ASICMINER: Entering the Future of ASIC Mining by Inventing It

[dexX7]

Just FYI:

Immerson cooling is awesome! DeathAndTaxes's thread is great, to get some background knowledge: https://bitcointalk.org/index.php?topic=313087

The technology is already there, but as far as I know it's rarely used. www.allied-control.com and www.grcooling.com and are leading suppliers - if AM really partnered with Allied Control, it's even greater!



One advantage is a very high power density as shown in the picture above. To quote DAT:

What types of power densities are possible?
3M has conducted experiments cooling 4KW heat loads using 1L of working fluid so in theory heat densities approaching 4,000 W/L are possible.  The constraint on commercialization is that existing servers have relatively low energy densities (well low relative to the limits of immersion cooling).  Even a high end 3U server (4 CPU, multiple GPUs, 4+ 1200W PSU) may only have an energy density of 100W per Liter.    However SHA-256 ASICs have very high energy densities although current systems have server like energy densities due to the limits of air or water cooling.  Take a look inside the case of any 2nd gen ASIC design what takes up the most space?  Air.  The actual ASIC boards are very energy dense however there are surrounded by a significant amount of empty space.   Remember these are using boards designed for air/water cooling.  It may be possible to improve energy density by making custom compact boards.

[bitcoin.newsfeed]

Meanwhile no block found in last 40hours.

https://blockchain.info/blocks/ASICMiner

[SmiGueL]

100TH ASiCMiNER ChainForker[/center]

Looks like AM took you seriously

Whaha I was just gonna quote that!

If this is real, that's awesome news. _{*The blades look pretty real though.}

[Zubilica]

It seem FC is looking too far into the future

[JimiQ84]

Meanwhile no block found in last 40hours.

https://blockchain.info/blocks/ASICMiner

Now is probably good time to switch back to pool mining

[mutex]

You are misreading. Submersion cooling doesnt magically make the chips more efficient

Actually, it does. All else equal a cooler chip can work at lower voltage due to lower current leakage.

But I don't know what the actual increase in efficiency would be in the case of BEs.

[jimmothy]

It doesn't reduce the chip prices at all but what it does do is make them much more efficient in the long run. This means that if the difficulty skyrockets many asics will cost more in electricity to run than the amount of btc they produce. Since AM uses less electricity it can operate at a higher difficulty and still churn out a profit.

You are misreading. Submersion cooling doesnt magically make the chips more efficient, it makes the cooling potentially much more efficient. If you dont need AC, cooling is just the fans and maybe 2% of the total power consumption, reducing that by "97%" means you drop overall power consumption by, well almost 2%.

If you compare it to AC, it all depends on what AC system you compare to. Savings there can be substantial, easily 50% if you assume inefficient domestic AC systems (a lot less for typical heatpump cooling in a datacenter), but thats still nowhere near enough to make a chip that uses 1000% more W per GH suddenly competitive.

Not too mention the fact anyone can buy submersion cooling, its not like AM invented and patented it. For instance:
http://www.grcooling.com/

I don't know what you mean by 1500 gen1 asics do you mean the usb asics? The AM cube is nearly as efficient as a Knc Jupiter and that is with 144nm vs 28nm.

Efficient? Now you confuse the market price with efficiency and cost. AM cant charge more per GH than KnC. Thats not good for AM, thats BAD.  Because the fact is their current chip produces 0.3GH where the competition provides 100+GH and soon 400+GH per chip. You really think AM can produce their chips 1200x times cheaper than their competitors? Of course not, so soon they can no longer sell anything based on that chip at a positive margin.

First of all it does "magically" make the chips more efficient because they don't need a massive fan hogging up all the electricity. When you consider how much energy is spent keeping the chips cool, a 97% reduction would have an enormous effect on efficiency and power density.

Secondly, I think you are confusing knc rigs with chips. There is no such thing as a 400gh chip. Also by 0.3gh I assume you mean the usb which is by far the least efficient asic by asicminer.

Thirdly both asicminers new cube and kncs fastest miner have nearly the same ROI. This means when asicminer does get 28nm chips like knc they will be much more efficient due to manufacturing costs being lower and immersion cooling.

[Puppet]

Actually, it does. All else equal a cooler chip can work at lower voltage due to lower current leakage.

But I don't know what the actual increase in efficiency would be in the case of BEs.

You are correct, but that effect is so marginal, particularly for bitcoin asics running at 100% load, so dynamic power is so much more important than leakage, it can be safely ignored in this context.

[Puppet]

First of all it does "magically" make the chips more efficient because they don't need a massive fan hogging up all the electricity.
When you consider how much energy is spent keeping the chips cool, a 97% reduction would have an enormous effect on efficiency and power density.

How much do you think the low speed fans on a KNC Jupiter consume ? 2W per fan? Total of 8W. 12W if you count the case fans which KnC users disconnect because they (weirdly) get better performance at higher temperatures. 12W out of a total of ~650W at the wall is 1.58%. Thats not massive in my book, thats statistical noise.

