On edit:  forgot to reduce chip size converting from 55nm to 28nm, the corrected fabrication guestimate is ~$0.25 per GH/s instead of original $1.00 per GH/s.


The silicon itself has a cost.  My data is somewhat old but a 300mm wafer runs ~$5K in large volume (think AMD or NVidia), lets be generous and say for small runs that it is only double that or $10K.  I would guess it is probably higher but lets go with that.  If someone has a better small run 28nm wafer cost let me know. 


300mm wafer has 70,685 mm^2 (Pi*150^2).  Lets look at the most advanced chip with known specs, bitfury.  I am just using bitfury because it is the chip with smallest feature size and highest performance that has actually been demonstrated. The die is estimated at 14.44mm and it gets 2GH/s nominal.  The same chip at 28nm would be 14.44 * (55/28)^2 = 3.74mm.   Now for the cost per GH/s of silicon fabrication the size of the chip doesn't really matter.  This hypothetical Bitfury28 could be produced as lots of 3.74mm 2GH/s chips or fewer 14.44 8GH/s chips.  Regardless the cost per GH/s is roughly the same.

 So an entire wafer (assuming no loss due to incomplete chips or defects) would be 70,685 mm^2 / 3.74mm^2 * 2 GH/s = 37,690 GH/s.  Simply put we are estimating that all the chips on the wafer will have an output of ~37 TH/s.  The real number will be lower because there is wasted space on wafer but for an estimate this is close enough.

At $10K per wafer you are looking at ($10,000 / 37,690) ~$0.26 per GH/s just for the silicon.  You still have the cutting, packaging, testing, defect losses, and shipping costs.  Now that gets you a box of working chips.  You now have to put them into boards so you have BOM (balance of materials), company profit, labor, defects, etc.  If you want to ignore the small stuff look for high capacity high quality DC to DC Power supplies as one expensive component.  I don't see how you are getting assembled ready to user miners at less than $1 per GH/s.*  

Prices will come down but that is just silly.  Now don't take the numbers as gospel but they should provide a ballpark look at silicon costs.

As for Intel that is just a red herring, Intel doesn't use foundries, they build their own fabs for internal use from the ground up so they are getting both the chip designer AND the foundry's share of the profit.  

* This assumes no chip that is radically more efficient (GH/mm2) than competitors, improved process (14/20nm), or larger production runs (think $100M in chips) to bring down fabrication costs.

Probably not.  Silicon wafer's aren't that cheap.  Even in high volume we are talking $5K a piece.   So there will always be a chip production cost.  Unless you get a design which is significantly more efficient (GH/s per mm) at the same process (28nm) you are probably looking at more than that just for chip production cost.  Of course that ignores the BOM, assembly, testing, defects, profit margin, and amortizing the NRE.  I don't know how low chip companies can go but I think $1 per GH/s is like hoping gold will go below $300 an ounce (current minimal production cost).

This isn't legal advice but I would like to point out that the FinCEN guidance isn't the law, it is simply how FinCEN interprets existing law.  FinCEN guidance on virtual currencies is contradictory and breaks the precedent of prior administrative rulings on what is and isn't in the scope of money transmission.   It is possible that eventually someone will get involved in a lawsuit with FinCEN and the guidance merely the playbook on how FinCEN intends to proceed.  If they can't convince an old guy in a robe that their playbook has any merit then guidance is worthless.

FinCEN took too much of a stretch trying to shoehorn this new technology into an existing framework that they covered activities that in the past they had declared the same activities to NOT be covered.  I don't see this highly capricious and vague guidance standing the test of time.

It is a shame because they were not forced to act.  They could have taken their time, they could have revised the MSB statutes to create a new category or revised the regs on foreign currency exchange to become "currency exchange" including virtual currency.  Instead they decided to act rashly using some very tortured legal opinions to blanket cover all fiat to virtual transactions.

The end result is a stiffing of innovation, a loss of productivity, and barriers to entry. The only winners will be lawyers (no offense to Mr. Santori) as it takes years to figure out what hell FinCEN "means" (pretty tough when I am pretty sure FinCEN doesn't know).

Well in theory it certainly can.  It is important to not make that claim because it is just easy for BFL to refute.  Case in point, the 7990 is 375W TDP and it is a "reference" dual slot design using a similar "squirell cage" blower as their photoshop card.  So it is "possible" to have a SHA-2 ASIC which dissipates 350W within the dimensions of a high end 2 slot GPU form factor.  


