Weekly pool and network statistics

[os2sam]

I would use the "7 day window" line on the graphs at http://bitcoin.sipa.be (Pieter Wuille's site

I had known about this site but never got around to bookmarking it.  Got it in my list now.
Thanks,
Sam

[organofcorti]

24th February 2013 weekly pool and network statistics

Welcome, miners.

Last week my estimate of BTCGuild's hashrate was a slight underestimate. BTCGuild pool op eleuthria kindly made a quick adjustment to the block history on the website, so this week's estimate is correct (within the listed 95% confidence interval). BTCGuild now has 25% of the network, a percentage not seen for any pool since DeepBit in July last year

As well as ASICMiner continuing to add hashes to BTCGuild, Avalon ASICs began to appear on various pools.

Ozcoin's hashrate has constantly increased during the week, and only Ozcoin and 50BTC have had any singnificant increase in the network. Other pools have had steady hashrates, but because the overall network has increased so much their percentage of the network has declined.

This was likely to happen - each new Avalon ASIC increases the variance in hashrate between pools until each pool has a steady increase in hashrate from ASICs. I estimate that by May 2013 the network hashrate will increase by an order of magnitude or more - so right now all pools are starting with approximately the same hashrate.

This would have been a good opportunity  for very small pools to make sure they have variable difficulty and local work available, and to offer rebates to ASIC miners. So far only Ozcoin, Eligius, BTCGuild and 50BTC seem to have attracted significant numbers of the new ASICs, but that may change.

Ozcoin did overtake Deepbit this week, and Coinotron is still down - apparently an ongoing DDoS.

The pie chart shows a significant drop in the "unknown" hashrate (sources of network hashes other than pools included in the weekly summary). The hashes from the unknown sources do not seem to be using ASICs, so perhaps they have mostly been botnets mining their own bitcoind? Or a dark pool for botnets? (I'm making this up as I go, but some reader will have a better idea about this than I).

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

You can view this weeks charts at http://organofcorti.blogspot.com/2013/02/24th-february-2013-weekly-pool-and.html

You can view all previous charts and other posts and fun things at http://organofcorti.blogspot.com

[organofcorti]

3rd March 2013 weekly pool and network statistics

Welcome, miners.

Top 10 pools which are "ASIC ready" (in order of size):

BTCGuild
Slush's pool
Ozcoin
EclipseMC
Bitminter
Itzod
p2Pool


Top 10 pools that have had an increase in percentage of the network hashrate this week (in order of size):

BTCGuild
Slush's pool
Ozcoin
EclipseMC
MTRED


There are still only a small number of ASICs running, and these are having a disproportionate affect on the network and on the pools hosting them. I expect all pools that are "ASIC ready" to displace all the "not ASIC ready" pools from the top 10.

MTRED is a bit of a mystery - where is it's increase in hashrate coming from? Has redditorex actually made it "ASIC ready" just just not updated the MTRED website? Or are the GPU miners that are attempting to mine the last few coins at current difficulty jumping on to the only zero fee PPS pool?

The "unknown" portion of the pie continues to reduce. It will be ineresting to see if that reduction accelerates as difficulty increases, or if the solominers and dark pools upgrade to ASICs. If they don't they were probably botnets, and we'll all be better off without them.

BitMinter's drop in hashrate is a bit of a shock. This is a pool that belongs in the top 5 but has seen a constant reduction in hashrate since their own DDoS experience a while back. I hope DrHaribo is able to turn this around. Well, I'm sure he will, since BitMinter is one of only a few "ASIC ready" pools. And in other DDoS casualty news, Coinotron is still down and under DDoS attack. This must be some kind of record, and my best wishes go out to the Coinotron team.

Ozcoin gets the "Quiet achiever" award for February. The hashrate at Ozcoin has increased consistently since well before ASICs were being used on the pool, and their percentage of the network averaged at 50% more than DeepBit for the past week. If hashrates continue to change as they have for the last four weeks, Ozcoin could be the second largest pool in the network within a month.

If DeepBit is unable to improve their "ASIC ready" status, EclipseMC or BitMinter may replace them in the top 5. A year ago, when some miners were making their stange "Boycott DeepBit" demands, no one had any idea that things could change so drastically in twelve menths.

Finally - look at BTCGuild's percentage of the network history. It's actually much greater than that - more like 30% of the network. However the pool has had poor luck this week, which affects the method I use to estimate the hashrate. BTCGuild pool op eleuthria and I are working on a way to make this an accurate result rather than an estimate with confidence intervals.

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

You can view this weeks charts at http://organofcorti.blogspot.com.au/2013/03/3rd-march-2013-weekly-pool-and-network.html

You can view all previous charts and other posts and fun things at http://organofcorti.blogspot.com

[Graet]

Thanks again ooc
Next week's statistics will be interesting after ~1.5 more terahash of asics joined the pool this morning
Interesting times

[organofcorti]

10th March 2013 weekly pool and network statistics

Welcome, miners.

