12.4 Miner hashrate distributions 28th May 2013

28th May 2013

Recommended reading: 
12.1 The pre-ASIC network hashrate distribution
12.2 Pool and network miner hashrate distributions
12.3 Miner hashrate distributions 14th May 2013

Changelog:

• Everything! A complete change the way I think it's easiest for most people to visualise data. Also, I was running out of room for the sequential violin plots

Errors:

• From Mid April until 27th May Itzod's miner count is significantly inaccurate, although the grouped hashrates will be essentially unaffected (see below for details).

0. Introduction
Although the violin plots used previously were quite good at showing trends, they weren't so good for showing clear changes to total pool hashrates and total numbers of miners. As well as this, the violins were not a good method to present an open ended chronological series.

So instead I grouped user hashrates and plotted by time:

1. The number of user accounts in that group.
2. The total hashrate attributable to that group.

All the charts in this post represent this, using various colours to represent various groups of hashrate. Think of them as a layer cake, building upwards to a total number of miners or total hashrate and to the right over time.

In doing this I gained new insight into how users' hashrates were changing, and also noticed a problem with Itzod's miner count - it seemed to top out at 500 miners. On investigation I realised this was a problem with the way I'd been scraping the data which is now fixed.

1. Change in miner count and grouped hashrates by pool.

Each coloured layer in the charts below represents either the number of miners (left plot) or the total hashrate contributed by that group (right plot) - ASICMiner solo will of course only ever have a miner count of 1.

Some trends are evident:

• Nearly all miners have a hashrate less than 5 Ghps, but user accounts with more than 5Ghps contribute far more to the total hashrate of each pool.
• The amount of very low hashrate miners at BitMinter has decreased and the amount of hashrate contributed by high hashrate miners has increased significantly, and the same is true for BTCGuild.
• The number of low hashrate miners at Eligius has increased.
• At Ozcoin and HHTT, miners with hashrates greater than 5 Ghps make up more of the user base than at other pools.
• 1 - 4.99 Ghps miners at P2Pool, Itzod and Polmine contribute more to the total hashrate than any other group.

I'll leave it to the reader to find the other trends visible in the charts.

It should be noted that the data from the end of January to mid April is interpolated. I have no data point in that time frame. After that point data points are weekly (or every few days).

2. Change in miner count and grouped hashrates, all pools combined.

By combining data for all the available pools (and now ASICMiner's solomining efforts), we get an idea of how the network as a whole might appear. 

Since I only have a sample of the network, the total hashrate and total miner count isn't important. However percentage due to each hashrate group should be similar to that of the entire network, and that is what the first plots below show. 

Applying the percentages in the "Percentage of total hashrate" to the known network hashrate results in the final plot "Summed hashrates as a proportion of the network hashrate". The "Unknown proportion is due to the lack of data from the end of January to mid April.

• The percentage of miners at lower hashrates has decreased since late April.
• The 50 to 99 Ghps contribution to the network has increased only a little since April, consistent with only a few new Avalons being added since then.
• The 10 to 49.9 Ghps contribution to the network has increased slightly - possibly due to the 10 Ghps blades from ASICMiner.
• The percentage of 1 to 4.9 Ghps user accounts has increased consistently - people still purchasing  GPUs?

3. Summary
This update was mostly to present and explain the new method of charting data, and allow readers to understand it before I move to estimating the network's electricity use. If you have any questions, please post a comment.



The data in the combined plots from section 2 is in the table below,  and also available as pastes:
Number of miners in sample at hashrate
Total hashrate in sample at hashrate
Percentage of total hashrate in sample at hashrate

organofcorti.blogspot.com is a reader supported blog

BTC:  12QxPHEuxDrs7mCyGSx1iVSozTwtquDB3r
LTC:  LPXnETNoCBr16GduvyWRzFP83rZNeEgMuB