As a practical matter I doubt BFL will be able to deliver on time, on spec, and on wattage.  Everything would need to be perfect from the logistics, to the claimed performance per watt gains to the thermal specs to the fabrication.  It is running so close to the limit of what is possible that any failure would almost certainly result in costly delays.  Lets put this into perspective, AMD pushed back the launch date of the HD 7990 twice for over six months because they (with decades of GPU design and cooling experience) were running into power/thermal issues.

Simple version: 350W in that form factor is possible but it is close the redline.  Worse the specs are based on simulated chips (which for BFL have been horribly wrong in the past).  If it uses more power you can't simply go bigger because they decided to use a specific form factor.

A smaller issue is that lets say for the sake of argument they can delivery on time, on spec, and on wattage.  Despite the compact dimensions of the card there are going to be real limits on what one system can do. As an experiment if someone wants to plan their future 4U BFL hashing rig

Thermal Testing Load
1x BFL card (350W) ~= 2x7870 (350W)
2x BFL cards (700W) ~= 3x7970s (750W)
3x BFL cards (1050W) ~= 4x7970s (1000W)
4x BFL cards (1500W) ~=6x7970s (1500W)

Take a 4U case (or full tower and seal any top/side/bottom vents.  Drop the cards in and try to cool the rig inside a sealed case only using the cards fans plus front intake fans.  It isn't going to happen.  There is a reason that people starting using extenders and open frame rigs ("ghetto miners").  Trying to remove 1KW or more heat from a closed case is a challenge.  The high energy density of GPU cards (x watts isn't evenly distributed across the case area but is instead concentrated in the GPUs) make it very difficult even with high output fans. So you get all the cost/complexity of the GPU form factor while not really gaining higher GH/u (thats GH per rack unit) density.

Still I am sure BFL will lock down $1M in orders or upgrades within a day and that was the point.  That will improve their cashflow situation and is enough people upgrade (and delay their delivery till "December") they may even make their backorder promise.

I never said it can't be done just that datacenter space is expensive, probably more than you think.  A datacenter has so much power, cooling capacity, and rack space.  They want to sell all three "commodities" in equal amounts so they don't run out of one before the other.  If they give you a higher electrical load per sq/ft that means either they need to sell underpowered racks to someone else OR they run out of power or cooling before they run out of space.  No datacenter operator (who just spent tens of millions building the thing) is going to do that.  It traps them into an underutilized capacity scenario which upsets all the bean counters. The fact that miners are energy dense doesn't really help.  They are going to want you to rent an equivalent amount of space for the power/cooling you need and then distribute your power/heat load across it.  That likely means a lot of expensive empty space.

The rate the datacenter pays for power isn't the whole story.  That class I building cost a fortune to build you are buying power, cooling, floor space, security, connectivity, redundancy, fire suppression, natural disaster resistance, etc.  It doesn't come cheap.  It is some of the most expensive real estate in the world (any data center).  I doubt it is going to be cheaper than space in China.  Maybe it is but I remain unconvinced it is so massively cheaper that giant PH/s operations will be operating at a tiny fraction of the operating cost.  


I didn't think Freidcat was renting space in a datacenter, I always thought (maybe incorrectly) it was just industrial/warehouse space.  Also I don't think ASICMiner is using as much power as the 900KW in the scenario we are responding to.

As a non-lawyer that is the route I would go.  Until FinCEN gets their act together trying to comply is an excercise in futility.  If you registered as an MSB and sold coins on MtGox you would be obligated to collect THEIR KYC information (due to a lack of any exchange exemption).  Of course MtGox isn't going to give you KYC information so you couldn't comply if you wanted to.  Hopefully in a decade all this will be settled law and old news, until then you are on the cutting edge not just in the technical world but in the legal world too.


Disclaimer: the above statements are merely the opinions of the author and should not be considered legal advice.  If you have a specific legal question you should retain qualified legal counsel.

TDP of 5970 is 294W and 5870 is 224W.   The card is reported to be 350W which is significantly higher not lower.

Also this pretends away the challenges of the form factor and ignores it was AMD with three decades of experience, and the HD 5000 series was their 12th generation of graphics cards. AMD/ATI's first graphics card looked like this and consumed 10W.  


Even AMD isn't immune to the challenges of working in a compact unforgiving form factor, the 7990 (375W TDP) was delayed by six months due to power/thermal issues that they found challenging to resolve.


