[Archive] BFL trolling museum

[MrTeal]

okay so 1 SC Single @ @$1299 is equal to 11.42 SC Jalapenos.  Round it down to 11 Jalapeno's and that would cost $1639.  If you can swing for an SC Single, it is like getting two extra Jalapeno's for free to process the same amount of hashes.

As for the SC mini-rig, you would need to buy 25 SC Singles to reach 1TH/s for a total of $32,475.  You save the cost of two SC Singles for a total of $2,576 if you just go with the rig.

$30K is a bit steep though.  I'd rather just buy 12 SC Singles for a little more than the cost of a single previous generation $15K mini-rig which would still let me process 480 GH/s.  It makes absolutely no economical sense anymore to get the current mini-rig for $15K but process 25GH/s.

The only way it makes sense is if the SC's take another year to get here.

But man, 480 GH/s would produce 274.46 BTC per day at the current difficulty.  That would be 8,233.8 BTC per month x $6.53 mt gox exchange = $53,766.71 per month.

Yeah...talk about disruptive technology...lol  Most of the 21 million bitcoins would be mined in no time compared to the original estimates...

No, I think the timeline will be pretty stable.

[sadpandatech]

Yeah...talk about disruptive technology...lol  Most of the 21 million bitcoins would be mined in no time compared to the original estimates...

the difficulty changes every 2 weeks. So unless we could online this massive amount of hash power in that timeline, difficulty will quickly increase to keep the block generation as close to 144 per day as possible.

[interlagos]

My best guess is that trade in policy will only be unit for unit within a product category.
Meaning you can only trade in 1 "old" single for 1 "new" single, 1 "old" minirig for 1 "new" minirig. And jalapenos will be sold without any trade ins. This way they will still be getting $$$ flow with every sale or trade in. Mark my words

[bulanula]

My best guess is that trade in policy will only be unit for unit within a product category.
Meaning you can only trade in 1 "old" single for 1 "new" single, 1 "old" minirig for 1 "new" minirig. And jalapenos will be sold without any trade ins. This way they will still be getting $$$ flow with every sale or trade in. Mark my words

words marked.

It can be true. They said "full terms of tradein to be announced at a later date" = they can screw you hard if you are stupid and believe their marketing.

[P4man]

I would assume the same, that the single ASIC is 3.5GH/s. Still, that doesn't change the fact that they are going to have fixed costs. Even if the chips are free, they need to place 286 of them in a minirig. That's going to have no insignificant expenses for PCBs, power supplies, and cases. Do you think they would be able to ship a Minirig even with free ASICs for $300?

Current minirigs? Barely I guess. If nothing else, those heatpiped heatsinks arent exactly free.
The minirig SC? who knows. Lets wait and see if there really are 200+ asics in there, and how much or how little support logic etc they need; but if you doubt its at all possible to cram tons of asics on a PCB cheaply, look no further than some memory modules. They cost like a dollar more than the chips on them.

Regardless if 100x lower end user products are feasible, even if its "only" 20x or 50x it doesnt really change the issue. A miners ROI will still go up from, say, 1 year to his lifetime.

[||bit]

Whatever happens, I suppose it would make more sense that they ship the smallest units first and more frequently. Not just because they can be made faster, but also to distribute the hashing power across more individual miners. So, if one is wanting to beat the curve on this upcoming ASIC frontier, it might be best to order the $149 modules in bulk rather than go for a $30k rig.

I'd rather have 200 of those small cards now, than a nice slightly faster rig at the same cost 6 months from now.

[Hawkix]

It seems 2013 will be very interesting year for Bitcoin. (No, they will not deliver in 2012, I can bet on it).

[MrTeal]

I would assume the same, that the single ASIC is 3.5GH/s. Still, that doesn't change the fact that they are going to have fixed costs. Even if the chips are free, they need to place 286 of them in a minirig. That's going to have no insignificant expenses for PCBs, power supplies, and cases. Do you think they would be able to ship a Minirig even with free ASICs for $300?