Changelog:

BTGGuild pool op eleuthria provided me with a fancypants new API last week, which means that not only is the calculated hashrate now accurate (now confidence interval required) but I can also calculate the pool luck, CDF, shares per round / Difficulty, and chart the intraweek hashrates. All have been included this week.

Since HHTT has become a top ten pool (in terms of percentage of network hashrate) I spent some time including them this week. The hashrate has a confidence interval since the current data is insufficient to calculate the pool's luck, and I'm not certain that my script for it wont break as the HTML for the site doesn't seem to be standard. Be prepared for HHTT to be missing every now and then.


The "Predictions that came true" department
From last week's post:
The hashrate at Ozcoin has increased consistently since well before ASICs were being used on the pool, and their percentage of the network averaged at 50% more than DeepBit for the past week. If hashrates continue to change as they have for the last four weeks, Ozcoin could be the second largest pool in the network within a month.
This has happened somewhat sooner than I expected - Ozcoin miners now pool the second largest amount of network hashing, after BTCGuild. A large number of ASICs joined them over this past week. The top three pools now gather more than 50% of the networks total hashes.

The sources of "unknown" hashes has declined even further, down to 10% of network blocks solved this week. Once difficulty is increased to the point that GPU mining is prohibitive, the "unknowns" will represent only solominers and dark pools without botnets - so I expect this percentage to drop even further over the coming weeks.

Deepbit is being hopped a little, and Slush's pool is being hopped is a way that hasn't occurred for a long time. Although I haven't checked, I wouldn't expect this to have a very significant effect on miner earnings.

MTRED is actually using variable difficulty, but I haven't found any details about it on the MTRED website yet so I'm not sure they're "ASIC ready".

Top 10 pools which are "ASIC ready" (in order of size):
BTCGuild
Ozcoin
Slush's pool
Bitminter
EclipseMC
HHTT
Itzod


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

You can view this weeks charts at http://organofcorti.blogspot.com/2013/03/10th-march-2013-weekly-pool-and-network.html

You can view all previous charts and other posts and fun things at http://organofcorti.blogspot.com

[organofcorti]

Thanks again ooc
Next week's statistics will be interesting after ~1.5 more terahash of asics joined the pool this morning
Interesting times

It certainly was an interesting week

[eleuthria]

Thanks again ooc
Next week's statistics will be interesting after ~1.5 more terahash of asics joined the pool this morning
Interesting times

It certainly was an interesting week

You haven't seen anything yet.  This month is going to be one hell of a ride.  Avalons are starting to show up in greater numbers.  At the very least it will be fun to see the Top 10 pools shuffle around in rank as the hash power finds new homes.

[Graet]

Interesting is an understatement lol

[organofcorti]

17th March 2013 weekly pool and network statistics

Welcome, miners.

Changelog:
Nil

Pools missing from results:

Eclipse Mining Consortium - API not working.
p2pcoins.net,  Mkalinin - these pools were hacked and the owner has taken them off-line.


"ASIC ready" pools:

BTCGuild
Ozcoin
Slush's pool
Bitminter
EclipseMC
HHTT
Itzod
Bitparking


Pool hopping:

Slush's pool


Lots of drama this week. The hard fork had a significant effect on the midweek hashrate, so the retarget difficulty was a little less than estimated.  Figure 4 shows how pool's hashrates were affected by the downtime.

Earlier in the week while BTCGuild pool op eleuthria was upgrading a merged mining server to bitcoin 0.8.0, a bug allowed miners to credit their accounts with far more bitcoin than they were due. This personally cost eleuthria 1250 btc - about US $60 000 at current exchange rates. So far only one miner has returned any of the coins. CryptoCluster returned the 47.8 btc (US $2 300) without even a request email - that's certainly one way to increase your standing in a community that is mostly anonymous. On the other hand, a miner (whose details eleuthria has not released) responded to eleuthria's polite request to return the overpayment with scorn. I hope we learn his username, since I wouldn't want to do business with such a community member.

Eclipse Mining Consortium was DDoS'd for much of the week. I'm not sure whether his response to this has affected the API was using for his pool data, but I expect to have it fixed for next week. This means the "Unknown" section of the pir chart is about 1.5 Thps larger than it would be if Eclipse had been included. This means that "Unknown" is down to about 6%. It appears not much solomining mining is being done by ASICs.

Both Eligius and BitMinter overtook Deepbit this week, and thanks to some hefty ASIC power added Eligius' hashrate jumped to 5 Thps for a day or so.

Bitcointalk forum member gyverlb posted an excellent guide to mining on p2Pool. Since mining on p2Pool can net you more coin than mining on other centralised pools, this post is important reading for any geeky miner.

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

You can view this weeks charts at http://organofcorti.blogspot.com.au/2013/03/17th-march-2013-weekly-pool-and-network.html

You can view all previous charts and other posts and fun things at http://organofcorti.blogspot.com

[DrHaribo]

Since mining on p2Pool can net you more coin than mining on other centralised pools, this post is important reading for any geeky miner.

