I think this graph says it somewhat better:



Largest spike in the past 2 months at halving. Looks like whatever caused it is dying down now though.

I can't read chinese so I had to use google translate. That says 40 computers not 60. I don't see why it would require 350Watts per computer for an Internet cafe, seems mildly unreasonable, as it is not necessarily designed to be a 3d gaming cafe. I would expect less capacity than the 14kW (close to 120A on 120V, not sure what China uses, but that is a huge breaker box for a single store) that your 350W calculation provides. That is usually for Industrial applications, not commercial property. 40 computers averaging 180W would be closer to 7.2kW (somewhat more reasonable 60A load). Assuming you could find > 100 locations with similar capacities for the same price (I'm no China expert) that's closer to $30,000 / month in cost for space. $54000 in Electricity cost seems reasonable for 750kW, however that is with 0 cooling costs, for 750kW of heat. Probably don't want to just leave your doors open all day and night with some fans on (especially come August when it's ~90F). Assuming that there is already Industrial cooling installed (very unlikely, need more machine costs there), I'm not a cooling expert but google suggests 1TON AC for every 3.5kW of cooling, or for 750kW ~220TON units of AC. 2.5ton of central cooling googles out to about 2520kwh / month, so we're at 221760kwh for cooling, or another $22,176 a month, and we haven't even included that power draw on our infrastructure (increasing the spread requirements).
I'm guessing that you would require more than 5 local yokels to administer to > 100 sites, but I'm not going to bother estimating that.

Without any logistical issues, we're at rent + asic cost + cooling cost = $30,000 + $54,000 + $20,000 =  >$100,000 / month to run 750THash/sec.

Another note to keep in mind:


China is struggling to meet Electrical Capacity nationwide, doubly so in the "Industrial Park" areas such as Chongqing (your listing), this does not make for a bastion of reliability for a fairly extensive mining operation, especially when you are utilizing cut-rate commercial sites putting huge loads on the local infrastructure.

So, again, not a trivial matter, but yes, assuming that they have some sort of connections in China (and ignoring that cost), ignoring the fact that the PCBs and other parts need to be shipped to China (they have stated that those are not MFGd there), ignoring the costs of all parts of the chips, ignoring the costs to have the parts assembled, ignoring the costs of the chips, ignoring the cost to retrofit their cutrate buildings rentals to handle a huge constant electrical load + cooling, ignoring the infrastructural problems in China, assuming no one minds your constant heavy usage for minimal pay, assuming cheap efficient labor keeping mining going at all times, assuming the network does not grow beyond 750THash of "legitimate" miners, ignoring the hassle of translating massive quantities of BTC into usable operating capital, etc., then yes, it certainly is possible to run a very hefty and profitable large scale mining operation.

More likely they would just set up shop not in the bargain basement street shops, but in some tech development zone in large warehouses with existing power and cooling infrastructures. The cost would be higher, but it would be less ridiculous a proposition.

It is and has been a concern, but it is not inevitable. There are lots of costs and headaches involved, and requires a belief in bitcoin itself, rather than selling on the belief of bitcoin, which may or may not be evident in a scammer who sees greenbacks, not digital bits.

Not really true... or even partially true.



Only passed ~.5% in Sept and lasted a little over a month, after which the rate of growth has been steadily decreasing from Mid-Oct until Late Nov (where it was down to about .35%/day) when there was a sudden reversal and seeming surge of new hardware right around block halving.

The reasons remain mysterious, super-coincidental crazy network-luck, miners turning on to grab the last 50block and forgetting to turn off, asics, who knows? But it definitely bucked the trend right at halving, when many people have claimed to be switching off. Mysterious.

Everything is a matter of perspective. One neat thing about ASICs is that if they work as intended, even the small fry will be profitable for a long time. To whit:

http://tpbitcalc.appspot.com/?difficulty=993438908.9602&hashrate=60000&exchangerate=12.42&bitcoinsperblock=25.00&rigcost=1300&powerconsumption=60&powercost=0.10&investmentperiod=355

A BFL Single SC @ 1Billion Difficulty, still makes money. Not much, but not nothing. At a Billion difficulty (300x now). Something current day GPU miners could never say.

Que? When did it become some sort of crazy notion to want to know who is behind a business?

