1GH/s, 20w, $500 — Butterflylabs, is it a scam? - page 47.

makomk

hero member

Activity: 686

Merit: 564

Quote from: DeathAndTaxes on November 08, 2011, 08:52:48 PM

The wattage is not "similar". Other FPGA designs are getting about 20MH/W not 50MH/W. 250% improvement in electrical efficiency is massive.

Other designs use Spartan-6, and Xilinx's tools make a real pig's ear out of placing and routing Bitcoin miners on it. I suspect this is probably quite bad for power efficiency. Other FPGAs are better at this, the trouble is they're also quite a lot more expensive at least if you're buying them from distributors.

Quote from: DeathAndTaxes on November 09, 2011, 07:45:58 AM

There aren't magical ultra cheap and ultra powerful FPGA lying around. While there are volume discounts, "volume" means thousands if not tens of thousands of chips not 200.

There might actually be. For examle I have a suspicion that the manufacturing costs of Spartan-6 chips might actually be higher than a better-performing Virtex-6; Bitcoin mining is very inefficient at making uses of resources on Spartan-6 and Virtex-6 should be a much better fit. However, because of the way Xilinx segment their markets and price their chips it's cheaper to use the inefficient Spartan-6.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: gmaxwell on November 09, 2011, 10:03:54 AM

Quote from: DeathAndTaxes on November 09, 2011, 07:45:58 AM

maybe 2GH/s+ from a custom built ASIC, so the nominal speed claim isn't what is unbelievable.

An OT tangent really, but this is ridiculously and stupidly conservative. FPGAs are quite neat— but in terms of silicon and power they are enormously wasteful. It's trivial and expected to get >>10x better on both fronts going from FPGA to a proper ASIC. Without the inefficiency of the FPGA in the way the circuit for a miner is rather small so a mining ASIC on modern process would be thermally limited. Even in cheap packaging you're going to easily do better than 2GH/s.

Sorry for being clear I was talking about structured ASICs (i.e. Altera Harccopy series). Your right a true custom ASIC would have massively higher throughput but honestly that is a pipedream anytime in next decade. Bitcoin mining is simply too small to warrant the enormous up front costs and risks of a custom ASIC design.

On the topic of ASICS it also makes no sense to use both FPGA and ASIC to do hashing so BF Labs talk of using both FPGA and ASIC is simply marketing double speak. IF they have a $1.85 USB controller it is an ASIC chip so technically any board is going to be a blend of ASICs and FPGAs.

P4man

hero member

Activity: 518

Merit: 500

Quote from: ?? on ??

You think I make less than $3000 a year? I mine at nearly 10Ghash/S

And yeah, I wont be able to bet if no one is willing to bet with me. I stated the terms of my bet before this last option was set up. I wanted all bets done in 10 days, now 8 days. My terms are very reasonable. I will give you 10 to 1 odds as well if you want. I will have the coins to Inaba in less than 24 hours if you want.

I will also bet anyone 10 to 1 odds on this just to prove that I will bet. Just one person. First come first serve.

however Taxes, BF Labs and P4man I will also do the 10 to 1 odds if any of you are willing to risk even 1 coin on this

i already bet 5btc on a bet i didnt formulate and didnt want to make . waiting for your 50

gmaxwell

staff

Activity: 4242

Merit: 8672

Quote from: DeathAndTaxes on November 09, 2011, 07:45:58 AM

maybe 2GH/s+ from a custom built ASIC, so the nominal speed claim isn't what is unbelievable.

An OT tangent really, but this is ridiculously and stupidly conservative. FPGAs are quite neat— but in terms of silicon and power they are enormously wasteful. It's trivial and expected to get >>10x better on both fronts going from FPGA to a proper ASIC. Without the inefficiency of the FPGA in the way the circuit for a miner is rather small so a mining ASIC on modern process would be thermally limited. Even in cheap packaging you're going to easily do better than 2GH/s.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: ?? on ??

BLAH BLAH BLAH BLAH I am big and powerful.

BTW I am betting on the side of it being a scam not legit. It still doesn't make your trolling and lame behavior any less trollish and lame.

Raoul Duke

legendary

Activity: 1358

Merit: 1002

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Goat you aren't going to bet anything so just shut up please.

