You wish!

You seem to think hardware prices will remain steady, in reality they will drop as fast as difficulty shoots up. Cointerra preorders are already being discounted at a rate of 50% per month now, not surprisingly perfectly in tune with D doubling per month,  and thats not going to stop until marginal production cost approaches mining profitability per TH.

The latter is easy to calculate (I went with cointerra's specs), but the question is, how much does it cost to make more 28nm asics ? I did the math in another thread given a $5000 price tag per 28nm wafer  (generous) and 2-3x cost for stuff like testing, packaging, PCBs and vendor markup, and it  results in a difficulty somewhere between 100 and 200 billion at todays BTC exchange rate and with EU electricity prices. Thats a ballpark figure, it may end up being half that or double that, but not 1/15th of that, no way.

I cant predict how fast we will reach those levels, it could be many years,  but the only thing limiting that is the ability of all asic vendors combined to produce and ship their products. There is no other limit. Considering the amount of vendors and the fact that at least some of them will be able to manage a supply chain better than BFL, I wouldnt be too surprised to see this happen in less than two years.

I used to have a 6 number ICQ account and 2 of the original devs in my "friend" list.
Stopped using it soon after the take over by AOL, and I cant for my life remember what I registered it with, and whatever email was associated with it is long gone. So meh.

LOL."Soon" as in in when difficulty approaches 100 billion, no sooner:
https://bitcointalk.org/index.php?topic=286126.msg3143855#msg3143855

As long as difficulty is well below that point, the mining potential/market value of 28nm asics exceeds their variable production cost,  and so one way or another,  more will be produced and added to the network, limited only by how fast they can be produced and shipped.

"bullet run" ?
Never heard of that term.
Seems like Josh still hasnt even caught up with the vocabulary of this industry, its called a hot lot.

Speed (TH/s)   1003.00
http://mining.thegenesisblock.com/


One petahash is a fact now.
Im betting on 3 PH before the end of october, just 6 weeks from now.

yeah, I just wonder if there will be a market to make good on the multi million dollar investment for a 20nm maskset.
By the time difficulty levels out, mining profitability will be so marginal, I cant see a lot of demand even if electrical efficiency of a new chip is substantially better (but not orders of magnitude better).
More likely is that the mining will shift towards miners with very cheap/free electricity.

To put it differently, if 55nm asics had been in production for a longer time and  the providers able to keep up with demand, allowing the market prices to approach production cost (which will happen inevitably), so that current difficulty would be 10-100x higher than it is today,   Im not even sure there would have been a viable market for 28nm products.

Anyway, we may or may not see another shrink one day, but we will never again see the difficulty explosion thats about to happen.  Fun times ahead .

puppy linux is really only useful if you have far less capable hardware than that. Ive run it on computers with 256Mb of RAM.
If you have 4GB, you can run any full fledged linux distro you want with any desktop environment you like; no need to go to puppy/DSL or even Lubuntu.

I run elementary OS on a laptop with 2GB, and it flies.

Big or small doesnt really matter a whole lot here, considering how "simple" bitcoin asics are, at least compared to CPU's, SoCs or even GPUs.
Even if Intel/AMD/nVidia were to design one, Im sure they could make it a bit more efficient or faster, but the difference wouldnt likely be very big.

The real differentiator is the process tech, and with 28nm all these companies are pretty much on state of the art for whats available commercially.

Anyway, just saying I dont expect anything spectacular to happen after the transition to these asics. I doubt it will even make sense to invest in a 20/22 or 14nm maskset in the coming years.

If you like apple, give Elementary OS a try.
http://elementaryos.org/

So you would define any company that would sell a product that works as advertised, is delivered on schedule, at a price agreed to by its customers as a scam, only because that product would not provide its customers with windfall profits?

Interesting. I guess that makes almost all companies a scam.

Depends want you want to achieve.
selinux / apparmor  or just plain old chroot

An SSD makes all the difference in the world.

New linux users always make the mistake thinking command line is both necessary and too complicated.  

Its neither.

Command line is rarely required, but  allows anyone to copy paste commands from a website, email or whatever in to a terminal, regardless of what GUI you use, regardless of what language your OS is in, regardless (to some extend) of distribution or version. It even works when the GUI doesnt.

Now you can do just about anything through the GUI too (obviously including apt-get update), but explaining how to do it would take dozens of lines, possibly require screenshots, would be different for every different GUI and distribution, wouldnt apply to french or chinese installations etc. Try explaining over the phone or email to a computer neophyte how to remove AMD video driver. Perhaps then you will see who incredibly useful a command like


really is.

In linux, like in windows, you use the GUI if you are trying to discover things intuitively. If you want to remove AMD drivers through gui, just use the "additional driver" app or ubuntu software center, its intuitevely enough for anyone with basic computer skills to discover that. But you use the command line to be much faster if you know what you are doing, or if you are just implementing commands someone else wrote.

Both approaches make sense and have their uses, both approaches are available in linux. But in windows, you dont have that choice, most things require a GUI.

