Topic: White Paper: Analysis of Large-Scale Bitcoin Mining Operations (Read 1795 times)

There are 6 mini-containers (tanks) inside each of the large ones. In the tanks there is a 19" rail. Your gear can be as small as 1U.

For very small setups it's probably easier to stick with air cooling?


I posted this in the AM thread earlier:

They used to be, but no longer. They stagnated a lot in upgrading their line, so their market share got pushed down considerably by all the new players with 28nm chips.

Thanks for sharing!

I agree that containerized solutions are the way to go similar to the direction of non-BTC implementations. However, I think there is room to challenge traditional DC design with BTC given the back-office and decentralized nature of BTC mining. 


Maybe there is room for smaller container solutions without the need for extensive redundancy

thanks milan, welcome back!

Seems to be from here:
http://cex.io/redeem

I am not sure actually hmmm.

Hi Antirack.

Nice to see you active here again

I saw your site. Where is this picture from? Didn't see that one before.

Damn this is good. Going to link the whitepaper from my new site: BitcoinMiner.net

Should have seen this earlier, would have saved so much time writing the articles on large scale mining :S

Are you reading the same document?

There is a difficulty chart on page 2, it says the difficulty will go up steep, estimates 175PH when ASICMiner gen3 is out in a few months and states that AM alone plans to dump 400 to 1600 PH/s on the market in 2014.

Difficulty will go up no doubt.

Isn't something that costs less to build and saves you money to run good for you in that case?


Lower Bitcoin ($500 vs. $900), rising difficulty and more people getting into the mining business.
Isn't this exactly the reason why an infrastructure that wastes time, resources and electricity is not a good idea?


Did you read the same white paper?

Is that why they are now all installing water coolers, refrigerators in cabinet doors, DX heat exchangers, compressors and chillers etc?

To cool 1MW of power electronics costs several (!) $$$ per W. Add to that the time and money you spend to build boxes with fans and radiators.

Immersion cooling costs less than $1 per W and you need nothing but boards and chips and a giant pool with liquid. Make your hardware dense and you can make more savings.

http://www.datacenterjournal.com/it/whats-stopping-liquid-cooling/

http://www.electronics-cooling.com/2014/02/bitcoin-2-phase-immersion-cooling-and-the-implications-for-high-performance-computing/

At high densities, liquid cooling is a necessity. Why not dump all of the gear in a liquid instead of using pumps, compressors, all that.

Supercomputers (Cray etc) and flying airplanes have proven that it works.

Bitcoin mining has proven that it's cheaper.

You may also want to watch this, they talk about all these problems and how this needs to be solved:

https://www.youtube.com/watch?v=JkBu5_hZgWI

Interesting white paper... barely mentions difficulty though, just in a couple of references. I think a lot of plans were hatched for large mining farms when BTC hovered around the $800-$900 mark, now with BTC in the $500, the ROI for these large farms is being stretched out.  Time to recoup an investment vs difficulty over time have serious implications for large mining operations.  While the focus on the paper was more on cooling, I think immersion cooling is a waste of money. Traditional air cooling seems to be doing just fine.

"1.2MW immersion cooled mining cluster with 6 tanks 200kW each tank"

Sounds good to me, heading over to the bank for a loan now...

Here's a new white paper about large scale mining ops and immersion cooling.

White Paper: Analysis of Large-Scale Bitcoin Mining Operations
http://www.immersion-cooling.com/publications/Analysis_of_Large-Scale_Bitcoin_Mining_Operations.pdf