Topic: Analysis of Realistic $ Earnings From Butterfly Labs Mining Hardware (Read 17082 times)

And now it is 40,300,030,328

omg.

Great table, creates some useful perspective.

Good. At 64 million according to that table, I make 0.008BTC a day per GH. A 50gh miner therefore makes 0,4BTC in 24 hours, making me 12 BTC a month, meaning it goes break even in... 2 months.

Come on guy, what the heck are you complaining about?

Of course, being first to market would be more awesome, making 70BTC right now, breaking even in 14 days. I get that. But come one... 2 months 100% ROI. I rest my case here... and I believe in 30-40 million actually, not necessarily 60.

BFL won't ship 10.000 50GH/s ASICs and Avalon have shipped between 300 and 600 65GH/s ASICs.

Assuming BFL did ship 10K ASICs, that would still put us only in the 60-70 million range, according to the "Ripple table".

Not factored in of course. However not designed yet and I doubt delivered before February.

^ Good post, interesting. But what about newer ASICs which are faster, cheaper, and more efficient in terms of mh/joule?

Let me argue the case why these numbers are completely bonkers.

These numbers assume an 18% increase over the former month, steadily. This is an exponentially growing function.
Bitcoins are generated in a steady fashion, meaning that the network adjust the difficulty to keep the hashrate in line with the time Blocks are mined.

So, that means that if you are mining with 1% of the network and then later, 0.5% of the network, the difficulty will have to rise accordingly to reward you with the now fitting portion of the network rewards, not counting transaction fees.

That means that if you were mining 36BTC at 1% of the network, it would mean you are at 18BTC at 0.5% of the network, per day.

Now, this leads us to the coming table. The table that will consider that if we assume difficulty increases like that, it will correlate largely with the hashrate. The following table assumes an absolutey 100% linear correlation in hash rate and difficulty, which we will find, is the theoretical ideal, just never met because of variance of people starting and stopping their devices. I have used the number from above, because the talks of "Difficulty is 1 billion, get out of Bitcoin" people tended to annoy me.



If you think, people will invest a whopping 1.2 billion dollars in mining equipment and get that started by 2014, by my guest.
If you think all the ASIC companies together will manufacture to the order of 500.000 50gh miners and get them sold, be my guest.
If you think that this is absolutely bonkers, like me, lets assume something more reasonable.

You can take the amount of money invested and pick one of the numbers that feels "right" for you. See how much money is flowing in bitcoin land and in miners' land.

+1

Everyone who's spending 2500+ USD on a mining rig will just hop onto the ASIC bandwagon and it will be business as usual for "serious" miners....I don't think it will take long either (ppl might take longer to switch if ASICs are delivered slowly but they will switch if they intent to stay profitable).

I don't agree that you only call GPU and ASIC a technological step.
All new forms of mining are a technological step.

ASICs are the end station in IC design at the moment. Nothing faster than ASICs.
I think that now that we are on ASICS the growth will start alligning with the usual moore's curves of current tech.
Which means about 2x better every 2 years or so.
We will probably see some more spikes when the ASICs go to smaller processes but nothing shocking after that. Untill quantum computers destroy criytography, that is.

Do you really see Difficulty rising at a linear rate?

Why don't you try calculating Difficulty based on something slightly more predictable, like network hash rate?

ASIC will reduce trust on newcomers, its already hard to explain to a new guy that this is not a scam, try to tell it now you have to buy especialised hardware for it instead of just using any vga card, im sure that the income of new miners will reduce over the time, and swift over to Litecoin and other scrypt based altchains, and BTC will remain in the power the ones that are mining today.

Also there is other sha256 altchains to mine on the 5ghash unit, if i get a 5ghash unit from BFL today ill put it to mine PPC... or solo mine TRC, them exchanging for BTC or even LTC...

