1GH/s, 20w, $700 (was $500) — Butterflylabs, is it for real? (Part 2)

[btc_artist]

Looks like you get to keep the coins goat. Congrats I suppose, although all along my point was that BFL were not fake as you wanted to bet $10,000 on, so I do feel I was closer to the truth than you

They're looking more and more legitimate, but I don't think anything has been established yet one way or the other.

[1713572519]

1713572519

[BFL]

Also, If I might ask BFL a second question, how can 32 x 1.05 = 50+ ?    What is the average guy missing?

We've never said or implied that 32 Singles constitute a 50gh/s Rig Box.  The street brawl logic of the previous thread came to that conclusion on it's own and left little opportunity for correction.

The Rig Box product uses different chips than the Single.  That's the simplest explanation.  The 32 expansion points allow different generation cards to be used along with the core interchangeably.

Regards,
BFL

[P4man]

Actually, I went back to reread the bet.
here is how Inaba phrased it:

I can confirm there is a 2 BTC bet for:

2 months +8 days left now until bet default end. 25% spread on wattage, 25% spread on hash rate from the stated baseline of 1050MHs and 50w. Bet is met/lost either when BFL provide '*acceptable' proof or disproof of claim, or defaulted at the end of 2 months and 8 days if no unit is provided to verify claim.

https://bitcointalk.org/index.php?topic=48863.msg612602#msg612602

(my emphasis).

Im a bit surprised to see 50W there. But as its written, the bet is still open no? And to quote you:

Just keep in mind, it is now 2 months and 7 days for a product to be delivered to Inaba. He then has 1 month to test it. And then if it works, BFLabs has two months after that to sell it at the price stated. 20% or whatever error is fine. Shit happens and my claim is its a fucking scam, so if this is just a bit off, then you win.

Assuming there will be a 50W 1050MH device +-20% within the above timeframe, Ill leave it to inaba to judge. I honestly dont know if the above 50W was a typo or not, but no one corrected him.

[Inaba]

Actually, I went back to reread the bet.
here is how Inaba phrased it:

I can confirm there is a 2 BTC bet for:

2 months +8 days left now until bet default end. 25% spread on wattage, 25% spread on hash rate from the stated baseline of 1050MHs and 50w. Bet is met/lost either when BFL provide '*acceptable' proof or disproof of claim, or defaulted at the end of 2 months and 8 days if no unit is provided to verify claim.

https://bitcointalk.org/index.php?topic=48863.msg612602#msg612602

(my emphasis).

Im a bit surprised to see 50W there. But as its written, the bet is still open no? And to quote you:

Just keep in mind, it is now 2 months and 7 days for a product to be delivered to Inaba. He then has 1 month to test it. And then if it works, BFLabs has two months after that to sell it at the price stated. 20% or whatever error is fine. Shit happens and my claim is its a fucking scam, so if this is just a bit off, then you win.

Assuming there will be a 50W 1050MH device +-20% within the above timeframe, Ill leave it to inaba to judge. I honestly dont know if the above 50W was a typo or not, but no one corrected him.

50w came from this post/bet:

https://bitcointalk.org/index.php?topic=48863.msg612144#msg612144

[fizzisist]

Can the betting discussion be moved to another thread?

[makomk]

The static draw is under 19.8w

It would be, yes. Running a fast Bitcoin mining bitstream tends to result in quite a lot of dynamic power consumption though - apparently it has a higher toggle rate than normal applications of FPGAs or something. (This was documented in the various threads about it a while back.) I actually ended up having to upgrade the power supply on my toy 25 MHash miner to get it to run at its full potential without falling over!

We then have two different technologies that are used in our product.  One is primarily for the Single and the other is for the Rig Box.  Both can be interchangeable, but the more efficient system was specked out for the single where the 19.8 figure was established.  In the end we're going with the less efficient in initial production for the Single.

It may have been a good idea to avoid advertising the higher efficiency until you knew you'd actually achieve it.

[rjk]

BFL: I am assuming there isn't much processing needed on the host, which means that the only limitation is the 127-device USB cap? Do you have a "recommended specs" for the host computer? This seems perfect for my pico-itx test build.