I googled "20 person company" and the first result was a google news report about a 20 person company that landed 80 million dollars in funding, named Anandale Construction. Googling Anandale Construction founder first result is a business journal Profile about the Founder (who is also President and CEO), a brief rundown of his personal history, his business partners names, the sources of his funding (including an upcoming contract with An Airforce Base for $10Mil), I stopped there but I suspect there's plenty more where that came from. (Side note: Not a publicly traded company)
Business journals routinely run profiles on businesses of all sizes who are more than happy to tell you their whole life story. Why in the world this seems CRAZY to people is... interesting.

Most businesses don't have the direct personal cozy relationship with their customers that Bitcoin and ASICs do, and so it's not often that anyone would have a chance to directly request that information, however freely making that information available is not at all uncommon.

Whether or not they choose to do so is their business, and I will be the first person to admit I will not tell any of you anything about who I am if I can help it, but it is hardly some sort of madman's snooping prying-eye request. I shrug I guess.

*Never* get your money back? That would be hard based on the devices efficiencies. Would require high Elec prices, dropping BTC value, skyrocketing difficulty, late delivery, and low Hash/J+$/Hash ratios.

A long ass time to get your money back? That's very probable.

Tons of calculators out there, bitcoinx.com, http://bitcoineering.appspot.com/simulator, http://tpbitcalc.appspot.com/, etc.

Plug in numbers that sound right to you and then have a think about it. No one really knows what will happen, so your guess is good as anyone elses.

That dude above me sounds reasonable, but he could be totally wrong.

Small correction: 750kW not 7.5MW. Still large and expensive, but not impossible. Certainly uninteresting I would imagine, both for expense and logistical issues.

More practical might be hiding 50TH/sec (or whatever number) of their own within the network.

Dude, see like literally 1 post above yours...

Nope!

Forum users are a small small minority of the hashing network it often seems, surprising as that may be.

I expect a delay of a week to a month for results to become truly apparent, as people either notice their bitcoin wallets slimming/their pool rewards not being what they once were or they get their next power bill and realize "hey this shit isn't paying itself off so well anymore".

I also expect that the "big boys" (> 100GHash/sec) had to be supremely efficient in the first place to even get that large, and are still making significant enough bucks after the halving that it is worth it to keep going (people with 4GHash/sec might think making just $1-$2 / day is annoying, but multiply that by 25+ and you're still a happy camper).

I was hoping to see some decrease instead of a huge increase in hashing power though, for a graceful decline in hashing power, rather than a big drop that might incite panic.

Different clients function differently, in order to send/receive bitcoins you do not need to be a node, or have the full blockchain.

In this context to be a "full bitcoin node" you need to have the Full BlockChain on your machine and be actively receiving and transmitting bitcoin transactions/block finds from many peers, assisting in dispersing blockchain information to as many people as fast as possible.

I perhaps make it sound complicated, but it can be as simple as you suggest, running a bitcoin client like Bitcoin-qt 24/7 on a machine that has open access to the internet (has an open port that isn't firewalled for listening for peers).

Well there is a whole subforum for speculation, admittedly this wasn't posted there but it's obviously not unwarranted. With the drama of block-halving come and gone (to really very little effect), there is little else to do but speculate on ASICs now anyway.


Josh said that the week of the 11th (the 11th is a Tuesday(?), meaning it could be anywhere between the 11th and the 15th, and Josh has stated that it is likely to be towards the end of the week, here) is the "fuzzy date" when the first batch of re-vamped chips will arrive, not when they will be shipping. They have said they are still going to be doing the 1/3rd plan batch first, so assuming they are really committed to putting out a significant quantity of orders, that could/would take a week+ to generate (I believe there was some talk of building each rig and testing for 8-24 hours before declaring it "ready for shipping", this requires many man-hours).



Whether they will begin shipping before the end of the year I do not know, but they certainly will only be able to get a paltry amount of orders out before the end of the year. A poster asked on 11-20, if he ordered that day, when would his order ship, to which Josh replied "End of Jan or Beginning of Feb", this was of course, before the delay was announced.

These are all company statements, not my own speculation.


Now for fun, I will do some speculation:

20k Chips were enough for all pre-orders until 10-31 (and more). So for fun I will say that 150THash (the full chip amount) = pre-orders 7-1 to 10-31 (4 months), or 37.5 THash/month. Tack on 20 days for 11-20, that's 25 more Thash, for 175THash in orders by 11-20. Expected time of arrival of chips being ~25days from 10-31, is ~11-25. So...