Also especially on the 5ghash unit from BFL, thats for ones that cant affort the biggers ones like me, there is not other option, tell Avalon to produce 7200 5ghash units at 200 dolars instead of making stupid 600, 60ghash units at 7000 dolars, Avalon is harming the network more than helping it. And the chips they are selling will not reach market in form of working smaller units until the end of the year...

Not to mention that ASICminer is not helping at all either.

Although organofconti has done a good job of predicting short term difficulty changes, I think it's unrealistic to extrapolate the short term trends far into the future.

As this graph shows, there have been a few different regimes in the historical difficulty trends.

1. Only CPU mining (steady growth)
2. Initial GPU mining (sharp difficulty spike)
3. GPU mining common, FPGA mining available (steady growth)
4. Initial ASIC mining (sharp spike) - The current regime

At each new technological step, the hash rate per dollar spent increased by an order of magnitude. FPGAs don't count as a technological leap, because they only incrementally improved the hash rate per dollar of TCO.

Once ASICs become common, I expect the difficulty growth to level off to a steady rate, until the next major technological leap which increases hash rate by 10x per dollar (or more). ASICs 2.0 will probably not be that leap.

Based on that, I think it's quite reasonable to predict the following regimes:

5. ASIC mining common (steady growth)
6. Initial New Technology (sharp spike)

I think Regime 4 is likely to last through the summer. It's impossible to say how long Regime 5 will last, but I think first generation ASICs will remain profitable for the duration of Regime 5 and into the early stages of Regime 6, in the same way that GPUs from 2011 are still profitable.

The other site seems mostly decent, i was criticizing the three graphs you posted and their obvious deviation from reality.

Actually, you're right in that you can somewhat predict hashrate.
But what your graphs also show is that if the action gets away from the mean your graphs lose predictive power.
Due to all the recent movements now is an important time to know something about hashrate.
And it is exactly now that your graphs are starting to diverge a lot.
It's not difficult to predict that something will continue straight on when it is already going straight.
What makes a model powerfull is when it can predict a change from the usual.
Now that there is turmoil your model becomes increasingly clueless.
How can you take a prediction seriously if it already diverges by 20% from a measured point and only seems to diverge more in the future?
If your model gave you 1% untill now, wouldn't you say that it must be missing something realy realy big for it to give you so much deviation now?

While you both make some good points, I find organo's predictions to be a useful tool. While they have been overshooting the difficulty mark somewhat it's easy to use his projections and then adjust slightly as events unfold. ASICMiner is deploying hashes somewhat slower than they predicted and Avalon batch 1 took longer to get out the door than anticipated. Meanwhile BFL is still BFL.

Read the posts before you criticise. All the data is there, previous errors, etc. The datasets are also available for you to check.

I wrote The four week forecast network hashrate has been quite accurate recently. The average error for the four week forecast for the last 10 ten weeks is 1%, and the absolute error is 5%.

I'm not sure what reality you share, but in mine a four week forecast with error that low is pretty good.

If you're interested in alternate forecasts using estimated ASIC delivery dates and hashrates instead, I have that covered too:
http://organofcorti.blogspot.com.au/search?q=%22ASIC+choices%22-canary

Thats what I can do with my mathy.

Everyting is predictable.
You just need to take into consideration how often the prediction is informative.

For one, what is the error rate of these models?
If you don't know anything about the error you have no way of assessing the quality of the predicion.
Effectively, these graphs are useless.

But is that information better that the information you gain from knowing that BFL hasn't delivered yet?
Because i've been waiting for a couple of days now for the hashrate to flatten while the models above keep thinking it will grow more fast.
And they already overestimate points that are actually measured and disagrees with it by 20%.
So the models say that reality is incorrect according to their predicions. Go figure.

You say quite acurate but in reality (the one you share with other people) the predictions are all overestimating by about 20%.
I mean, i can predict without any mathy that hashrate will increase in the near future. But the question is, by how much and when in time. If your models cannot predict that then they are useless for the purpose.
They don't add any information that could not have been obtained by simply knowing that ASICs wll be delivered.