Topic: CoinTerra announces its first ASIC - Hash-Rate greater than 500 GH/s - page 108. (Read 231002 times)

Need to be lower than 4 gh/s to be reasonable.

It sounds like they are finally coming around to reality. While the drop is still not bringing the miner into a truly profitable "buy" scenario, let alone to actually mine with, it is a first step in the right direction.

They also seem to (I assume) be listening to advice that they need to 'pave a path' for their future products beyond the immediate December delivery window. It sounds to me like they might be starting designs on scaling their chips.

Either way, if they continue to listen and adapt to reality, they are so far among the best poised due to prudence. Lets hope.

Price drop to $13999, still not profitable with a Dec delivery

http://mining.thegenesisblock.com/a/5fe987161b

http://www.coindesk.com/cointerra-cuts-price-of-terraminer-iv-bitcoin-mining-rig/

Why 27 PH/s? Isn't that in the neighborhood of 4B difficulty?

Yeah, but you have to consider the marginal cost for people who already have chips. Labcoin estimates it'll cost them about $2/Gh/s to add capacity. Once the masksets are done for the 28nm chips they'll likely be able to add capacity for far less.  Maybe even 50 cents per GH. 

With HashFast's 4x miner protection plan, they expect they'll be able to make a profit for $15/5 = $3/Gh (since they'll have to send 4x chips for every one chip they sell)

So that's just $81m at HashFast's "Miner Protection Plan" cost.
Or $51m at Labcoin's claimed 130nm chip cost.

Total amount of bitcoins mined a year is worth about $130m I think. So it could potentially be profitable to create that many chips.  Plus companies don't know about eachother's plans.

Not a chance.

The funny bit is *no one has to buy the hardware for that hardware to mine*.  Self-mining is already alive and well, and mining gear will not collect dust in warehouses while manufacturers see profitability windows closing.  Self-mining will remain profitable to ASIC manufacturers long after it bankrupts their retail customers.  Common sense.
As far as "protecting the bitcoin ecosystem" in a bout of enlightened self-interest?
We'll watch prisoner's dilemma played out with with consumers as free cannon fodder

Simple multiplication.  It is what 27 PH/s would cost at the various price points offered by various vendors. (i.e assumming they could deliver today buying 27 PH/s from BFL would cost you ~$1.35 billion).  In order for the hashrate to rise to 27 PH/s that means someone had to buy 27 PH/s of gear.  At the current hardware prices and exchange rates I don't think that is likely.

I guess another way to look at it is how much hashing power (in USD) do you think has been bought.  My guess (SWAG) is that it isn't enough to push us to 27 PH/s as fast as you think. Note this doens't mean buying more hardware is a good idea certainly not at current prices but I think the law of large numbers is going to slow the rate of hashing power growth.

And with a single post you have me turned 180 degrees on the profitability of mining. It kind of puts everything in perspective, doesn't it?

Simple conversion from a previous poster's formula: 1 trillion diff @ 42K USD/BTC -> x diff @ 110 USD/BTC

x diff = 1 trillion diff * 110 / 42000 ≃ 2.6 Billion diff.

Factoring in the miner's price is a bit tricky as you don't know when it was bought, how much electricity it consumed over time, what the sale price was (+ other associated costs) so it's more like an upper limit that probably won't be reached (never say never though...), unless you disregard the payback time (and then we're back to uselessness).

That does seem high. How do you get those numbers?

Right. Don't forget the private miners: ASICMINER, bitfury (http://ghash.io already at 80+TH/s right now) and undisclosed private ASICs.

Costs for private mining could be well under $1/GH.

Note that manufacturers likely have to book a certain amount of chips (I heard sets of 25 wafers) at the factory. So the number of chips actually hashing in the future might be larger than the number of orders that the manufacturer actually manages to land.  (Also, no KNC??)

Difficulty will overshoot by a lot into deeply unprofitable territory before the flow of chips reduces.

At $8 per GH that would be $216,000,000 in hardware. (Monarch & Cointerra)
At $15 per GH that would be $405,000,000 in hardware.  (Hashfast)
At $20 per GH that would be $540,000,000 in hardware.  (Bitfury October)
At $50 per GH that would be $1,350,000,000 in hardware (BFL, AsicMiner, Avalon, Bitfury August).