11-25 -> 1-25 = ~2months to create and ship 175THash/sec of orders, giving us a production capacity of 87.5Thash/sec per month, or roughly 21Thash / week, so they can take weekends off and stuff (best case scenario).

With the new arrival date being considered, likely to be ~12-14, if they want to do a 1/3 shipping plan first batch of 21THash (seems reasonable, approx equiv to current hashing power), giving an extra few days for testing (the first week of production will need a slight additive for testing that will later be rolled into the consecutive weeks of prod), they can under the best circumstances be expected to have out the door about 21THash/sec by the 26th of december, juuuust beating the new year (but they have to work through Christmas ).

Once things get rolling I expect 21/week sustained becomes easier to manage, so assuming demand picks up (once people see an actual product in hand, they will be less wary of buying), so how long to whip through the rest of those 100k chips?
Well 100k chips = 750THash, at 21Thash per week that is 35ish (rounded to 36) weeks, or about 8.3 months to whip through the whole batch.

Interesting that if Tom begins shipping Mid Jan (say ~1-14) and he produces half of what BFL does, and Avalon ships end of Jan (~1-31 say) and they produce half of what Tom does (based on first pre-order batch sizes), by end of August we will potentially see a network of > 1.25Peta hash  

Or they could all be significantly delayed again. Time will tell.

EOF (end of fun).

EDIT: Looks like so far so good on my above estimates. Avalon recently said that they would try to finish their 300 preorder batch by end of feb, which means about 20 THash/month, roughly 1/4 the rate of 21THash/week. Damn I'm good!

As I recall from my days long ago, 5970s were far more efficient devices for $/Mhash and even Mhash/J than 6950s, and I assume by extension 7970s. Or are the 7970s a lot more power efficient? I haven't kept up on cards in a while.

Kudos to you sir. And you've given me a good idea too. I've long been thinking about getting one of those cheapie little low-power computing devices like a BeagleBone or RaspPi, to play with, and potentially run future mining operations on, but I never really could think of anything entirely USEFUL I'd actually want to do with them. Running a full node on such a device would be low cost, a mildly interesting learning experience, and quite useful as well (I'd like to support bitcoin as well).

600W was just a broad guess, worst-case scenario type of thing. The latest word I've heard is they're expecting something like 160W? Don't quote me on that though, but it's definitely way down from the scary 600.

Will Avalon ship first? Or nearly first? Good Q., it's looking more and more likely though.  We might end up with some very surprising first-run winners chomping up all the easy BTC (assuming network hash rate drops to like half its current rate right around mid-Jan, oh myyy).

There are tons of 58XX fan replacements, not all of them cost an arm and a leg either, that you can easily replace the stock fans with.

You're one of the few names I remember from my old mining days (I've been gone a long long time, $0.34/kWh is not friendly), and I didn't particularly like you, but it is a harkening to times I quite enjoyed, so I tip my hat in salute to end of an era.

I'm personally very interested to see where ASICs lead us, so I'll be around for a bit myself (and why I'm back after all this time).

Just for fun I calculated this out, at 315MH/sec, no electrical costs, current difficulty, and 10 week payoff, you would have had to get each board + chip for $40/chip (free board), or for a ~950MH/sec setup, board + chips + fans + etc would have to be $120.

That sound right to you, or do you think you need to re-calc?

I laughed.

210000 (Main Chain)    2012-11-28 15:24:38    000000000000048b95347e83192f69cf0366076336c639f9b7228e9ba171342e    Slush      457    194.46
209999 (Main Chain)    2012-11-28 15:01:40    00000000000000f3819164645360294b5dee7f2e846001ac9f41a70b7a9a3de1    BTC Guild      543    234.18

so close.

If he can make $1200 out the gate, which is possible as I have no idea what he's running, then certainly he is free to do so. I was not responding to the idea that he forever mine, I was responding to his response to a post about heating. Which does not rely on A vs. B as listed above.

A better way to look at it would be:

Option A: Sell the hardware and get $1200(?) now.
Option B: Heat the house, make $13 a month, pay off all other electric usage, and take what, 3 months of extra mining time until spring before selling for $1200-X? (x potentially being negligible, I doubt 3 months will make them valueless).