BTCGuild luck
Is not unusual at all this week. I think I've got the calculation bug beat!

2. Network hashrate and forecasting
The average weekly network hashrate has continued on its exponentially increasing path. I've been working on forecasting methods since my last post on the subject and I've been testing several that have been so far much more accurate than my previous method. I'll post an update when I'm confident that I have a reliable method.

3. The effect of GHash.IO on other pools
I'm seeing many complaints about the increased variance experienced by many pools at the moment. If you look at figure 3, you'll notice that from a few weeks before I started recording GHash.IO's hashrate, all pools except BTCGuild (and Ozcoin, not on there tho week) suddenly lost - and in many cases continue to lose - a significant proportion of the network hashrate. As difficulty continues to sky-rocket, pools who cannot maintain a consistent proportion of the network will experience increased variance.

It's amazing to think that much of this increased variance can be attributed to just one pool (or hashing entity, I'm not sure about who owns the hashrate there just yet).

As usual, please post comments if there's anything you don't understand, with which you disagree, or just think is wrong.

The charts

Table: Table of all pools with public data and their various statistics averaged for the last seven days - for smaller pools the average may be more or less than seven days, depending on number of blocks solved for the week. Network hashrate and that of some pools are estimates, the upper and lower 95% confidence interval bounds are included.
Figure 1: Pie chart of the percentage of network blocks hashrate by pool. "Unknown" combines those pools for which I can't scrape statistics, solominers and private pools. The percentage of network hashrate will only be approximate since the exact network hashrate is unknown.
Figure 2: Chart of network hashrate, hashrate of the largest mining pool, combined hashrates of the three largest mining pools, and a line representing 50% of the network hashrate. Handy if you're worried about 51% attacks. The upper and lower 95% confidence interval bounds for the network hashrate are in between the shaded areas.
Figure 3: Chart of chronology of pool hashrates, averaged per week.
Figure 4: Chart of average hashrates per pool per round for the week, and per 144 rounds for the network. The upper and lower 95% confidence interval bounds for the network hashrate are in between the shaded areas.
Figure 5: Chart of chronology of negative binomial CDF probability of shares submitted and blocks produced for the week.
Figure 6: Chart of chronology of round length divided by difficulty, averaged per week.
Figure 7: Chart of hashrate vs round length for hoppable pools (the larger the hashrate increase at the start of a round, the larger the loss to strategic miners).
Figure 8: Chart of pool user hashrate distribution. Note that for some pools this average is over twenty four hours, some pools are averaged over an hour or more and some for only fifteen minutes, so expect some variance in the results.


























Thanks to blockexplorer.com for use of their network statistics.

organofcorti.blogspot.com is a reader supported blog:

12QxPHEuxDrs7mCyGSx1iVSozTwtquDB3r
<weeklypoolstatistics>
at

No comments:

Post a Comment

Comments are switched off until the current spam storm ends.