There is a very easy to use escrow system already. You could bet 1 to 1000 BTC. Right now if you bet less than 2BTC you would get better than even money. Remember payout is based on final bet amounts. A large bet (say 100 BTC) would give people taking the other side huge odds and that will increase demand in the bet driving up the reward. So it isn't like even if you bet 100 BTC it would be locked in a 50:1 odds. 50:1 odds would encourage bets and bring that close to even money.

So either bet or just shut up. Stop making stupid and outlandish claims of thousands of BTC. Nobody is impressed.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: bulanula on November 09, 2011, 06:40:49 AM

Price is not an issue right now. I just cannot see how anyone can get 500 mhash/s out of a single ASIC or FPGA. Not even the 5870 gets that much. The numbers are all a big fairy tale.

Um the speed isn't the issue at all. Price & wattage are. Homegrown developers have gotten 200MH out of Spartan 150 which is pretty low end FPGA. A high end FPGA like a Virtex6-300 would certainly be capable of 500MH/s+. Of course Virtex6-300 is going to cost $600 ea so you are talking $1200 just for the pair of chips when you add rest of components, labor, and retail markup probably $1600+ not $500.

Hell you likely could get 1GH/s+ from the most powerful FPGA and maybe 2GH/s+ from a custom built ASIC, so the nominal speed claim isn't what is unbelievable. The wattage and price as the unbelievable claims. 50MH/W is 250% of the best anyone else has achieved and $0.5 per MH is incredibly low price point. There aren't magical ultra cheap and ultra powerful FPGA lying around. While there are volume discounts, "volume" means thousands if not tens of thousands of chips not 200.

For a company which didn't have a website until this year and wasn't incorporated until August these are some high claims but don't confuse it with impossible. It is 100% possible today just not at the price points and wattage they are claiming.

P4man

hero member

Activity: 518

Merit: 500

Quote from: ?? on ??

Blah blah I do not really like the idea of putting money down and not knowing what I will get back if I win blah blah

blah blah blah

You were so sure you would bet $10.000 but you dont like odds for betting 10BTC

blah blah blah

JonHind

full member

Activity: 154

Merit: 100

Wouldn't it be a great idea for the scammer to get lots of money in 'escrew' for a 'bet' and do a runner with that money along with all the pre-order money. Double win.

bulanula

hero member

Activity: 518

Merit: 500

Quote from: ?? on ??

Quote from: P4man on November 09, 2011, 06:22:19 AM

Quote from: ?? on ??

I am interested in this, but I do not understand why the dead line to bet is Jan 1st. Why would anyone place a bet before Dec 31st just in case some new info comes out?

Because its time weighted. Early bets weigh much more than late bets.

Quote

Also, if I put in lets say 1000 bitcoins, I will only win 5 or so? (assuming no on bets heavy the other way)

Of course the assumption is wrong. If you bet 1000 bitcoins, I and others will bet against that, even people who might be skeptical. The final ratio is a good indication of what people actually expect. So why dont you start with 100 BTC, heck, just 10BTC and see what happens?

I do not really like the idea of putting money down and not knowing what I will get back if I win. There is no way to know until after. Also only 90% of the losing bets are distributed.

I will stand with my original offer of a bet with Inaba holding the coin and testing the product. I will also give Butterfly Labs 2 to 1 odds just for the fun of it:)

Heck for the first 10 bitcoins I will give anyone 2 to 1 odds with Inaba holding the coin and him testing the product. He will have to get it before 2 months from this date (he can take longer to test it, but no more than an extra month) and also the product must be sold at the stated price. and have a 20% variance on what is claimed. If anyone is interested let me know!

Price is not an issue right now. I just cannot see how anyone can get 500 mhash/s out of a single ASIC or FPGA. Not even the 5870 gets that much. The numbers are all a big fairy tale.

P4man

hero member

Activity: 518

Merit: 500

Quote from: ?? on ??

I am interested in this, but I do not understand why the dead line to bet is Jan 1st. Why would anyone place a bet before Dec 31st just in case some new info comes out?

Because its time weighted. Early bets weigh much more than late bets.

Quote

Also, if I put in lets say 1000 bitcoins, I will only win 5 or so? (assuming no on bets heavy the other way)

Of course the assumption is wrong. If you bet 1000 bitcoins, I and others will bet against that, even people who might be skeptical. The final ratio is a good indication of what people actually expect. So why dont you start with 100 BTC, heck, just 10BTC and see what happens?

P4man

hero member

Activity: 518

Merit: 500

Quote from: bulanula on November 09, 2011, 06:02:12 AM

What are you talking about ?