While 350W is possible in that form factor one would have to be willing to bet that unlike every other time the simulations aren't lower than reality AND that the company doesn't run into any cooling/power problems due to the extremely high energy density.   As for 350W being conservative?  I don't see it.  350W is 0.6 w/GH.  BFL current chips are 3.1 w/GH correct?  A die shrink conservatively means at best a 40% reduction in power (miners tend to be always on so we are really only interested in active load).  28nm is two die shrinks from current chip.  So 3.1 w/GH * 0.6 * 0.6 = 1.1 w/GH.  If the current generation was just a die shrink (Intel's tick/tock strategy) we would be looking at 1.1w/GH (660W for this card).  You stated you will both shrink and optimize (something Intel split up to reduce risk) but that is a rather significant optimization wouldn't you say?  Nearly an 86% (1.1/0.6) improvement in performance per watt outside what is gained from the die shrink.  Intel (that small rookie ASIC designer) is happy for a 10% improvement in performance per watt from architectural changes.

Given the aggressive improvement in performance per watt necessary, combined with the lack of any headroom (if it misses by even 20% then it can't be cooled in that form factor at that speed), it would need to be a nearly flawless design and execution from start to finish.   It certainly "can" be done (it isn't beyond the theoretical limits of silicon on forced air cooling) but given BFL past promises on power and cooling well one would be betting that "this one will be different".  

Not at any reasonable costs.  Make sure to look into amperage details in datacenter contract.  Most datacenters drop 20A @ 120V (20% derate) = ~2.4W per rack.  Now you usually can get 20A @ 208V single phase for not much more which gives you ~4KW.  Going to 30A @ 208V bumps you into an entire different cost range and still "only" gives you 6KW per rack.  Datacenters are carefully designed to avoid hotspots so they really aren't wired for higher heat loads then that. Nobody is going to let you dump 20KW into a rack just because it "fits".  It might fit physically but it doesn't fit from a power or thermal standpoint.  If your heat or power load compromises other clients the datacenter is going to suffer massive fines and it is simply not a risk worth taking.  Plus the datacenter only has so much space, cooling capacity, and power.  They want to get top dollar on all three.  Running out of cooling or power capacity and having half the datacenter empty isn't going to make any beancounter happy.

Try pricing out what 150 racks with 30A 208V single phase costs.   I am not saying it can't be done but I think most people have casually looked around and saw "deals" for $400 for 40U and are greatly underestimating what a datacenter will charge for higher density loads.

That would be a bad idea (freezers not designed for continual heat loads).

Still I would expect (hope) that KNC already thermal load tested the cooling equipment.  They don't need chips for that, just an test heat source (heat plate).  They should be able to properly measure both chip and overall system cooling without the need for any chips.

Yes.  The newly formed subsidiaries which will handle the exchange related business are setup and I finally got bank accounts done properly.  Working capital has been transferred from TC, LLC to the subsidiaries, and we have been doing internal testing of payment processing.  There will be a small private beta launching within a week to ensure we have a smooth (re)launch.  Tentatively I am looking at the new site being open to the public* by September 2nd.  There should be an official update by the company on Monday.  


* With the exception of clients in the states of VA, NY, and CA at this time.

Vote with your wallet.dat and the free market will reward good companies and punish bad ones.   Nothing else is really worth anything.  Companies with "good" customer service don't do it to be nice, they do it because they have good management who have learned that exceeding expectations boosts profits.  It means they keep customers longer, those customers provide referals, and those customers are less likely to trust a competitor even when they seem to be a better deal.

Companies exist for the sole purpose of increasing the wealth of owners.  Period.  Vote with your wallet.dat and let the free market decide.

You can transfer but not at 100%.  There is a 10% transfer fee and you go to the end of the line.

Immaterial.  Like I said try it.  Crank your room AC down to 60F/15C for all I care and try to cool 3x 7970s maxed out inside a sealed 4U chassis or tower case (w/ only front fans).  Pretty simple experiment.  

Now remember the wattage for two of these "mining cards" would be more than 3x 7970s and that is if the specs are right.  What if they are 10% or 20% higher.  That is just two cards.  Kinda undermines the beauty of PCIe if you are limited (due to cooling) to 1 or maybe 2 (but results with 7970s would suggest otherwise) cards per chassis.

Simple answer... nobody knows.  By the letter of FinCEN guidance you would be a money transmitter if you exchange your BTC for USD (or other currencies).  You would not be a money transmitter if you exchange your BTC for goods or services.


That being said.  FinCEN guidance is well retardedly vague and it is highly likely that had a very flawed understanding of Bitcoin and mining when they wrote the guidance.  FinCEN also doesn't clarify the distinction between solo miners and pools.  Yes we routinely calls "workers" in a pool, a "miner" but technically they are not producing any new currency.   Only the pool is.  Even that standard can be disputed, you could say only Satoshi created the currency units, miners simply "unlock" them.