Current minirigs? Barely I guess. If nothing else, those heatpiped heatsinks arent exactly free.
The minirig SC? who knows. Lets wait and see if there really are 200+ asics in there, and how much or how little support logic etc they need; but if you doubt its at all possible to cram tons of asics on a PCB cheaply, look no further than some memory modules. They cost like a dollar more than the chips on them.

Regardless if 100x lower end user products are feasible, even if its "only" 20x or 50x it doesnt really change the issue. A miners ROI will still go up from, say, 1 year to his lifetime.

Memory makers have much greater economies of scale that BFL does. I'd imagine BFL will be looking at 1000s of units, not tens of millions.

I'm making a couple assumptions here, but I don't imagine I'm far off. First, the MiniRig will use an array of singles PCBs, similar to the current setup. I would guess that the future SC MiniRigs will go in the same cases as the current MiniRigs, which probably cost BFL well over $100 in the small quantities they buy. I'm going to estimate power draw for 1TH at 500W-1kW or so, so there's going to be a pretty beefy power supply, as well as each of the 25 single boards carrying a 12V to Vcc power supply. Add on assembly and test costs, and I don't think a small outfit could reasonably get Minirigs out the door for less than $1000 with any kind of profit.

[cablepair]

1)    BitForce SC Jalapeno: a USB powered coffee warmer providing 3.5 GH/s, priced at under $149

1. A USB port can deliver 2.5 Watts.

2. This means they are claiming 1,400 MH/J.

3. The current record is under 25 MH/J, so they are claiming a 5600% improvement in power efficiency.

Does anybody here believe this?

I call bullshit.

thank you.

on a side note:

How long do you think BFL will mine on their wonder ASICs before making them available to the public?

How long before gigavps buys 51% of the network with his American Express card?


If this PAID-ADVERTISEMENT :http://news.yahoo.com/butterfly-labs-announces-next-generation-asic-lineup-054626776.html

is only half true Bitcoin is completely screwed.

[P4man]

I'm making a couple assumptions here, but I don't imagine I'm far off. First, the MiniRig will use an array of singles PCBs, similar to the current setup. I would guess that the future SC MiniRigs will go in the same cases as the current MiniRigs, which probably cost BFL well over $100 in the small quantities they buy. I'm going to estimate power draw for 1TH at 500W-1kW or so, so there's going to be a pretty beefy power supply, as well as each of the 25 single boards carrying a 12V to Vcc power supply. Add on assembly and test costs, and I don't think a small outfit could reasonably get Minirigs out the door for less than $1000 with any kind of profit.

Well, considering their minirig SC will sell for $30K, some of those assumptions make sense. But once difficulty has gone up enough to require prices per GH to go down by >10x or 50x, it will make equal sense to redesign for cost efficiency and not have 25 daughter boards with their own power circuitry etc.

Anyway, this discussion is missing the big picture. Even if you are right and prices would bottom out above $1000 per TH, thats still, what,  500x lower than today? And therefore, difficulty would be ~500x higher. Good luck estimating the ramp on that and earning back the investment of your minirig sc.

[sadpandatech]

Anyway, this discussion is missing the big picture. Even if you are right and prices would bottom out above $1000 per TH, thats still, what,  500x lower than today? And therefore, difficulty would be ~500x higher. Good luck estimating the ramp on that and earning back the investment of your minirig sc.

I think anyone would find it hard to disagree that difficulty will go up to match the $/MH. But would not the miners for profit still do the same costs/roi caluclations. I.e., if difficulty is then at 14.8mil that would be the number I would use when considering whether or not 15 grand worth of hardware investment is worth it. Also considering whether or not difficulty is on the climb, etc, etc.

Everything you say is accurate except that it assumes miners for profit are all retarded. ;p

cheers

[jl2012]

Wrong. The difficulty changes every 2016 blocks, not 2 weeks.