Total agree bets: 3.25
Total disagree bets: 6.00

The statement is BFL will NOT ship a card that matches their specs in time. 6 BTC disagree with that, so the majority so far believes they will ship in time and meet specs.

bulanula

hero member

Activity: 518

Merit: 500

Quote from: P4man on November 09, 2011, 05:54:36 AM

Naysayers put your money where your mouth is. Bets so far are clearly in favor of BFL not only being legit, but even going to live up to all its claims and in time. Stop saying you are sure its a scam unless you put some coins on it:

http://betsofbitco.in/item?id=141

What are you talking about ?

Total agree bets: 3.25
Total disagree bets: 6.00

P4man

hero member

Activity: 518

Merit: 500

Naysayers put your money where your mouth is. Bets so far are clearly in favor of BFL not only being legit, but even going to live up to all its claims and in time. Stop saying you are sure its a scam unless you put some coins on it:

http://betsofbitco.in/item?id=141

JonHind

full member

Activity: 154

Merit: 100

And after all these pages, they still haven't given any information that can identify anyone from the company? That smells like 100% scam to me. If I were a designer of a new amazing device, I would be proud of my invention.

Seriously, if anyone pre-orders one of these based on the information given to date, then please don't be surprised if you lose your money. I'd seriously reverse the charges now. You can always order them at some point in the future.

bulanula

hero member

Activity: 518

Merit: 500

Quote from: ?? on ??

Quote from: DeathAndTaxes on November 08, 2011, 08:52:48 PM

Quote from: punningclan on November 08, 2011, 08:04:24 PM

FPGA is used exclusively with the actual costs of the boards being designed somewhat equivalent. Mhash's are about a quarter of the Butterfly labs card though but wattage is similar, they do say they use a combination of ASIC and FPGA.

The wattage is not "similar". Other FPGA designs are getting about 20MH/W not 50MH/W. 250% improvement in electrical efficiency is massive.

Yeah, this is why I am so shocked that people think this can work out. The are stated outputs that are off the wall.
Assuming this is not just an outright scam this guy is off somewhere with his math.

If this guy can deliver he will be a multimillionaire within a year as this new technology could be used in a number of fields. This guy also would odds are be hired by IBM, the NSA or something like that.

I do not think people understand how revolutionary this product would be if it works and costs as stated.

I am willing to put my money where my mouth is unlike BF Labs.

How do you know they are not already working for the CIA ? Grin

This is just impossible IMHO. To get 1 GH/s with just 2 ASICs or FPGAs or whatever is just too damn improbable right now. Not even a damn 6990 gets that with 2 GPUs on it etc. Keep dreaming BFL. IMHO if anyone buys into this scam then they damn right deserve to be scammed like there is no tomorrow. $50 000 easy money for you BFL ? What will happen with this capital if your project fails and you realise you used up all the money on a prototype that does not work in the end ? Hey, I am not saying that this is a scam but anything can happen right ?

P4man

hero member

Activity: 518

Merit: 500

My bet was rejected because "scam" is apparently a curse word or something Roll Eyes

But someone else started a bet, which should be a lot safer for the naysayers than I what I was willing to bet, since it includes the 8 week deadline and no margin on the specs. Anyone thinking this is a scam can make a few bitcoins by betting this:

http://betsofbitco.in/item?id=141

Goat, Im sure you will put 100 or more BTCs in there, there is no way you could lose that bet.

Inaba

legendary

Activity: 1260

Merit: 1000

So here is the first BFL Escrow address: 1CzKrTnncqVuviR9iyBBL5KBJkHUtv4Sjp

If anyone takes an initial wager, let me know how much you are sending. If someone else wants to bet against it, I will generate a new address for that.

DeathAndTaxes

donator

Activity: 1218

Merit: 1079

Gerald Davis

Quote from: punningclan on November 08, 2011, 08:04:24 PM

FPGA is used exclusively with the actual costs of the boards being designed somewhat equivalent. Mhash's are about a quarter of the Butterfly labs card though but wattage is similar, they do say they use a combination of ASIC and FPGA.

The wattage is not "similar". Other FPGA designs are getting about 20MH/W not 50MH/W. 250% improvement in electrical efficiency is massive.

Topic: 1GH/s, 20w, $500 — Butterflylabs, is it a scam? - page 47. (Read 123107 times)