[BFL]

The static draw is under 19.8w

It would be, yes. Running a fast Bitcoin mining bitstream tends to result in quite a lot of dynamic power consumption though - apparently it has a higher toggle rate than normal applications of FPGAs or something. (This was documented in the various threads about it a while back.) I actually ended up having to upgrade the power supply on my toy 25 MHash miner to get it to run at its full potential without falling over!

We then have two different technologies that are used in our product.  One is primarily for the Single and the other is for the Rig Box.  Both can be interchangeable, but the more efficient system was specked out for the single where the 19.8 figure was established.  In the end we're going with the less efficient in initial production for the Single.

It may have been a good idea to avoid advertising the higher efficiency until you knew you'd actually achieve it.

Agreed.  However, I'd like to take the opportunity to point out that we didn't start this thread or announce ourselves at all.  

We were discovered in mid development by a few in the bitcoin community...  that interaction altered our development course significantly.

In order to accommodate their friendly requests of early delivery units, we put up a pre-order page to keep it orderly.  That worked well, then at some point...  someone (gmax) heard about our project and posted...  the rest is history.   In sum, we weren't polished for commercialization because we hadn't expected to enter the market at the time our project was discovered.

[BFL]

BFL: I am assuming there isn't much processing needed on the host, which means that the only limitation is the 127-device USB cap? Do you have a "recommended specs" for the host computer? This seems perfect for my pico-itx test build.

Host computer requirements are nill.  If you can power a modern browser and you have a healthy USB port, you'll have more than enough performance capacity to host a chain of BitForce Singles.

BFL

PS.  Someone had speculated at some point that the Single itself could run an OS. Yes, it can.  However, in this version we allow the host miner to validate the nonce before network submital.  This is not resource intensive but it does require a host.

[O_Shovah]

It seems im becoming a collector
As my own solution is unlikely to come to a frutation i will watch the competitors.

So far ive got one copy of ethery FPGA or simmliar mining solution that has made it to the market.
I'll be sure to get at least one of you small ones.Altough i would love to get one of the rig boxes, but my student wallet will not allow that.

But if you seek testers..... 


I wish you good luck with you development

[eldentyrell]

God bets of bitcoin sucks.  ... if you are listening b.o.b. please consider moving to an intrade type format.

+1

We really need a bitcoin intrade.  B.O.B. is not a substitute.

[legolouman]

Curiously monitoring and posting in this thread.

[BFL]

I wish you good luck with you development

Thanks, I sincerely appreciate your encouragement.

BFL

[eldentyrell]

but it was slightly unstable, often returning extra nonces that are invalid

This is pretty bad news.

Keep in mind that (unless something truly weird is going on) that on average every invalid nonce returned by flaky hardware means that one valid nonce was missed.  The valid nonces are essentially randomly distributed (SHA-256 wouldn't make a good hash function if this weren't true) so the probability of a bit error causing an invalid nonce to look valid is equal to the probability of a bit error causing a valid nonce to look invalid.

So three bad nonces per good nonce means you are only going to see 25% of the expected hashrate.

[Turbor]

Can the betting discussion be moved to another thread?

+1

[coinjedi]

Goat, it's all but certain that we'll be above 19.8w in our final power specification.

Since BF Labs officially changed their claims, I'm closing the bets. Thanks everyone who participated.
We will make the final decision of winning side once their website is updated with new claims or at Jan. 3rd whichever happens first.

PS: I can get into my view of intrade type of markets, but I don't want to do it here. Cheers!

[Inaba]

Can you elaborate on what you mean by making a final decision once the website is updated?  What does that have to do with the betting decision?  I'm not being coy, I am honestly unclear on how they are related at this point?

[uck]

BFL, in  your demos planned for the future, please be sure to make some youtube videos showing the device running. This will help immensely.

Thanks

[coinjedi]

Can you elaborate on what you mean by making a final decision once the website is updated?  What does that have to do with the betting decision?  I'm not being coy, I am honestly unclear on how they are related at this point?

Before irreversibly moving couple hundred btc between accounts, I would like to wait for something more official. It seems that BF labs will update their specs when they are sure. Even if that doesn't happen I'll distribute the coins when the "event date" (Jan 3rd) comes.

[sadpandatech]

  In all fairness the title should be changed to $700 (was $600). As that was the initial price for units, not accounting for the 'discounted' preorders.....