Topic: 1GH/s, 20w, $500 — Butterflylabs, is it a scam? - page 28. (Read 123112 times)

Off the shelf sha256 units plus an FPGA controller sounds about right me, both in terms of performance/watt and cost/performance.  Just saying.

Especially since many of the same people in this thread are completely willing to write off FPGA products like the x6500 which are using the _same_ FPGA as some other products because they aren't yet claiming the higher performance numbers that they should be able to achieve but have not yet (because of time spent working on building boards to ship to customers, instead of on promoting more vaporware).

If BFL demoed a FPGA running at a fraction of the speed and using a lot more power, I'm pretty sure many of you would say that was more evidence that it was a scam. Worse for them, they'd probably alert their competition to identity of their supplier (and the non-scamness of their business). So it goes either way.

This controversy is fantastic marketing. We're all watching for the drama— and a lot of people will buy buy buy when there is any evidence of it being real, even less evidence than they'd demand had we not had this little adventure. This all sounds like a reasonable enough approach, scam product or real :-/

sub'ing to this to see how it turns out 

I never said it makes sense if you intend to sell 1 or 10 chips. I only said the per chip cost could be dramatically lower. Amortizing the NRE is just a matter of selling enough of them. I dont have numbers to back that up, but I would guess its a low triple figure number. Not exactly out of reach if you look how big the bitcoin mining market alone is.

You always point to the low unit price as suspicious, you think the hashrate too high and the power consumption too low. But all these indicators are perfectly compatible with s-asic with low per unit cost and relatively high NRE..


?
I dont think they will get a single working chip without paying altera or whomever they are sourcing from. Also, its not because Altera might require you order 1000 units minimum that you will get those all at once. Id be very surprised if they didnt do a test run with an absolute minimum number of wafers first. 


Yes, I agree, if this is a sasic, they should have working FPGAs for quite some time. But why would they show them? Obviously they are not going to meet the performance and power claims and they would have no intention of selling them. Whats the point?

Anyway, so far I have not seen a shred of evidence that what BFL is saying is untrue let alone impossible. Obviously we havent seen a lot that validates their claims either, the only sensible attitude IMO is to wait and see.

No because FPGA are produced in bulk.  There are millions of identical Spartan6 produced each year.  The cost of the mask is amortized over a large number of units.

With sASIC the run is much smaller.  You do gain economies of scale but only in large volume (so you can amortize those production run costs and mask costs).


Prefinancing what?  You prebuy a run and a run of sASICS is going to be 1000+ units.  You aren't going to get very good cost reduction (due to high fixed costs) without 10K units. 

So lets say they already bought 10K units and they have been delivered.  What do they need funding for?

A company which could design, prototype, and fab 10K custom processors (like $250K+) needs to sell a handful of boards to keep the lights on?

Also where is the protype?  The whole point of sASICS is you can prototype on a FPGA, work on the kinks, improve performance, optimize power load and ONLY then drop the big money on a run of chips.  The sASIC run generally takes 6 to 8 weeks (assumming no backlog at the fab) so obviously to meet their timeline they already had 100% functional and mining prototypes 3 months ago.

Exact copies of the sASIC design but using FPGA.  Yet nothing.  No demos? No photos of the prototypes.  Nothing.  Suddenly at the finish line (after fabbing 10K units at costs in the hundreds of thousands of dollars) they need pre-orders?

Compared to FPGAs you are looking at a reduction of die size of up to 80%. That actually means a cost reduction thats potentially more than 80% per chip (higher yields, more candidates per wafer etc).


I dont think anyone said they would only receive 100 chips. The point of presales is usually prefinancing.


Actually, an "implementation of both FPGA and ASIC technology" as they put on their website, fairly well describes a structured asic, without spilling the beans.

I disagree with that.  I know you disagree with my disagreement but you asked so I answered.   

Structure ASICS would have the potential to achieve those performance per W numbers but not those performance per $.  I don't know about others but Alterra doesn't do runs of less than 1K units.  There would be no need to limit the number of pre-sales if they have 1000+ units.  Volume discounts really don't kick in until 10K units.

Further the double speak of FPGA and ASICS is even more nonsensical.  If you have access to the kind of capital and resources to run off 10K+  units of sASICS there is absolutely no need for FPGA, nor is there need for scammy looking pre-orders on unverified products.

Im a believer of a few things;
- I believe its possible to achieve the kind of performance/W and /$ that BFL claims, if nothing else, using structured ASICs. Is anyone disagreeing with that?
- I believe it much more likely that BFL will ship a working product than you putting your money where your mouth is. Why dont you start by offering those 10% of purchase price you promised?

I dont think I ever said anything definitive about BFL.

Here's my two bit-cents.

Real scientists with an academic degree I would think have matured to a level where they really don't care about what people think, they will deliver if and when they have a product. They would therefore not lower themselves to participate in a shit-storm on any internet forum, if they did, that would raise a lot of red flags with me.

