LargeCoin C200 Integrated Mining Unit

[Inspector 2211]

I just got this via email, I'm not affiliated with them:


LargeCoin is pleased to announce that we are now taking deposits against pre-orders of our first dedicated, ASIC-based mining appliance, the LC 200C Integrated Mining Unit (IMU). Deposits will be held in third party escrow until final product delivery, which is expected in July 2012, and will be refunded if the product cannot be delivered on time. And yes, you can pay with Bitcoins if you prefer. Additional product details, along with terms and conditions, are found toward the end of this message.

If you would like to be contacted about making a pre-order, please use the following Google form to let us know:

[url removed]

PRICING AND DEPOSIT DETAILS

The C200 IMU is priced at USD $30,000. We are selling 25 units initially, with more to come later in 2012. To secure your place in the line, you must make a $4,500 deposit, which will be held in escrow by a third party until product delivery. You and LargeCoin will sign an escrow agreement to this effect before any money is transferred. The deposit will be returned to you if product delivery does not take place by July 31, 2012.

==============================================

PRODUCT DESCRIPTION

The LargeCoin C200 is the world's first purpose-built Bitcoin mining appliance. Designed to fit within a standard 1U of rack space, the C200 connects to the network using Ethernet, and starts mining as soon as it's plugged in to the wall. Mining is controlled via an online control panel hosted by LargeCoin, which allows you to direct mining shares to the pool of your choice and manage your entire LargeCoin cluster in one convenient place. Each C200 mines at 20GHash/s, consuming a mere 100W. Designed for high density operation, the C200 provides efficient movement of air and is suitable for operating in a fully loaded 42U rack (up to 40 units per rack).

Despite the incredible power efficiency and spatial density of the C200, it's priced competitively with GPU mining. When compared with GPU mining, the C200 consumes 100 times less electricity and 14 times less rack space, meaning there's virtually no operating cost associated with this device.

TERMS AND CONDITIONS

Your deposit guarantees that LargeCoin will ship you a C200 IMU by July 31, 2012, that the specifications of the final product will not deviate materially from the specifications shown here, and that the device will function properly when it is plugged in. When you receive the unit and connect it to the network, you will be granted a temporary mining license enabling the system to mine for a period of 30 days. When we receive the balance of your payment, a permanent mining license will be issued to you. C200 IMUs may be transferred and re-sold -- just let us know before you make the sale so that we can transfer ownership of your license key.

SHIPPING

LargeCoin ships to North American destinations for free. Shipping costs for other destinations will be born by the customer and must be paid in advance. We will notify you when the shipping date approaches and provide a shipping quote for your review and final approval. International customers should be aware that air freight and insurance may cost $2,500 or more.

90-DAY LIMITED WARRANTY

LargeCoin warrants that the C200 will function materially consistent with the specifications for a period of 90 days following device activation, which occurs when you plug the unit in to a network with Internet connectivity for the first time. If the device fails, you may return it to us at LargeCoin's expense for a full refund, or replacement unit. Devices that are dead on arrival will be replaced or refunded.

NO GUARANTEE OF MINING REWARD OR FINANCIAL PERFORMANCE

Due to the random nature of Bitcoin mining, and the large uncertainties in the Bitcoin economy and network, LargeCoin cannot guarantee that the C200 will solve Bitcoin blocks at a particular rate, or that it will generate a financial benefit of any kind. Mining profitability depends on a number of factors, including the selection of a mining pool, which may or may not charge fees that reduce the mining reward; changes in the Bitcoin mining reward calculation including but not limited to scheduled reductions in the mining reward; and fluctuations in difficulty factor. LargeCoin's warranties and guarantees extend only so far as the hashing rate provided by the appliance, and its level of average power consumption.

[Inspector 2211]

My initial assessment is: It's more than twice as expensive as it should be. Thank you, but no, thank you.

[bulanula]

Is this the second unicorn project

Even BFL appears to be WAY more legit than this cheap PR trick.

Who is ordering ?

20 GHash/s at 100W only possible with dedicated ASIC.

You think they have the $$$ millions needed for that ?

I doubt it ...

[nmat]

Is this the second unicorn project

Even BFL appears to be WAY more legit than this cheap PR trick.

Who is ordering ?

20 GHash/s at 100W only possible with dedicated ASIC.

You think they have the $$$ millions needed for that ?

I doubt it ...

They announced it many months ago (it was before this, but I didn't find the link). Also, there is another thread here.

[Raoul Duke]

I love the part about the specs coming from the chip manufacturer simulations.

