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

[DeathAndTaxes]

oh, sorry sir.
i think i made some math mistake at the "POLL: Miners, do you pay for electricity?  " thread, no 25MH/W, i test them again today, and it's about 18-20MH/w as i say on the main thread : "each board has 2 XC6SLX150 -2FGG484I on it, generates a 360MH/s hashing power. 19.5W on wall power consuming."

No problem.  Thanks for the correction I will update my post.  I like your business ethics more and more each day.

[Inaba]

To Inaba:

I know you can't give exact numbers but can you give us more of a ballpark.

When running stable (no false hashes) was performance in the 900 to 1000 MH range or 600 to 800 or <600 MH?

When you say power was more but not 200W well now FPGA on the planet (not even one for 10 years ago) uses 200W.  Hell most CPU don't use more than 200W.

So is "more power" =
still <30W?  
30W - 40W?  
>40W?

Less hashes & more power could mean anything from slightly worse than expected but still decent to worse than products already on sale by other manufacturers. I know you are prohibited from giving exact numbers but not providing a firmer ballpark range on stats does a disservice to the 3+ FPGA developers who have always supported the Bitcoin community and have real products w/ real verified results.  They have had sales impacted as people wait for results from this magical unicorn.  If you feel you can be firmer but not exact and still meet the spirit of your agreement that would be the right thing to do.

Can someone run down the current MH/s and power draw of the current offerings?  I can give you an idea of how it compares when compared to what's currently available for sale.  From what I know (which may be inaccurate, which is why I ask for the run down) the BFL offering exceeds everything available that I know of by a substantial margin in hashing power (at stable speeds), although I'm not sure about power draw.

[xaxik]

Someone asked that I measure the actual chips - they are 30mm^2.

Thanks Inaba.
Are they exactly 30 or rather 29mm^2?
Some chips from altera are 29mm and some are 31mm...
I'm just curious, what may be unter the hood heatsinks 

[Inaba]

xaxik - Sorry I wasn't that exacting... I will see what I can do to get an exact measurement, but it's kind of hard to get at the chips with a ruler with the caps and such blocking the way.

[DeathAndTaxes]

Can someone run down the current MH/s and power draw of the current offerings?  I can give you an idea of how it compares when compared to what's currently available for sale.  From what I know (which may be inaccurate, which is why I ask for the run down) the BFL offering exceeds everything available that I know of by a substantial margin in hashing power (at stable speeds), although I'm not sure about power draw.

ztek board is 190MH (looks like he pushing that to 192MH by improving power).

Prices are:
1-4 units: 327 EUR (about 460 USD)
5-9 units: 305 EUR (about 430 USD)
10-24 units: 282 EUR (about 395 USD)
25-49 units: 259 EUR (about 365 USD)
50-99 units: 236 EUR (about 330 USD)
100+ units: 213 EUR (about 300 USD)

License production programs can be offered too. The customer would purchase the empty PCB including a license fee and gets assembly data (stencil data, bill of material, pick and place data, ...). Estimated costs (including PCB, license fee, parts, assembly) based on the prices here in Germany (from official Xilinx Distributor) would be:
100 units: 170 EUR (about 240 USD)
250 units: 140 EUR (about 195 USD)

ngzhang board is 360MH (he indicated 380 MH is possible but it currently runs hot)
price is $590.

rph is experimenting w/ simplified board design targeting $1/MH but currently isn't available

All the boards get ~20MH/W.

[DeepBit]

ngzhang is advertising his boards as 360 MH, not 200. May be you were thinking of ztex ?
And ztex is advertising his boards at ~1.6$ per MH, not $1 per MH (100 unit order)

ztek offers lower price is the buyer provides all capital.  A licensed production run is $1.26 per MH (in 100 board runs) and $1.02 per MH (in 250 board runs).

May be I misread his post, but I think that "licensed production run" means "no assembly included", so the price will be higher to the end-user.

[ngzhang]

To Inaba:

I know you can't give exact numbers but can you give us more of a ballpark.

When running stable (no false hashes) was performance in the 900 to 1000 MH range or 600 to 800 or <600 MH?

When you say power was more but not 200W well now FPGA on the planet (not even one for 10 years ago) uses 200W.  Hell most CPU don't use more than 200W.

So is "more power" =
still <30W?  
30W - 40W?  
>40W?