I think you meant to say that since true zero-fee pools like BitMinter, MtRed and P2pool will pay better in the long term, you may want to look into that. And on BitMinter you don't need to be geeky, just click a button.

[organofcorti]

Since mining on p2Pool can net you more coin than mining on other centralised pools, this post is important reading for any geeky miner.

I think you meant to say that since true zero-fee pools like BitMinter, MtRed and P2pool will pay better in the long term, you may want to look into that. And on BitMinter you don't need to be geeky, just click a button.

Absolutely, on both points. Hence the "can" - p2Pool can net you more if you set the donation to the dev lower than whichever pool at which you mine currently. For p2Pool to run well, you need to be a bit sysdmin-ish, (or be prepared to learn), and there are other geek requirements. I wouldn't recommend it to someone (like me) who wants to mine and just check once a day that everything is copacetic.

I personally find the idea of low maintenance, easy to manage solution (like yours) more appealing, but for miners who love fiddling with their set ups, p2Pool could be a good choice. I have other things with which I prefer to fiddle (mind out of the gutter, DrH!).

[DrHaribo]

... everything is copacetic.

Word of the day? I had to look up that one.

[organofcorti]

... everything is copacetic.

Word of the day? I had to look up that one.

It's an interesting word - I heard it first when I was little. It was in some Seventies song (anyone name it for me? I 've forgotten).

I liked the word even more when I found out no one knows exactly what the etymology of the word is - it's the creature of the black lagoon of Americanisms.

[ldrgn]

Do pools publish information about the number of miners submitting work?  It seems like this would be an important number to estimate as decentralization is key to Bitcoin's security.  It would also be interesting to see how many people drop out when ASICs hit.

[organofcorti]

Do pools publish information about the number of miners submitting work?  It seems like this would be an important number to estimate as decentralization is key to Bitcoin's security.  It would also be interesting to see how many people drop out when ASICs hit.

It's possible to get the information, but hard to do synchronously since for most pools you have to request the data from the pool ops. I have actually been working on something along those lines for the pre-ASIC era, so keep and eye out for it.

[DrHaribo]

Number of active users mining at BitMinter is available at http://bitminter.com/api/pool/users

We went from 900 to 1500 active users over the last two months.

[organofcorti]

Number of active users mining at BitMinter is available at http://bitminter.com/api/pool/users

We went from 900 to 1500 active users over the last two months.

Well now, that's handy. But if you're worried about decentralisation, the number of miners isn't as important as the number of miners plus as the hashrate distribution of the miners. That's what is hard to get synchronously, and since some pools average the hashrates for 15 minutes and others for 24 hours, it's very hard to compare the pools. That's why it's taking a while.

[DrHaribo]

Well now, that's handy. But if you're worried about decentralisation, the number of miners isn't as important as the number of miners plus as the hashrate distribution of the miners. That's what is hard to get synchronously, and since some pools average the hashrates for 15 minutes and others for 24 hours, it's very hard to compare the pools. That's why it's taking a while.

I have data on how much work each user did in each shift and each block round. I know the duration also so I can use that to calculate each user's average hashrate during a shift or round. It gets a little crazy for 1-second rounds though.

I'm not sure how to massage this into some useful metrics to make available through the API.

Perhaps if you let us know what you need, pools can deliver.

[organofcorti]

Well now, that's handy. But if you're worried about decentralisation, the number of miners isn't as important as the number of miners plus as the hashrate distribution of the miners. That's what is hard to get synchronously, and since some pools average the hashrates for 15 minutes and others for 24 hours, it's very hard to compare the pools. That's why it's taking a while.

I have data on how much work each user did in each shift and each block round. I know the duration also so I can use that to calculate each user's average hashrate during a shift or round. It gets a little crazy for 1-second rounds though.

I'm not sure how to massage this into some useful metrics to make available through the API.

Perhaps if you let us know what you need, pools can deliver.

It's a bit chicken-and-egg. Hard to know exactly what the data format should be until I've analysed it, and it's hard to know how to analyse it if the data format varies from pool to pool.

But from what I've seen so far, I'd need the 24hr average D1 hashrates, or proofs of work with average difficulty attached. The miners don't have to be in order or named, but it does need to be the total hashrate per account rather than per miner.

I wouldn't expect pools to go to the hassle of creating an API just for that until I can show some use for the data. Hopefully once I finish my current work and post it there'll be enough interest from enough pools that I'll be able to include the weekly pool user hashrate distributions in the weekly pool stats.

[Graet]

pool users and workers are already in our API.
but
we had 287 users and 800 or so Gh a while back
right now we show
Pool: [ 4,858.78 GHash/s | 300 Users | 775 Workers and 1 user with 2.8Th/s
and a couple of days ago 7Th with 309 users

personally I don't think user numbers are a good indicator currently unless you also look at user hashrates.