If the simulation is as good as the one BFL made then we know they are dreaming

[BlackPrapor]

Make a discount on those specs, and you'll get an approximate spec of this system. Is it worth $30k? For me not at all.

[finway]

The price of Bitcoin is the key to the ASIC miners.

[DeathAndTaxes]

I think the OP needs to reconsider the pricing.

LargeCoin C200 IMU as offered:
Price: $30,000
Hashrate: 20 GH/s
Power: 100W
Capital Efficiency: $1500 per GH/s
Power Efficiency: 200 MH/W
Size: 1U rackmount space

BFL RigBox
Price: $30,000
Hashrate: 50 GH/s
Power: 2500W
Capital Efficiency: $600 per GH/s
Power Efficiency: 25 MH/W
Size: ?? Likely would fit in 4U of rackmount space ??  Maybe BFL could release a rackmount chassis one day.

Pros - Lowest $ per GH.  
Cons - Not released yet, specs unproven, potential delays

50 BFL Singles
Price: $30,000  (50x singles @ $600 ea possibly less w/ bulk pricing)
Hashrate: 41.5 GH/s (50x 830 MH/s)
Power: 4500W (50x 90W likely less w/ high efficiency power supplies)
Capital Efficiency: $720 per GH/s
Power Efficiency: 9.2 MH/W  
Size: ?? Physically could likely fit in 3U of space but removing 4500W will require more space, airflow, fans, etc.  A guess to put it into rack units 6U?
Pros:  Specs proven in the wild.  
Cons: Worst MH/W
Notes: Doesn't make much sense to get 50 of these instead of a RigBox but if delivery is in 4 weeks one could start hashing sooner on 50 BTC blocks than waiting 3+ more months for RigBox or IMU.  

144 ztek boards (Why 144?  Just a guess on max one could put in an industrial 2U chassis w/ separate chassis for host)
Price: $36,720  (144 single FPGA boards @ $255 ea possibly less with a custom licensed run)
Hashrate: 30.67 GH/s (144x 213 MH/s)
Power: 1400W (144x 9.8W likely less w/ high efficiency power supplies)
Capital Efficiency: $1190 per GH/s
Power Efficiency: 22 MH/W
Size: 2U or 3U?
Pros:  Specs proven in the wild.  Similar MH/W than rig box.  Could start mining in 2-4 weeks.
Cons:  Worst $ per GH OTHER THAN IMU.

100 Hypothetical "next gen" 28nm ztex board
(doesn't exist yet don't bug ztex, just a back of napkin guestimate based on conservative 28nm FPGA performance)
Price: $30,000  (100x 28nm FPGA boards @ $300 ea possibly less with a custom licensed run)
Hashrate: 40 GH/s (100x 400 MH/s)
Power: 1100W (100x 11W likely less w/ high efficiency power supplies)
Capital Efficiency: $750 per GH/s
Power Efficiency: 36 MH/W
Size: 2U or 3U?
Pros:  Competitive w/ all offerings.
Cons: Doesn't exist.  Availability will depend on economical availability of 28nm parts which is likely 2013+
Assumptions:  33% per FPGA price increase due to demand.  80% increase in MH/s and 40% decrease in MH/W due to die shrink.


So lets normalize all these for a hypothetical 1 GH/s fractional rig (@ $0.10 per kWh)

Code:

Board     Upfront   Annual    1yr     2yr    3yr    4yr
C200 IMU     $1,500     $4  $1,504  $1,509  $1,513  $1,518
BFL RigBox     $600    $35    $635    $670    $705    $740
BFL Single     $720    $89    $809    $899    $988  $1,078
Ztex cluster $1,190    $40  $1,230  $1,270  $1,309  $1,349
28nm cluster   $750    $24    $774    $799    $823    $847

The year costs are the purchase cost plus total electrical cost.  (i.e. to run 1GH/s of BFL singles for 4 years will cost $1078 total).

On edit: fixed decimal place error (which massively changes results) and the update is even worse for LargeCoin's relative value.  It doesn't even offer a lower total cost of ownership compared to the power hungry BFL Single.

Price needs to be cut by 50% for it even to make economical sense under any conditions.  Sure 1U of space is nice, and sure 100W power draw is incredible but it doesn't matter if the upfront price is so far out of line as to be noncompetitive.  To be viable it likely doesn't need to be the lowest cost offering but it has to at least get in the ballpark.

[DeathAndTaxes]

The price of Bitcoin is the key to the ASIC miners.

A myth.  Price is irrelevant.  Only price/difficulty matters or in terms easier to deal with revenue per GH/s.

IF price went up to $30 and stayed there difficulty would go up 6x making equipment no more profitable then today at current difficulty and $5 BTC.

