Topic: ASICMINER: Entering the Future of ASIC Mining by Inventing It - page 416. (Read 3917543 times)

Has 2 phase immersion cooling existed for decades? I think we will see a revolution as soon as the big players like google realize how much money/space/heat savings they would get from a little 2 phase liquid.

Coolermaster, Intel, and I'm sure many other companies are already working on prototypes/concepts for 2 phase. I think it is only a matter of time before it becomes the norm unless a significant drawback is discovered.

Computer chips that have needed cooling have existed for decades, and immersion cooling has existed for decades, and immersion cooled systems have already been used where it makes sense to do so. I'm not seeing any particular reason why the there is suddenly going to be an imminent revolution in how cooling is done for the majority of datacenter applications.

You don't need to drain the tanks. You can simply pull out the malfunctioning board.

Training shouldn't be too different from a normal datacenter.

A few % of savings may not seem worth it yet while the technology is still immature but once it is refined and perfected there will be no such thing as air cooled datacenters (no large ones at least)

Id like to see a better picture and macro shot of that chip.

Immersion cooling might make sense for a super reliable set-up and forget type system. But for a real datacenter with essentially prototype equipment? People need to go in there to perform maintenance, replace aging/malfunctioning units, diagnose errors. Ease of use, accessibility, having a setup more familiar to a wider range of personnel with only standard training/experience are all of great value. How long does it take to drain the tanks if someone needs to go in and fiddle with one of the boxes? Or do you leave malfunctioning units alone to not disturb the rest of the system, and if so, what's the cost of that over time? What kind of training do personnel need to safely work with these systems and what is the cost of these personnel relative to those who can work with a standard data center?

In short, immersion cooling might be of interest in specific applications such as military, security agency, and some of the largest companies where maximum performance density trumps everything else and billions of dollars are available for specialized facilities, procedures, and personnel. But for a typical start-up company that has to worry about capital costs, is juggling personnel among all kinds of tasks that they are learning on the fly, and is working with prototype systems that don't have years of reliability heritage already established... not worth it.

People seem to forget all the complications associated with working with non-standard prototype systems and just look at a few % savings in electricity cost, but that's not nearly the whole picture.

FC stated he needed to address PR and that hasn't occured. From there on I don't get the impression anything has derailed from the plan which offered three courses: selling, mining, and franchising; the greater emphasis of which is going to be based on best return at the time production runs were projected to hit the street. I am hopeful a high percentage will go to mining and that a plan is put in place to create a sustainable revenue stream for the company. Regardless the company should be able profit now that Gen3 is establshed and show an income. By comparison there is no income to be had from BTC sitting in a wallet.

If you go back through this thread from the origins of AM the nexus of issues are PR related so what we are currently experiencing with respect to PR is not unprecidented.

O/T From their twitter account it looks like RM got the 4 X ASIC board stuffed and are testing.

Yeah, you are absolutely right.
Sell me all your shares right now, I'll take that risk for you.

What's your point? It's obviously a strategic partnership that they both stand to profit from.

Well, you asked if AM is "officially building any mining farms", and the answer is "Yes, they are officially doing so." I imagine that after they are officially finished building it, it will officially be put to some use.

Engineering samples sent out to customers doesn't count as "results"? In fact, there have been plenty of "results" and FC actually keeps us relatively updated on the progress and is pretty transparent about the operations. Three weeks may be longer than you (and I, for that matter) would like to wait between updates, but considering the pace at which these things move it's not too bad.

In that case... Stop whining and sell your shares already.  Or do you just come on here saying a lot of stupid angry things in a lame attempt to drive the price down?

maybe everyone is busy optimizing before they come out with another statement on speed on chips......

Partnership? It was only one project and ASICminer was a customer. Now AC can sell their technology to anybody. No share from profit for ASICminer, its completely different company.

Promises. And what after that?


Because I see no results and no communication about whats happening? No news were always bad news here.

A partnership.

They have (repeatedly) said that the HK datacenter will be used.

If he hasn't been communicating, how can you conclude he hasn't been doing anything?

What has ASICminer in common with immersion cooling from Allied Control other than one small room filled with "proof-of-concept" which is anyway in every day loss right now? Is ASICminer officially building any mining farms ? No. Unofficially behind our backs? Maybe. What is crucial is timing. And I see only delays and empty promises over last 6-8 months here ! Timing is everything and Friedcat is very slow. He knew like 6 months about tape-out timeline, he should have everything in place ... and now, when chips are finally out he is unable to do anything for 3 weeks? And no communication with shareholders? What the... ? So much for "overpromise and overdeliver" ... yeah. 

Even if you could get the building for free and managed to acheive an impressive air cooled PUE of 1.15 you would still be better off with immersion cooling.

You would get your money back in only a year from electricity savings alone.

Iceland, Cheep electricity... cold climate.  If I were to set up a new major mining operation, I would go to iceland.

As ridiculous as it may sound I would be looking at cold war leftovers in the form of abandoned hangars and silos.

Googles mega expensive, highly engineered, sea water cooled datacenter in finland has a PUE of more than 1.06 so knc must have a PUE of around 1.15 where as immersion cooling can be as low as 1.01

What seems like a better choice:

Option A
10 x $100,000 1MW containers @ 1.01 PUE

Or

Option B
1 x giant $5,000,000 10MW aircraft hanger @ 1.15-1.3 PUE

The beauty of mining is it can be situated anywhere there's network access and power. 30% or far more can be saved on where it's situated.  That also is offset by lower startup costs.

10 times more space and 30% extra electricity is not cheap.

Which really is a function of where you want to situate your data center and that typically would be established as a function of energy costs, climate and real estate costs. I believe it was already established that immersion cooling was not a cost effective solution so it's interest other than novelty would have one conclude it has little bearing on future AM operations. If you already have tanks full of Novec use it but IMHO KNC took the more cost effective approach.