Most skilled people in any business is more interested in expanding their knowledge and working in their field, rather than wasting their time arguing with negative people on a forum.

Maybe nothing comes out of this whole spectacle, but they've set a date, the 25th of Nov. We shall see if BF Labs deliver at this date, or at least if they don't deliver, they should invite a member of the community that could tape a tour of their 'offices' and interview one of the head R&D people.

If we see more delays and no real product, this is nothing but an elaborate scam to hoist as many gullible people in as possible.

If I lived nearby, I would insist on getting on a visit and see what they're up to. If they're legit and honest, they'd have nothing against having some genuine bitcoiner coming by and having a chat and a look around.

There could be at least a couple of scenarios:

1. Some university drop out with a half backed electronics degree and some knowledge about computers figured this would be a nice way to extract some money from the bitcoin crowd, if you have the knowledge, making a front that seems legit should not be impossible. So we need knowledgeable people on this forum to ask hard questions to find wheter this is all a scam or if it's legit.

2. It's actually some weird scientists/tech people working on an invention and actually have adequate funds to fund the research and development, and for some odd reason they've set up shop in a loft, instead of renting a business location. I've seen and heard weirder things happen.

Sadly I don't have the technical expertice to judge if BF Labs are scammers or not, but I have sound sceptism, and no money would ever leave my hands to a shady party unless I was convinced I'd be doing a good deal, and I'd absolutely use an escrow service, or a trusted local party!

Just beware of the classical scammer signs: "It will happen in 14 days...", "You will have the product in 3 weeks". Always delays and excuses.

BF Labs would have everything to win by working with the community and be upfront about what they are doing, and everything to lose (except if they just want to extract funds from stupid people that preorder a product that's not sure to be known from a perhaps shady company).

We'll see on the 25th what happens, maybe we will be asthonished, but more likely there will either A.) be no response or B.) the same day or a few days later there will be a message that the product is delayed, and they will set a new date, which again will be pushed etc..

We must also be aware that even if a box is produced and we see a box showing a hash rate on a display, this is not proof. We need independent testing, by someone like the pool-operator Inba that lives close by.

The product should also be inspected to verify that the hashing is not provided by an external party through a cellular internet-connection. :\

There was recently a case where someone wanted to rent a huge amount of hashing power on the forum, and I have absolutely no clue if it's a connection there, but I'm just saying it would be possible to build a 'black' box, with some impressive looking circutry that connected to an external hashing farm, and got it's hashing power from there, and when Inba or another tester booted up the box and connected it to his pool, he'd see the advertised hashing rate and deem the product legit.

So if the product could be tested, also with a radio-jammer present, that would make for a better testing environment.

Sorry if these last sentences makes me sound paranoid, but I'd truly and sincerely hate for anyone to be scammed in a major way again withing the bitcoin economy, there's already been a lot of cases, and scams not only makes peoples pockets empty, but may also make people lose faith in bitcoin, and that's not good at all.

Only time will show, but I'm not holding my breath.

Also I would like to apologize in advance if I in some way is misinformed or if some of my points have already been discussed in this thread already (I didn't read every comment in this thread, although I've tried to follow it since it started).

I am suggesting that it be demonstrated at a conference with hundreds of technically-minded individuals, most of whom are quite capable of independently verifying these claims, and that makes me a believer?
I'm not sure that I follow your deluded train of thought.

I don't consider most of the Prague conference-goers to be gullible. They could deliver the first pre-orders there at the same time.

What would be really cool is if they would demo it at the Bitcoin conference in Prague.

Thanks! I was really just kidding, didn't mean to fish for compliments!

The output really just depends on the code, of which ours isn't really optimized yet. It looks like ztex, rph, and ngzhang made some big improvements there lately. I'm confident that we'll get ours closer to that in time. The FPGAs are identical, after all.

The real question is, what's ngzhang's price? Hopefully it's not as good as the BFL price or we're dead!!

What about our stuff?

Waiting to see the magic, though one forum member zhang also did something that is quite good.

48 hours to go mate. I think they will blow our socks off !

Maybe they also managed to do like :

-$250 price
-2 GH/s
-10 W

It is magical, I'm telling you ! Anything is possible. Maybe they used FPGAs from Egypt and Libyan routers LOL.

Hurry hurry and get those pre-orders in.

Website indicates they will ship in 4-8 weeks (strangely it said that 4-8 weeks ago).

3 days left. TICK, TOCK, TOCK 

I am waiting to see the magical unicorn on the 25 Nov. Can't wait . Prepare your socks to be blown off.

Yes, we'll do it retroactively. We coded this feature for our admin interface. Many bets like this one will turn to chaos, if we are not willing to intervene. Let us know when you think that one of our statements can be concluded.