Rapidly increasing price can provide a short term benefit as difficulty lags price by a couple months (takes time for miners to commit to near hashing power, build rigs, and get them online) but in the long run difficulty follows price.

$5 BTC, $50BTC, $500,000 BTC it doesn't really matter difficulty is going to follow price.

Look at how much both price & difficulty have changed in last 90 days but the daily revenue per GH/s trades around $4  (+/- $2).  I love these charts that bitcoinx made but chart is in 100MH/s which is just a weird increment still just multiply values by 10 to get daily revenue per GH/s.



If BTC was $100 and the block subsidy was down to 25, and difficulty was 12 million 1 GH/s would be making ~ $4 per day.

[finway]

Yeah, i know what you mean.

But as the price rise 10 times,
you can't make your GPU farm 10times bigger than it is now ,right?
No mention to electricity cost efficiency, and price dropping due to the scale effect.
That will make ASIC grid flourish.

The price of Bitcoin is the key to the ASIC miners.

A myth.  Price is irrelevant.  Only price/difficulty matters or in terms easier to deal with revenue per GH/s.

IF price went up to $30 and stayed there difficulty would go up 6x making equipment no more profitable then today at current difficulty and $5 BTC.

Rapidly increasing price can provide a short term benefit as difficulty lags price by a couple months (takes time for miners to commit to near hashing power, build rigs, and get them online) but in the long run difficulty follows price.

$5 BTC, $50BTC, $500,000 BTC it doesn't really matter difficulty is going to follow price.

Look at how much both price & difficulty have changed in last 90 days but the daily revenue per GH/s trades around $4  (+/- $2).  I love these charts that bitcoinx made but chart is in 100MH/s which is just a weird increment still just multiply values by 10 to get daily revenue per GH/s.



If BTC was $100 and the block subsidy was down to 25, and difficulty was 12 million 1 GH/s would be making ~ $4 per day.

[Brian DeLoach]

I think the OP needs to reconsider the pricing.

LargeCoin C200 IMU as offered:
Price: $30,000
Hashrate: 20 GH/s
Power: 100W
Capital Efficiency: $150 per GH/s
Power Efficiency: 200 MH/W
Size: 1U rackmount space

It looks like your GH/s numbers are all off by a factor of 10.

[DeathAndTaxes]

It looks like your GH/s numbers are all off by a factor of 10.

Oops thanks.  Sad thing is that error was the only thing even keeping it "close". 

Updated post.

[Brian DeLoach]

Oops thanks.  Sad thing is that error was the only thing even keeping it "close". 

Updated post.

Electricity would have to be $.47 per kilowatt for the C200 to overtake the Rig Box in five years. Or, it would take 24 years at $.10 per kilowatt for the C200 to overtake the Rig Box. Either scenario is not very likely. I am completely lost with LargeCoin's strategy. They know the demand will greatly exceed 25, only if they wouldn't limit the initial run to 25 thus pushing the price way up.

[Inspector 2211]

Oops thanks.  Sad thing is that error was the only thing even keeping it "close". 

Updated post.

Electricity would have to be $.47 per kilowatt for the C200 to overtake the Rig Box in five years. Or, it would take 24 years at $.10 per kilowatt for the C200 to overtake the Rig Box. Either scenario is not very likely. I am completely lost with LargeCoin's strategy. They know the demand will greatly exceed 25, only if they wouldn't limit the initial run to 25 thus pushing the price way up.

There are many foundries that offer multi-project wafer runs, for instance http://www.globalfoundries.com/services/global_shuttle.aspx, and you receive anywhere from a few dozen to a few hundred chips.

So, maybe that's all they [will] have: A few hundred chips, enough for 25 boxes (each containing, say, 10 chips).

[DeathAndTaxes]

Oops thanks.  Sad thing is that error was the only thing even keeping it "close".  

Updated post.

Electricity would have to be $.47 per kilowatt for the C200 to overtake the Rig Box in five years. Or, it would take 24 years at $.10 per kilowatt for the C200 to overtake the Rig Box. Either scenario is not very likely. I am completely lost with LargeCoin's strategy. They know the demand will greatly exceed 25, only if they wouldn't limit the initial run to 25 thus pushing the price way up.

There are many foundries that offer multi-project wafer runs, for instance http://www.globalfoundries.com/services/global_shuttle.aspx, and you receive anywhere from a few dozen to a few hundred chips.

So, maybe that's all they [will] have: A few hundred chips, enough for 25 boxes (each containing, say, 10 chips).

Then don't use ASICs OR use the 25 units as loss leaders to show the product is viable and secure large enough order to get full wafer production run of 10,000 chips.