Secondly, I think you are confusing knc rigs with chips. There is no such thing as a 400gh chip.

Yes, I was confusing with hashfast which claim 400GH per chip. Not that the comparison with KnC's 100GH chip looks so good.

Also by 0.3gh I assume you mean the usb which is by far the least efficient asic by asicminer.

They all use the same chip. Its not because a blade contains 32 asics that the per asic performance is somehow better, nor that 32 asics would cost less than 32x the price of 1 asic.

Thirdly both asicminers new cube and kncs fastest miner have nearly the same ROI.

Who cares? Someone buying it may care, but as a sharesholder dont you understand your company is selling 96 asics (+PCB, assembly, etc) for ~$500 ? <$5 per chip whereas hashfast is charging $1500 for a module.  How much do you think these chips cost to produce?

This means when asicminer does get 28nm chips like knc they will be much more efficient due to manufacturing costs being lower and immersion cooling.

Hu?

[antirack]

How much do you think the low speed fans on a KNC Jupiter consume ? 2W per fan? Total of 8W. 12W if you count the case fans which KnC users disconnect because they (weirdly) get better performance at higher temperatures. 12W out of a total of ~650W at the wall is 1.58%. Thats not massive in my book, thats statistical noise.

Try putting 10, 50, 100 or 1000 of them in a room and you will see what "statistical noise" sounds and feels like.

[Puppet]

Try putting 10, 50, 100 or 1000 of them in a room and you will see what "statistical noise" sounds and feels like.

Sure, but it would still only be 1.58% of the power consumption as long as you dont need AC. Which of course isnt likely, but if you followed the thread, that was my point.

[Dexter770221]

Difficulty growth seems to be faltering finally.

It's just a little silence before storm. It will flatter in 50PH range....

[asdfghj]

One thing no one has mentioned and i would like to hear opinion on.

With immersion cooling (I expect) the temperature is constant, deviating only slightly from the boiling point of the liquid.  Air is different because its low heat conductance, the more btu output the hotter the chip gets because it can't be carried away quickly.  With such high thermal densities being tested above this won't be a problem.

So what will being able to lock the chips temperature in at a cool 40C be?
(Marginally?) More efficient?
Over clocking potential?

The writeup mentions a longer timeline for large scale production but maybe a consumer device (enclosed cube) with immersion cooling is doable and possibly improves w/GH? Wishful thinking?

Also, could this rescue a gen2 design which had poor thermal properties?

[Rival]

I think the most overlooked statement is that there are currently no commercial data centers available that are designed to to handle the requirements of of a world-class farm. And definitely none that can scale to the farms of the future.

The brick wall that AM hit was not an inability to create chips and boards. It was finding a place to put them. The datacenters of today were not designed for the sort of high-heat high-density equipment that make bitcoin mining profitable. They were built for commercial servers and hard drive arrays... very different. Trying to build a mining farm can quickly overwhelm the resources and capacity of a datacenter built for common blade servers. Expansion starts to get really expensive when you count electricity, cooling, and especially infrastucture and even the building itself.

So he solved it.

Building? Poof, it's gone. Replaced by a $3000 shipping container. Cooling? 97% of costs eliminated. Not just high density of chips, but high density of boards. The very definition of optimization.

AM is doing for farms what ASIC technology did to video cards. Making anyone using generic facilities outdated, under-performing, uncompetitive, and obsolete.

He created ASCF: Application-Specific Computing Facility. (Coined here first!)

FC is so far ahead of the curve he is beyond the horizon.

[binaryFate]

Building? Poof, it's gone. Replaced by a $3000 shipping container.

What are you referring to? Franchising? Or did I miss something?

[hlynur]

Building? Poof, it's gone. Replaced by a $3000 shipping container.

What are you referring to? Franchising? Or did I miss something?

i think he's referring to this:

Not sure if this has been posted before. Some updates from AM:
https://drive.google.com/folderview?id=0ByWHHc0u_thNSWhHS18xM2ZXSUU

[muyuu]

Difficulty growth seems to be faltering finally.

It's just a little silence before storm. It will flatter in 50PH range....

[binaryFate]

Building? Poof, it's gone. Replaced by a $3000 shipping container.

What are you referring to? Franchising? Or did I miss something?

i think he's referring to this:

Not sure if this has been posted before. Some updates from AM:
https://drive.google.com/folderview?id=0ByWHHc0u_thNSWhHS18xM2ZXSUU

Thanks! I had missed that... 

[chairforce1]

Meanwhile no block found in last 40hours.

https://blockchain.info/blocks/ASICMiner

Block mined today.

[Puppet]

yeah just a $3000 container and some  bottles of fluorinert. lol
for the record, similar solutions exist commercially, and are sold (in containers) for $2.per watt.  that's more affordable than a brick and mortar datacenter but the price of the container itself doesn't really matter

I would also dispute that datacenters suitable for mining  don't exist  the requirements for hpc and gpu compute are quite similar.