Pages:
Author

Topic: 1GH/s, 20w, $500 — Butterflylabs, is it a scam? - page 17. (Read 123109 times)

hero member
Activity: 546
Merit: 500
Any news from Inaba ? The live test should be running by now...  Tongue

Very eager to see what Inaba has to say.  Sadly I expect him to report that there were "problems" that prevented him from seeing advertised performance (or any performance) and they will need to reschedule.
legendary
Activity: 1666
Merit: 1057
Marketing manager - GO MP
Well the Asynclab products look at least genuine, from common sense.
Arduino clones, even fancy ones are all around.
legendary
Activity: 1022
Merit: 1000
BitMinter
Any news from Inaba ? The live test should be running by now...  Tongue
hero member
Activity: 504
Merit: 500
Question: Why is butterflylabs using the same logo as asynclabs which subsequently closed this year? Coincidence?

Both created in 2009, both WHOISs are blocked.

  Very interesting;  Though the logos are different. Very similar though. Gonna go poke through it and see if there is anything revealing..




 Edit1; no related technologies from what I can see..
 Edit2; Async is based out of San Jose. https://github.com/asynclabs
 Edit2; And closed with this statement, "I'm sure that at some point in the future I'll be back at this again, tinkering my way to something else interesting, but for now, my family needs some of my attention (as well as my day employer )."  http://asynclabs.com/forums/viewtopic.php?f=6&t=681

  Edit3; Logo Comparison;
           Async                        BFL
donator
Activity: 1218
Merit: 1079
Gerald Davis
2. should we be scared ? I certainly am.

No.  Even w/ GPU one can 51% attack the network w/ roughly 20M invested.  This brings the cost closer to 10M (be sure to include labor, warehouse space, electricity, cooling, security, etc).

It simply isn't worth it for someone to spend that kind of money to destroy Bitcoin.  20M or 10M.  If/when Bitcoin becomes larger the network will hopefully be even larger thus as the potential of Bitcoin rises the cost to destroy it rises also.

No blockchain will ever be immune to the threats of governments.  Even if the cost is $100M it is "nothing".  The US has wasted something like $500,000 million on the three decade long failed war on drugs and almost $1,00,000 million on the war against terror.  (Side note wars against nouns are usually futile).

The point of the high cost of a block chain is to make it ECONOMICALLY NONVIABLE for an attacker to spend the amount of money necessary to destroy the network.  It isn't a magic gain auto and unlimited immunity against every govt on the planet.
rjk
sr. member
Activity: 448
Merit: 250
1ngldh
Where's the demo please, Inaba?

EDIT: I guess I should calm down  Grin
It has only been 3/4 of an hour since he said he would commence testing.

*jumps in anticipation*
hero member
Activity: 518
Merit: 500
I guess the two questions we should be asking are these :

1. are the quoted figures ( price, power, performance ) possible and achievable with a current sASIC or FPGA out in the market right now ( eg Hardcopy )

2. should we be scared ? I certainly am.

Think about it. A startup like BFL Labs can come up with a custom ASIC or a sASIC or whatever new revolutionary hashing technology and do a 51% on the network with just 4-5 million USD ( eg current network of slow GPUs is about 8000 ghash and one rigbox supposedly does 50 ghash and costs 25000 USD ). I wonder what the US government could come up with to take us out, spending less than 10 million in the process ( bet they have spent much more than that fighting the war on drugs etc. ). In any case I still think it would be cheaper for them to drive the price under $1 and every miner quits then they attack with server farms etc. but my point still stands.

My point is that the network right now is mostly GPU while in the past was mostly CPUs. If somebody makes a custom ASIC / product like BFL Labs that gets each unit minimum of 1 Ghash/s until we all upgrade to that technology the network is vulnerable to a 51% by the guys with the new tech etc. Can someone confirm ? Thanks. 
hero member
Activity: 588
Merit: 500
Hero VIP ultra official trusted super staff puppet
Question: Why is butterflylabs using the same logo as asynclabs which subsequently closed this year? Coincidence?

Both created in 2009, both WHOISs are blocked.
hero member
Activity: 504
Merit: 500

What is going to happen if this is indeed real ? Will all the other FPGA garage designers quit and let BFL have a hardware monopoly etc. ?

  Not much. The guys that spend big bucks now to buy high end GPU's will spend big bucks to buy these. The current devs that are approaching their developments from a business perpective will adapt, modify and move on from there.