ASIC simply doesn't make sense in small runs.  It never has and it never will.  This is the entire reason or economic basis for developing FPGA.  In large runs FPGA get destroyed by ASICS but lots of products don't require tens of thousands of hundreds of thousands of chips per year.  That is where FPGA shine.   It allows a lower NRE and thus makes it more economical in small runs.  

So
ASIC - high fixed cost & low per unit cost
FPGA - low fixed cost & high per unit cost

Obviously if you graphed these two options (total unit cost vs # of units) they will intersection as some # of units.  Where depends a lot on a lot of variables but generally speaking it in the thousands of units. 

Simplistically:
Intend to sell <1000 units - don't use ASIC
Intend to sell 1000+ units - don't use FPGA



Usually those small run multi-project wafers are not used for retail products.  They are used to provide real silicon prototypes so you can catch bugs, improve designs, perfect layouts BEFORE running off 10,000 potentially sub optimal chips.

[coblee]

Basically, they are currently looking for 25 suckers to help them

catch bugs, improve designs, perfect layouts BEFORE running off 10,000 potentially sub optimal chips.

Then they will sell thousands of these and difficulty will shoot up and these 25 "early investors" will never make back their money.

[Inspector 2211]

Basically, they are currently looking for 25 suckers to help them

catch bugs, improve designs, perfect layouts BEFORE running off 10,000 potentially sub optimal chips.

Then they will sell thousands of these and difficulty will shoot up and these 25 "early investors" will never make back their money.

Exactly.

Let's say a few friends come up with 20 grand each to finance a multi-product wafer run.
They get 250 chips.

So they build 25 miners, sell them for 30 grand each and now they have 750 grand (less expenses).

After that, it's a simple three-step process:
1. You have a proven ASIC design
2. You have about 700 grand for a real ASIC design
3. Profit!

[siggy]

So lets normalize all these for a hypothetical 1 GH/s fractional rig (@ $0.10 per kWh)

Board     Upfront   Annual    1yr     2yr    3yr    4yr
C200 IMU     $1,500     $4  $1,504  $1,509  $1,513  $1,518
BFL RigBox     $600    $35    $635    $670    $705    $740
BFL Single     $720    $89    $809    $899    $988  $1,078
Ztex cluster $1,190    $40  $1,230  $1,270  $1,309  $1,349
28nm cluster   $750    $24    $774    $799    $823    $847

The year costs are the purchase cost plus total electrical cost.  (i.e. to run 1GH/s of BFL singles for 4 years will cost $1078 total).

for an even better comparison.. lets compare it to a 7970 rig:
Board,     Upfront,   Annual,    1yr,     2yr,    3yr,    4yr
7970 rig    $800   $350   $1150   $1500   $1850   $2200

looks like the C200 takes 2 years to even compete with a GPU.  That is even assuming zero resale value of the GPU.

[Inaba]

I have to agree... as someone considering buying 2 RigBoxs and with 40 BFL Singles on order, it's not like I wouldn't drop the cash for something viable... but this is completely off the scale for payback and I wouldn't even consider purchasing one of these at the current $/MH.  This thing is worth no more than $10,000, if that... especially considering that the time frame for delivery is July, as assuming no schedule slips.

You'll never make you $30k investment back from one of these things.  Either ASIC takes off and difficulty rises and you're screwed, or difficulty remains the same and block reward halves... you might see a 1/3 bump in value, so your payback stretchs into years.  Naw... you'd be crazy to purchase one of these at that price.

RJK: You're datacenter can't handle a power request like that?  Ask them about private suites, they can usually handle larger power requests, at least the ones I use.  400 amps for a suite is not uncommon.  If you are willing to suck it up on the open floor, you can get 220v drops and pull down 200 amps fairly easily as well.  Did the specs on this thing say it was 220v or 120v?

[rjk]

RJK: You're datacenter can't handle a power request like that?  Ask them about private suites, they can usually handle larger power requests, at least the ones I use.  400 amps for a suite is not uncommon.  If you are willing to suck it up on the open floor, you can get 220v drops and pull down 200 amps fairly easily as well.  Did the specs on this thing say it was 220v or 120v?

I haven't seen mention of whether it is 120 or 220VAC.
I was basing my datacenter comment on a single rack. I'm sure I could get a suite, but when half of the $600/rack/month cost is for power, and when you only get 40 amps for that, I dread to find out the cost for 400 amps and associated cooling requirements.

Of course I could probably get a lower cost by requesting non-UPS power and minimal genset backup, but since I am not buying 40x of these, I won't be bothering to get a quote on that