This may also give an indication for volume:
https://bitmine.ch/?product=coincraft-ai-asic

If you buy the chips in bulk, they give prices for orders of up to 10K chips (400TH). That may be their entire production run, but that seems rather unlikely, since they would have no more chips to put in their own miners .
2PH for the batch is probably a more sane guestimate.

Dear God, I knew asic mining would be trouble before BFL even announced their first chip, but even I had never anticipated THIS many asic vendors coming out with 28nm chips in roughly the same time. Even if only a fraction actually end up shipping, this is going to be one hell of a bloodbath!

Two quick questions:
1) How many chips do you expect in the first batch?
2) This caught my eye on your website:


Ive never heard of anything like that, but it sounds like a (very) good idea. Just out of technical curiosity, Can you elaborate? Any pics you can show of this package?

There is no reason to assume that. On the contrary, I dont see why anyone would sell a used/old/ready to ship mining chip or device at a price that would guarantee a quick ROI, be it now or next year.
In the dumpster you can find GPU's and early FPGA's now, and you may be able to buy them "cheaply", but show me one that I can buy that will ROI in less than a few months. You wont find any. You cant find any thats likely to give a positive ROI at all. Nor will you next year, or the year after.

Also, 28nm asics is more or less the end game for bitcoin mining. We may (but probably wont) see another shrink to 20 or 14nm several years from now, but the sort of quantum leap we are seeing now will never happen again (barring a breakthrough in technology like quantum computers). Because of the dynamics you explained yourself, the prices for these chips will tumble and follow difficulty (with difficulty following chip prices, creating a feedback loop which is why we see exponential growth) until variable production cost meets mining profitability. At that point, which I guestimated in the neighborhood of 1 exahash,  the market will be saturated, production will mostly halt, future difficulty will be predictable and stable (well, influenced only by BTC exchange rate and transaction fees)  and if you agree miners are overbuying, then here is no money to be made by anyone. You will only see miners mining with cheap/free electricity that are operationally profitable, but with a breakeven point on their investment thats beyond their life expectancy.

The one winning horse you will have backed will be offset by all the losing ones.
Fact is the sum of all preorders and upcoming orders (far) exceeds the mining profit thats to be had. Buying equally from all suppliers just guarantees you get an average result, which is negative.

The only way for your strategy to work is if most companies do send out early orders in a mostly timely fashion, but subsequently struggle to produce and ship their long order queues, like BFL has been struggling for years now.
Its possible, but I wouldnt bet on most/all of them being equally inept managing a supply chain.

heh, you expect them to be abe to read a logarithmic chart?
 
“The greatest shortcoming of the human race is our inability to understand the exponential function.”
Albert Bartlett

Maybe this will help though:



2.5% growth per day is an exponential function (that happens to work out almost exactly to doubling every month).
And if that is not scary enough, the growth rate is still accelerating, and thats well before even a single 28nm asic was shipped.

Indeed I am.



I cant bet by ordering hardware, but I am willing to bet you or anyone else Cointerra will prove to be a real company with real products. Even if those products will end up losing money for most of its customers.

Experience and industry contacts

edit, found a public source for you:
TSMC decided to increase prices for 28nm chips "by somewhere between 15% - 25%. Given that a wafer processed using latest process technologies can cost $4000 - $5000, ..
http://www.xbitlabs.com/news/other/display/20110912192619_TSMC_Reportedly_Hikes_Pricing_on_28nm_Wafers_Due_to_Increased_Demand.html

Please note the date on that, 2011, when 28nm was brand new, very low volume and very few fabs offered it. 28nm is now mainstream, offered in volume by all major fabs. So If anything,  $4000 today would be a high estimate, but Im okay using it since Cointerra is not an AMD or nViida and pretty low volume for a company like TSMC.


Absolutely not. Masksets, especially for 28nm, are orders of magnitude more expensive, but its a one time, sunk cost. Once you have a maskset, to determine production volume all that matters is the variable production cost vs market value. Sunk costs will determine whether or not you end up making a profit, but not whether or not you should produce more batches. Its like when buying a miner, once you bought it, purchase price (sunk cost) no longer matters to decide to run it or not, you will run it until it earns you less than it costs in electricity, but the purchase price will largely determine if you end up making a profit or not.


"Bumping and binning" is part of testing and packaging. Its not included in the $5K/wafer price, but it is included in a 'a few dollar per chip", since the cost is typically closer related to the number of die's than wafers.
If you want a public reference for that too, this is all I can find quickly:
http://electroiq.com/blog/articles/2001/03/technology-comparisons-and-the-economics-of-flip-chip-packaging/

(scroll down to the last charts as the others are unreadable, and you may want to note the numbers they picked for average die cost too: $5. Not $50 much less $500).
Please note that article is more than 10 years old. Gives you an idea what BGA packaging costed back then, when it was state of the art technology. The numbers have dropped since, but are still within the same ballpark of a few dollar per chip.

For PCBs, you can get an idea from the prices of infinitely more complicated PC motherboards (that contain dozens of other asics) and still sell for $20 or so in bulk, or from a raspberry Pi that retails for $40 including the  SoC, RAM, IO etc.  
PCB cost doesnt amount to much in the big picture.


You can lead a horse to water...