Still I don't know of any miner who has registered with FinCEN.  Registering with FinCEN is actually pretty easy.  The problem is the minute you register with FinCEN your bank is going to call and want to charge you additional fees, want to audit your AML policy, etc.  You saying "I just did it because I am a miner" probably means they simply close your account for non-compliance.  However even if you get compliant with your bank it is very possible your state will send you a notice that you are operating as an unlicensed money transmitter and ask you to comply with state regs which generally means 12-18 months, $500K bond, a whole bunch of financial requirements (min capital amounts).  Given your lack of experience in handling money transmission the state could simply deny your application.

So yeah it is utterly confusing.  It is probably confusing because FinCEN is confused in the first place.   I would point out that FinCEN 's guidance is NOT THE LAW.  It is simply an explanation (a piss poor one in this case) of how FinCEN interprets the existing law.

BTW "dealing only with exchanges" provides no protection under FinCEN's guidance.  They carve out no exemption for exchanging with exchanges not even ones registered with FinCEN.  Yes it is unbelievably stupid but law/regulation isn't based on common sense it is based on what is declared and by the letter of the guidance any miner who uses Mtgox for example would STILL need to register as a MSB.

This following part should not be considered legal advice.  If you need specific legal advice you should retain independent qualified counsel.  I personally (and this is worth nothing if you are standing in front of a judge) don't believe FinCEN intended to word the guidance they way they did.  It would be a radical expansion of MSBs to regulate any business that exchanges currency as a course of their business (i.e. miners who mine and sell BTC to pay operating costs).  However the guidance is written as it is written.  I think the likelihood of them going after miners who don't engage in other traditional MSB activity to be very low however understand you are taking that risk.   

You have one option for clarification and that is to file requesting an administrative ruling.  Essentially you explain in plain english the details of your operation and ask FinCEN to rule on if that would be a MSB or not.  If you go this route PLEASE have competent counsel write up the filing.  It isn't required but you using sloppy language could result in FinCEN ruling based on a yet still flawed understanding of what economic activity is really occurring.  The bad news is flawed or not those AR will hurt future AR and court cases.

Nominal numbers are meaningless and arbitrary.

Bitcoin's money supply is still growing at >10% on an annualized basis.  That is high monetary inflation in any economic book.  The fact that Bitcoin maintains purchasing power is a pretty solid testament to the demand but monetary inflation is still high.   

Who said don't use AC.  Keep the room at 72F for all you want.  The room being 72F doesn't mean the GPU will be cooled properly.  It is a simple experiment.  Try it.  Why do you think open rigs became the "standard".  You think people just like having extenders and cables everywhere?  You think it never occurred to anyone to just close the case with 3, 4, 6 GPU inside?   

Because they are bleeding cash and need more of it.  
Giving refunds makes the situation worse.  
Upgrading orders makes the situation better.

In the US no sale is final until customer takes ownership, usually unless otherwise specified that is when shipped.  Now the seller can protect themselves by contracting a future date but if that date has passed and there is no shipment then there is no "sale". BFL "sales are final" would not stand up to a chargeback, investigation from FTC/attorney general, or a private lawsuit.   Sales may be final but there is no completed sale until product is shipped, only an order.

Nobody said include a power supply.  We are simply talking the exact same sheet metal case but wider.  The silliest part is the stated design was for rackmounting however the case isn't standard rackmount width (19").  So it is smaller than what is needed for room to house a PSU and due to its size it can't be rackmounted without a shelf.  If it was a 4U 19" rackmount case there would be more than enough room for a PSU and it would rackmount directly.

Personally I see the semi-useless case as a waste of money.  If you are going to use a worthless case design then just drop the case.  Simply ship the hashing boards and control board and let the user supply the case.

So do this as an experiment.  Take a 4U case* and put 3x 7970s inside.  Max them out and start mining.  That would use less power than 2 of these mining cards.  Now tell me how easy it is to keep the temps under control.  There is a reason people starting using extenders and ghetto rigs.  More than 800W or so 24/7 inside a sealed case is pretty tough to keep under control without liquid cooling.

For the record though KNC decision to use a case which requires the PSU to be outside is equally stupid.  The stupidest part is if they used a larger case there would be more than enough room for an internally mounted case.

* If you don't have a 4U case a full tower is roughly the same size so turn it on its side and it is a good enough estimate for this experiment.  However make sure you seal any side/top/bottom vents as those wouldn't exist in a rackmount case.  Just cool it using fans at the front and back of the case and on the GPU only.

Maybe the final product will look more like this (but with some extra fans bolted on).  Maybe someone can do some fun with photoshop.



Well no that really isn't realistic as the GPU in the picture maxes out at 240W not 350W.