Yeah...talk about disruptive technology...lol  Most of the 21 million bitcoins would be mined in no time compared to the original estimates...

the difficulty changes every 2 weeks. So unless we could online this massive amount of hash power in that timeline, difficulty will quickly increase to keep the block generation as close to 144 per day as possible.

[sadpandatech]

Wrong. The difficulty changes every 2016 blocks, not 2 weeks.

Yeah...talk about disruptive technology...lol  Most of the 21 million bitcoins would be mined in no time compared to the original estimates...

the difficulty changes every 2 weeks. So unless we could online this massive amount of hash power in that timeline, difficulty will quickly increase to keep the block generation as close to 144 per day as possible.

you know what I meant.....

[Transisto]

I'll be shorting AMD stocks.   The flood of use cards is not gonna help.

LOL at the June 18th HD7970 raffle,,,

[ailikun]

well...i know that everybody should have their own opinion on this...
but my guess that this asic thing will definitly ruin bitcoin as a worldwide system....

and its sure way to early to make those rigs so quick in calculations...

they couldve atleast make then twice as fast for the same price as bfl single...
but why make them so fast?

[P4man]

Everything you say is accurate except that it assumes miners for profit are all retarded. ;p

No, not at all. But profitability in the asic era will be determined mostly by the ones that are most retarded, and frankly, reading the forums they alone would pose a serious threat

On a more serious note, its not a matter of being retarded; you cant estimate the speed at which these things will sell and subsequently drop in price, and therefore the speed at which the difficulty will be going up. In a way, buying those asics at ANY price (well, above, say 5x marginal cost) is potentially "retarded". Its a bit like a dollar auction. let me quote wikipedia:

..a paradox brought about by traditional rational choice theory in which players with perfect information in the game are compelled to make an ultimately irrational decision based completely on a sequence of rational choices made throughout the game.
http://en.wikipedia.org/wiki/Dollar_auction

Buying ASICs is a bit like that. Except, you wont have perfect information, unless BFL opens its books and shows you the orders and backlogs and production schedule.
Did I mention a competitor yet?

[jl2012]

Wrong. The difficulty changes every 2016 blocks, not 2 weeks.

Yeah...talk about disruptive technology...lol  Most of the 21 million bitcoins would be mined in no time compared to the original estimates...

the difficulty changes every 2 weeks. So unless we could online this massive amount of hash power in that timeline, difficulty will quickly increase to keep the block generation as close to 144 per day as possible.

you know what I meant.....

They are completely different. If the difficulty change is time-based, we could theoretically mine 21 million bitcoins instantly. But this is not true

[sadpandatech]

Everything you say is accurate except that it assumes miners for profit are all retarded. ;p

No, not at all. But profitability in the asic era will be determined mostly by the ones that are most retarded, and frankly, reading the forums they alone would pose a serious threat

On a more serious note, its not a matter of being retarded; you cant estimate the speed at which these things will sell and subsequently drop in price, and therefore the speed at which the difficulty will be going up. In a way, buying those asics at ANY price (well, above, say 5x marginal cost) is potentially "retarded". Its a bit like a dollar auction. let me quote wikipedia:

..a paradox brought about by traditional rational choice theory in which players with perfect information in the game are compelled to make an ultimately irrational decision based completely on a sequence of rational choices made throughout the game.
http://en.wikipedia.org/wiki/Dollar_auction

Buying ASICs is a bit like that. Except, you wont have perfect information, unless BFL opens its books and shows you the orders and backlogs and production schedule.
Did I mention a competitor yet?

haha, maybe retarded was too strong a word, maybe not. ;p  The part I bolded would keep me from making any sort of disruptive financial decission on a new investment there. Hopefully I am not alone and will be watching that factor very, very closely..

[gyverlb]

The trade-in offer seemed odd to me from the beginning because I don't see how it makes economical sense for BFL. They can resell the FPGA, but they will take a huge loss on this : the margin on the FPGA units will be eaten by the work needed to remove the FPGA from the units before selling them at the same price (or lower) than the price they acquired them in the first place.
So a $ for $ trade-in is a net loss for them which they'll take quite some time to absorb.

