Estimating/Visualizing total mining network power consumption?

[finack]

Has anyone already done this or attempted it? With the dramatic rise in hashing power that has been added to the network I am curious what that might amount to in terms of raw power consumption.

I understand you'd have to make a lot of educated guesses or estimates, but:

Bitcoin charts gives a current number of 3.390 Thash/s

Just to get us started maybe we could assume a general network efficiency of 50%-75% of a 5850? Or if anyone had an opinion about how better to guess at that.

Then we'd need a good all in number of watts for something stable like 1 Ghash. Maybe the combination of a few decent size efficient mining setups could arrive at a number like that that included the whole room power, meaning fans, power supply loss, cpus, gpus, all in.

Anyone else interested in what the total usage might be? I'm just curious if it's the size of a small datacenter, a large datacenter, a small part of a power plant, etc.

EDIT:

Maybe I can make this simpler. Would 2 or 3 efficient miners mining in the say 3-50 Ghash range be willing to post their hash rate and total all in power consumption to maintain it? Along with a short description of what is running and if AC is involved?

[1715555063]

1715555063

[k]

I've seen some numbers in other threads before estimating this.

This site has a lot of data on the different hardware including J/MHash etc.
https://en.bitcoin.it/wiki/Mining_hardware_comparison

The 5850 card you mention seems be ~1.85 Mhash/J or ~1.85 Mhash/s per W. I think I've seen other people use a very rough number like 1 Mhash/s per W - probably not a bad estimate when you consider all the different types of hardware, AC/cooling costs etc.

Using 1 Mhash/s per W, we get 0.024 kWh/day per Mhash
3.390 Thash/s = 3.390*10^6 Mhash/s = (3.390*10^6 Mhash/s *0.024 kWh/day per Mhash) = 81360 kWh/day = 81.36 MWh/day

[finack]

I've seen some numbers in other threads before estimating this.

[...] 81.36 MWh/day

Thank you very much for taking a stab at a number, that's exactly what I was hoping for. Now I've tried to compare this to stated outputs of power plants, and I'm pretty sure I'm running into problem with time.

When a power station reports a certain number of megawatts or gigawatts of output, is it assumed that is per day? Or is there some other scale being used?

https://secure.wikimedia.org/wikipedia/en/wiki/Power_station#Typical_power_output

Large coal-fired, nuclear, and hydroelectric power stations can generate hundreds of Megawatts to multiple Gigawatts. Some examples:

The Three Mile Island Nuclear Generating Station in the USA has a rated capacity of 802 megawatts.
The coal-fired Ratcliffe-on-Soar Power Station in the UK has a rated capacity of 2 gigawatts.

Gas turbine power plants can generate tens to hundreds of megawatts. Some examples:

The Indian Queens simple-cycle peaking power station in Cornwall UK, with a single gas turbine is rated 140 megawatts.
The Medway Power Station, a combined-cycle power station in Kent, UK with two gas turbines and one steam turbine, is rated 700 megawatts.

80 megawatts / day = 10% of the rated capacity of three mile island nuclear generating station? I would have expected much, much less power being used so I think there is a comparison problem.

[smooth]

When a power station reports a certain number of megawatts or gigawatts of output, is it assumed that is per day? Or is there some other scale being used?

No they are talking about power output (continuous rate) not energy output (sum of power over time).  Watts are a measure of power.  Watt-hours are a measure of energy.  A 1 GW power plant operating continuously produces 24 GWh/day.

[finack]

[...]A 1 GW power plant operating continuously produces 24 GWh/day.

Makes much more sense, thank you.

So for visualization, with our very rough numbers:

3.390 Thash/s = [...] = 81.36 MWh/day

and

Ratcliffe-on-Soar power station is a coal-fired power station [with] a [peak] capacity of 2,000 MW or 48,000 MWh/day

So 81.36/48000 = 0.001695 or 0.17% of this:


or breaking it into renewable units:

wikipedia says: "The power generated by a large wind turbine is of the order of 1 or 2 megawatts."

so 24-48 MWh/day, or 2-4 of these:



Which seems pretty minor, all told. Does anyone have any interesting pictures of data centers with their known power usage? Or tighten up the numbers?

[matt.collier]

http://www.youtube.com/watch?v=BzBCbO8U45w

You might want to fast-forward to 6:30.

In this video, the miner says his 1 Ghash machine use 1 kW/hr, so 24 kW per day.

So, roughtly 1 Thash = 24,000 kW * 3.39 = 81,360 kW = 81.36 MW per day.

Which is exactly the number given by kirian.

[benjamindees]

Oh christ the wrong units, please make them stop.  Don't they teach physics where you people live?

"kW per day" is not a sensical unit

"kW/hr" is not a sensical unit

"MWh/day" technically makes some sort of sense but should be avoided

Kirian's answer is what I would estimate, on the order of 3-4 megawatts.  Could be higher depending on how many CPU miners there are.

For comparison, enough to power a mid-sized college:  http://www.energy.rochester.edu/us/list.htm

[k]

you're right maybe I should have just given the power number in W rather than MWh/day , so ~3.39 MW.

People are used to the unit of energy in kWh from their electricity bill, so units of power in kWh/day make some sort of intuitive sense.
There is a book I like called "Sustainable energy without the hot air" which has a good explanation of units of energy and power and the author explains his reasoning for using different units to explain things. http://www.inference.phy.cam.ac.uk/withouthotair/c2/page_24.shtml. The whole book is available for free by the way.

[kiba]

http://bitcoinweekly.com/articles/the-wasted-electricity-objection-to-bitcoin-part-ii

I believe Vitalik Buterin have an excellent article on this.