Possible but unlikely.

Sure, it's a worst case estimate. But don't forget, difficulty is going to hit 65M today, so only 40x gets us to 2.6B. 100% increase each month gets us to 1B difficulty by end of year. That's not that unreasonable. There's a lot of hardware that's coming online Q4.

Remember that assumes free hardware.  It is just electrical cost  = value of Bitcoins mined.  Any higher difficulty and one would be mining at an operating loss.  Most people aren't going to buy a $5,000 box which lets them covert $100 into $100 in BTC.  I mean imagine MtGox required a $5,000 fee just to open an account.  Would you?  That is essentially what we are talking about.  So as difficult rises buying more hardware will be less attractive and at the break even point it is completely unattractive.

You could consider it asymptotic, I don't think difficulty will reach break even electrical cost but it does provide an upper bound.  Obviously if break even electrical is ~2.6 billion it is kinda silly to project difficulty to reach 1 trillion.  Miners would be mining for a >95% loss (convert $100 in electricity into $5 in Bitcoins).  I don't think most people would do that given they can just flip the switch to off

Not sure what numbers were used for the 2.6 billion but I used an average GH/W value for the network based on all ASICs released/pre-ordered so far.  It is only a guesstimate but without hard numbers it is difficult to know more.  Another way to say it is that unless either efficiency significantly improves (i.e. 0.2 W/GH or lower) or Bitcoin exchange rate significantly rises (>$200 USD per BTC) it is highly unlikely difficulty will go beyond that.

But I doubt it's a reasonable estimate as it would represent about 50x the current hashing power, or add another 27000+ TH.

That's a lot of hardware, even at 400GH/s per unit (≃67K units).

That'll take about 14 difficulty adjustments at 30% each. I think we'll reach that by end of this year.

≃2.6 billion diff = ways to go.

1) It likely will be last to market.  Even if one or more companies are delayed (or never delivers) you can expect at least three well capitalized companies will be shipping in significant volume before this release date.  That is a bad starting position.

2) ROI% is highly correlated to starting difficulty.  The further out the start of mining the more uncertainty.  The bigger issue is that given all the pre-orders and all the other companies with release date over the next couple months there never has been more uncertainty to what difficulty will be in 90 days than at any other point in time.  Having a launch after that makes a buy right now very risky.  Since your ultimate ROI% depends heavily on what the difficulty is when you start mining even if you are 100% certain you will have the chip on 1 December right now you are essentially gambling that difficulty will be low enough.  You might be right but you also might be wrong.  As we get closer to December the uncertainty will be less (because we will see the rate that new hardware is making it to the market and can project future rollout rates).  Right now at this price point you might as well go to Vegas and put $14K on the pass line at the craps table, save the other $1,600 for one awesome weekend either way.

3) Once other companies clear their backorders they have no incentive to keep prices high.  The NRE is paid, the gross profit margins are extremely high and they all have a product which devalues based on difficulty.   That creates a perfect scenario for a pricing war.  Worst case scenario a company like bitfury clears their Oct pre-orders, sees Nov sales flat line and cuts prices 75% for delivery in two weeks because they estimate a competitor will be clearing their backlog by December and they want to capture the sales while they still can.  You took all the risk and some noob buying in late November gets a unit quicker and cheaper.

4) A lack of details.  Ok they don't have the chips yet I get that.  However the site has nothing.  It is a marketing brochure.  Give us something.  How large is the case?  Will it be rackmountable?   Do they have some initial PCB layout?  What do they intend to use for a controller?  How do they plan to handle the heat?  500 GH/s is a large hot die.  Some details on the chip, package, heatspreader would be nice.   Have they done any thermal simulations on the proposed case (airflow, heatsinks, etc).  None of that requires an active chip.  A watt is a watt.  Ceramic heating elements in a case would provide assurance they can remove the kind of heat necessary. Even simple stuff like how big is the PSU?  Can it be used on a 120V 15A outlet (max of 1440W)?  Would really suck as an American to find out that due to wattage it will require a 240V outlet if you don't have one.

For the record I don't think it is a "scam" or fake.  Given the people involved I do believe they have the talent and connections to launch a product. That doesn't mean it is a good buy though.

Simple version:
least amount of details
highest system cost
latest launch date