  It would be nice if the release of such a device would bring all the hardware devs in the community together to design their own competing product.  If you guys do I'll manage the IPO for you.  Grin I've got almost 0 experience in releasing my own but can't count the number of Bplans I have drawn up for others.. >.<
donator
Activity: 1218
Merit: 1079
Gerald Davis
U5 on that board is atmel AT45DB642D, a 2.5V dataflash.
for the FPGA config steam storage? (so not hard-copy?)
I had a feeling it was going to be something like that but couldn't quite read the markings. Almost certainly bitstream storage for an FPGA of some kind then - nothing else would need that much storage.

So is this or is this not using that Altera hardcopy chip ?

Could other people also use that magical altera sASIC chip or the $$$ needed is just too much for the average bitcoin fpga designer etc. ?

What is going to happen if this is indeed real ? Will all the other FPGA garage designers quit and let BFL have a hardware monopoly etc. ?

A dataflash chip would indicate it isn't an sASIC.  sASIC are mask programmed at the fab not flash programmed in the field.  Maybe the chip is used for something else but its presence indicates an FPGA under the heatsink.
hero member
Activity: 518
Merit: 500
U5 on that board is atmel AT45DB642D, a 2.5V dataflash.
for the FPGA config steam storage? (so not hard-copy?)
I had a feeling it was going to be something like that but couldn't quite read the markings. Almost certainly bitstream storage for an FPGA of some kind then - nothing else would need that much storage.

So is this or is this not using that Altera hardcopy chip ?

Could other people also use that magical altera sASIC chip or the $$$ needed is just too much for the average bitcoin fpga designer etc. ?

What is going to happen if this is indeed real ? Will all the other FPGA garage designers quit and let BFL have a hardware monopoly etc. ?
hero member
Activity: 686
Merit: 564
U5 on that board is atmel AT45DB642D, a 2.5V dataflash.
for the FPGA config steam storage? (so not hard-copy?)
I had a feeling it was going to be something like that but couldn't quite read the markings. Almost certainly bitstream storage for an FPGA of some kind then - nothing else would need that much storage.
legendary
Activity: 915
Merit: 1005
Hmm, this is getting better. Looking forward to the results
rph
full member
Activity: 176
Merit: 100
Don't get fooled by xilinx S6 perf/power, those are the worst of all 45/40nm FPGAs by a long shot; we pretty much only use them because they're cheap-ish and readily available in small qty.

S6 is worse than V6 on perf/W, but it's still optimal on perf/$ even in qty 100+.

A7 should be a very good fit when it arrives

-rph
hero member
Activity: 518
Merit: 500
I say we probe the chip by JTAG and see what it is so we know if they can really pull this off.

Time for demo is still Sunday at midnight GMT right ? Thanks.
hero member
Activity: 592
Merit: 501
We will stand and fight.

Yeah, that's pretty much what I figured; just the messy routing on those alone has got to cost a fair amount in power consumption.

but i noticed on virtex-6 there are still very high I-cores. Huh


ADD:

U5 on that board is atmel AT45DB642D, a 2.5V dataflash.
for the FPGA config steam storage? (so not hard-copy?)
hero member
Activity: 686
Merit: 564
As far as I can tell that would require a lot more software work. Like building a program for the Atmel chip that fetches work. I think this would, somewhat, be a waste of time, as this is what ckolivas has spent so much time perfecting with cgminer. Also, is there a TCP/IP implementation for this Atmel chip even? Or how'd they actually send packets with the PHY, let alone run a mining application on the board?
There are platform-independent open source embedded TCP/IP stacks out there, the most widely used of which is uIP. I've been meaning to write some kind of embedded miner for them for a while. Obviously USB or something similar would still be needed as an option for people that want to run large numbers of miners, though.

Not sure on power, someone bored enough to synthesize a design and run power estimations?
Don't get fooled by xilinx S6 perf/power, those are the worst of all 45/40nm FPGAs by a long shot; we pretty much only use them because they're cheap-ish and readily available in small qty.
Yeah, that's pretty much what I figured; just the messy routing on those alone has got to cost a fair amount in power consumption.
hero member
Activity: 504
Merit: 500
  I'm going out to buy a shirt tommorow that reads,

      I'm
     With
    Smartie
  <--v^-->

  Thanks for your information, Art, Ng, Gmax, and Big-Chip
donator
Activity: 980
Merit: 1004
felonious vagrancy, personified
You'd agree what we could probably identify (b),(c) from the power usage though, no?

Early board photos showed an exposed JTAG header.  Just hook it up and query for IDCODEs.

If they did a custom chip it can obviously ignore the IDCODE request or lie, but if they're using a commercial FPGA they can't change this behavior.  The chip will tell us what it is.
Pages:
Jump to: