Topic: BFL are expecting 100,000 chips... - page 4. (Read 6989 times)

My bad, I was doing it in my head and must have been biased by a previous response to someone claiming that ASICMINER could just print out a couple petahash/s for cheap. That one was almost 10MW.

You wouldn't even have to do it in China, if you think they would make enough equipment to actually try over a weekend to 51% the network. You could probably just cut a deal with someone that operates a large industrial facility to use it over the Christmas weekend or something. Still, what would that get you? A 51% attack doesn't let you just take all the coins. I find it hard to believe that they would be able to pull off any double spends large enough to justify the cost of the operation.

On the flip side, there is absolutely no incentive for BFL to mine on any hardware until they get to the point where there's product sitting on the shelf. For now and for the foreseeable future, they can sell the hardware and make more than self mining anyway.

You are assuming they will sell $15 million dollars worth of equipment. Highly doubtful in my opinion. But it doesn't matter. Competing with customers is still the wise move.

Why ship the chips out of China? It is a waste of money. The attack doesn't need to be conducted from NYC real estate.

You can rent an internet cafe capable of supporting 60 computers in China for less than US$300 a month. (e.g. http://cq.ganji.com/fang6/400435890x.htm)

At 350 W per computer minumum that is 21 kW. Now rent 40 such cafes. (probably just better to focus on really big ones) We are up to US$12000 a month.

What about electricity? Say it is US$0.10 cents a kWH. Then for 750 kW, that is $75 per hour or US$54000 per month. Here we are getting expensive. However, I think you could get away with just attacking for 24 hours and then managing your load so that you regulate on any asshole who tries to mine. So it is probably more reasonable to think of 2 days of electrcity per month or US$3600.

We also need say 5 employees to keep stuff running 24/7. I'd say that each will run about US$1 an hour per body. That is another US$3600 per month.

So our total variable costs so far are 3600+3600+12000=$19200 per month. It is the price of a Honda. We just need the hardware.

If it works, the revenue per month is up to US$1 million. You are trying to convince me that the managers of the company are braindead. Sonny, extradited from Italy, multimillion dollar scammer, an idiot. I don't think so.

Maybe they won't 51% attack. Instead they can just earn US$500,000 per month and keep their complete control of the network secret to avoid panic. That might be a smart move. No one will know the difference. You idiots will just think that people massively overinvested in hardware and celebrate how "secure" the network has become. It will demonstrate that people are willing to mine at a loss. LOL.

This ends badly.

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.

And this argument also applies to any other attacker.  There is not scenario that ASICs could be bad for Bitcoin at large.  The door is closing for a serious 51% attack even for an attacker with nation-state resources.

Once they ship 750TH/s, they'd need the equivalent of 500 Minirig SCs to 51% the network. Considering that just the power supply used in the Minirig SC costs $450, I would imagine they would have a hard time launching a 51% attack on the network for $500k.

They'd also need a datacenter that could supply them 7.5MW.

That's the wrong question, because if BFL can make ASICs, others can too.  And BFL isn't really making anything, third party manufactuers in China are.  Their ASIC design isn't safe, and they know it, even if it was unique, which it isn't.  The ASIC market is really a race to get the first batch to market, each new offering is going to limit the profitablility of all previous offerings, and thus the value of it's own offering.  That's what real competition is supposed to do, it's just in an extra fast cycle in Bitcoin.

I dont believe it

We've been through this before but again it makes no sense whatsoever to attempt a 51% since if they are caught doing that, it would crash the Bitcoin market. In that situation they would be doing much worse than they are if they play nice.

They have actually commented on this many times and said that it's in the best interest of Bitcoin to do it this way. I call BS on that though, it's in the best interest of them to do it this way. It just happens to be good for everyone else as well.

Really. After they sell the first batch, how much extra would it cost them to 51% the network? $500k? less than that? Can they 51% the network for the price of a new Honda? Might be interesting to take bitcoin for a test drive.

I think they will sell 100k chips by the end of 2013.  After the first generation products ship, they will drop prices 2-10x and sell more units throughout the year.  The marginal cost of producing an ASIC chip is extremely low.

Also, shouldn't the title be "BFL is expecting 100,000 chips..." ? Of course the chips are plural, but BFL is singular, is it not?

They're not going to actually have sold 750TH/s by the end of the year.

750TH/s is 12,500 Singles. At $1300 per unit, that's $16million USD in preorders. I don't think they have that many preorders, but I could be wrong.

They're going to get 100,000 chips in 2 waves, the first of those waves being stated previously as being ~20,000 chips. This will cover the first batch, which puts us right at about 150TH/s. That's about a 5-6x difficulty compared to where we are now.

Eventually we will get to 750TH/s, but not for another few months, at the earliest.

According to this:
https://forums.butterflylabs.com/bfl-forum-miscellaneous/437-asic-update-26-november-2012-a.html

Each single has 8 I think so that means 100,000 chips are 750THash/sec

So difficulty would jump by a factor of 30 to about 100million.

This means a single will earn you about $114 a month after the diff drop on a single (and cost you about $5-$10 in power).


Everyone agree with my maths?