Less hashes & more power could mean anything from slightly worse than expected but still decent to worse than products already on sale by other manufacturers. I know you are prohibited from giving exact numbers but not providing a firmer ballpark range on stats does a disservice to the 3+ FPGA developers who have always supported the Bitcoin community and have real products w/ real verified results.  They have had sales impacted as people wait for results from this magical unicorn.  If you feel you can be firmer but not exact and still meet the spirit of your agreement that would be the right thing to do.

Can someone run down the current MH/s and power draw of the current offerings?  I can give you an idea of how it compares when compared to what's currently available for sale.  From what I know (which may be inaccurate, which is why I ask for the run down) the BFL offering exceeds everything available that I know of by a substantial margin in hashing power (at stable speeds), although I'm not sure about power draw.

you mean how to test the power draw?

i use this one:



5 boards running @ max power

[DeathAndTaxes]

ngzhang is advertising his boards as 360 MH, not 200. May be you were thinking of ztex ?
And ztex is advertising his boards at ~1.6$ per MH, not $1 per MH (100 unit order)

ztek offers lower price is the buyer provides all capital.  A licensed production run is $1.26 per MH (in 100 board runs) and $1.02 per MH (in 250 board runs).

May be I misread his post, but I think that "licensed production run" means "no assembly included", so the price will be higher to the end-user.

No I clarified w/ him via PM.  The price is an estimate (using component prices @ prices for assembly house he uses in Germany)

per the quoted text:

Estimated costs (including PCB, license fee, parts, assembly) based on the prices here in Germany (from official Xilinx Distributor) would be:

You would only pay a small fraction of that to him and he will give you 100/250 blank PCB, license to run 100/250 of them, and all necessary datafiles for assembly house to build the boards.

When I asked him why price is lower the answer was obvious (in hindsight).  He doesn't need to invest the capital, build a large number of boards upfront and take the risk/stress of trying to sell them all so he is willing to accept a lower profit and you also get the lower part/assembly costs of building 100/250 units at once.

[Inaba]

Well stable, the BFL unit is > 4x the hashrate than ztek and more than double ngzhang. Power draw is also more than double ngzhang (Dunno what ztek power draw is).  Even allowing for a 10% efficiency decrease for "real world" scenario vs the test data, the numbers still hold.

[DeathAndTaxes]

Well stable, the BFL unit is > 4x the hashrate than ztek and more than double ngzhang. Power draw is also more than double ngzhang (Dunno what ztek power draw is).  Even allowing for a 10% efficiency decrease for "real world" scenario vs the test data, the numbers still hold.

Ztek, rph, and ngzhang's boards all have roughly the same power effciency ~20MH/W (+/-10%) which is mainly driven by the power demand of the FPGA.

It would seem this board has similar power efficiency?  I would just like to restate for my record it was way back on page 1 that I indicated that 50MH/W was implausible for a FPGA (45nm).  That is what lead to all the speculation of sASICS and the likelihood of scam given the costs involved w/ sASIC development, etc.

Just to clarify you were comparing the 0 error hashrate not the higher 50% error hashrate right?

Thanks for providing more ballpark details.  

What is somewhat alarming is not getting performance specs wrong. Shit happens but how does a company get the claimed wattage that wrong.  I mean if you are using FPGA x it's power draw is not going to vary by +/- 80%.   So what performance you get out of the chip may vary by bitstream, cooling, power regulation, the particular FGPA used, speed rating, etc the wattage shouldn't be that variable.

[Zotia]

What will the final odds be I wonder?

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

If they don't close the bet soon, it will go to 10,000 BTC to 112.71 BTC.  What a waste.

I put 75 BTC on the bet while there were only small bets on each side -- giving the believers 5:1 odds.  Now that the test results are in, someone put about 120-150 BTC on the bet and they will be receiving a similar amount of the winnings as me.



Someone really needs to make a better betting site.

[sadpandatech]

ngzhang is advertising his boards as 360 MH, not 200. May be you were thinking of ztex ?
And ztex is advertising his boards at ~1.6$ per MH, not $1 per MH (100 unit order)

ztek offers lower price is the buyer provides all capital.  A licensed production run is $1.26 per MH (in 100 board runs) and $1.02 per MH (in 250 board runs).

May be I misread his post, but I think that "licensed production run" means "no assembly included", so the price will be higher to the end-user.