There is one existing explanation : they don't want to antagonize their existing customers and have orders canceled which would deprive them of some most needed cash.

What isn't clearly explained by this theory is why selling castrated SCs (less efficient but by design) without a trade-in offer wouldn't be more profitable for them on the long term and as profitable short-term ? Don't obsolete the existing products (which are quite competitive today) but make the new one the best on the market by an healthy margin, wait a bit for competition and release new hardware when needed. This is assuming they can deliver ASICs in large quantities. But what if they don't, what if ASICs will suffer the same problems than the BFL's FPGAs (relatively low count in the hand of customers, large delays) ?

I just thought of another possibility that could fit the situation nicely : what if they have more orders than FPGA units they can actually deliver ? Isn't offering a trade-in solution, getting back FPGA units, testing them quickly and sending them to the waiting customers and letting the original purchasers wait for SC whatever time needed to remain profitable a nice way to handle the shortage problem ?

The more I think of it, the more it makes sense : you don't have to spend too much on FPGAs to fulfill orders paid upfront this way, you could send the same units to 2, 3 or more customers, benefit from the good feedback of the first wave to have a "they're not so bad" reputation as they manage to be profitable.

Then you find an existing ASIC in short supply that is better than FPGAs, inflate its capabilities in an announcement (rings a bell ?), rinse and repeat (first wave of customers satisfied, huge backlog of waiting orders, ...).

It's just a theory and I've no facts to back this up but I've learned than when things are to good to be true, usually they aren't. The implications are huge though : only a very small proportion of BFL customers would ever be profitable if the theory matches the real situation and it will end in a very ugly way. If I were about to send money to BFL, I'll ask for quite some delivery guarantees...

[BR0KK]

The trade-in offer seemed odd to me from the beginning because I don't see how it makes economical sense for BFL. They can resell the FPGA, but they will take a huge loss on this : the margin on the FPGA units will be eaten by the work needed to remove the FPGA from the units before selling them at the same price (or lower) than the price they acquired them in the first place.
So a $ for $ trade-in is a net loss for them which they'll take quite some time to absorb.

There is one existing explanation : they don't want to antagonize their existing customers and have orders canceled which would deprive them of some most needed cash.

What isn't clearly explained by this theory is why selling castrated SCs (less efficient but by design) without a trade-in offer wouldn't be more profitable for them on the long term and as profitable short-term ? Don't obsolete the existing products (which are quite competitive today) but make the new one the best on the market by an healthy margin, wait a bit for competition and release new hardware when needed. This is assuming they can deliver ASICs in large quantities. But what if they don't, what if ASICs will suffer the same problems than the BFL's FPGAs (relatively low count in the hand of customers, large delays) ?

I just thought of another possibility that could fit the situation nicely : what if they have more orders than FPGA units they can actually deliver ? Isn't offering a trade-in solution, getting back FPGA units, testing them quickly and sending them to the waiting customers and letting the original purchasers wait for SC whatever time needed to remain profitable a nice way to handle the shortage problem ?

The more I think of it, the more it makes sense : you don't have to spend too much on FPGAs to fulfill orders paid upfront this way, you could send the same units to 2, 3 or more customers, benefit from the good feedback of the first wave to have a "they're not so bad" reputation as they manage to be profitable.

Then you find an existing ASIC in short supply that is better than FPGAs, inflate its capabilities in an announcement (rings a bell ?), rinse and repeat (first wave of customers satisfied, huge backlog of waiting orders, ...).

It's just a theory and I've no facts to back this up but I've learned than when things are to good to be true, usually they aren't. The implications are huge though : only a very small proportion of BFL customers would ever be profitable if the theory matches the real situation and it will end in a very ugly way. If I were about to send money to BFL, I'll ask for quite some delivery guarantees...

That was my idea to. I wrote about this some pages ago!
They simply send old waiting cusomers your old single