No I clarified w/ him via PM.  The price is an estimate (using component prices @ prices for assembly house he uses in Germany)

per the quoted text:

Estimated costs (including PCB, license fee, parts, assembly) based on the prices here in Germany (from official Xilinx Distributor) would be:

You would only pay a small fraction of that to him and he will give you 100/250 blank PCB, license to run 100/250 of them, and all necessary datafiles for assembly house to build the boards.

When I asked him why price is lower the answer was obvious (in hindsight).  He doesn't need to invest the capital, build a large number of boards upfront and take the risk/stress of trying to sell them all so he is willing to accept a lower profit and you also get the lower part/assembly costs of building 100/250 units at once.

   Aye, the only catch to that is good fuggin luck finding the LX150-N3 here in the US for $100 per chip. hell, even $120 per chip is unlikely. Best quote from Avnet is about $141 with a 6 week lead time from Xilinx. >.<

[DeathAndTaxes]

What will the final odds be I wonder?

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

If they don't close the bet soon, it will go to 10,000 BTC to 112.71 BTC.  What a waste.

I put 75 BTC on the bet while there were only small bets on each side -- giving the believers 5:1 odds.  Now that the test results are in, someone put about 120-150 BTC on the bet and they will be receiving a similar amount of the winnings as me.



Someone really needs to make a better betting site.

+1. 

[Inaba]

Yes, those numbers are from the 0 error rate firmware not the unstable higher hashrate firmware.

For the record, I think they should probably cancel that bet and refund it at this point, since it's such a cluster fuck

[rjk]

Question: what are smaller (faster) processes than 45nm? 32, 28, 22?

What is the expected hash rate and power draw from simulations of some of these? It seems that everything is assumed to be 45nm because of cost, but what if...?

[fizzisist]

If they don't close the bet soon, it will go to 10,000 BTC to 112.71 BTC.  What a waste.

I put 75 BTC on the bet while there were only small bets on each side -- giving the believers 5:1 odds.  Now that the test results are in, someone put about 120-150 BTC on the bet and they will be receiving a similar amount of the winnings as me.

Someone really needs to make a better betting site.

When I checked betsofbitco.in a few hours before Inaba's post, the tally was about 275 to 111, so those big bets came in before that. Maybe it was Inaba himself? 

They said before that they will refund bets made after the outcome is determined:

If there is an official announcement and confirmations, we will close betting and cancel those large bets done after the announcement. Betting deadline is not strict. We'll try to protect you as much as we can not to dilute your initial bets. Let me know if you have any concerns.

BTW, for those of you who are not following, the bet reached almost 250 BTC. This is probably the largest public bitcoin bet ever:
http://betsofbitco.in/item?id=141

Make that 400 BTC now... crazy!

[P4man]

Question: what are smaller (faster) processes than 45nm? 32, 28, 22?

What is the expected hash rate and power draw from simulations of some of these? It seems that everything is assumed to be 45nm because of cost, but what if...?

There are no FPGAs (or much else) available on nodes smaller than 45nm. Its basically only intel and AMD shipping 32nm products, and those are CPUs only.  28nm FPGAs is for next year.

I still wouldnt rule out (65nm ?) s-asic's though.  If it turns out to be off the shelve FPGA's Id really like someone to run some simulations for an altera hardcopy implementation. I know enough people able to invest the kind of money it would require.

[coinjedi]

I see that the demo didn't meet the specifications, but the official statement from BF labs is still that this was a tuned down unit. So at this point I will not close the bets.

Various people suggested switching to intrade type of betting. We believe that it is not suitable for sustaining many small statements. Go ahead and look at intrade. Even they have many empty markets. It is almost impossible to keep every market liquid in that type of betting and illiquid markets are useless. Whereas we can offer statements where only two users bet against each other. Saying that we are considering contract trading where we keep the current betting style and let people trade their bets either if they change your mind or use the advantage of early betting to sell for more.

[coinjedi]

For the record, I think they should probably cancel that bet and refund it at this point, since it's such a cluster fuck

Can you elaborate Inaba? I think the statement is very precise and the company still claims that they will meet the specifications. Did they say anything about the shipping date that I missed?
I know that tension is high due to so much BTC is being at stake but I am doing my best to be fair to both sides.

[P4man]

t is almost impossible to keep every market liquid in that type of betting

Im sure you could attract the bitcoinica folks - who might actually make  some money on average!