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

[runeks]

Seems like it's the same computer. Both computers have the same name (z61p-ThinkPad-W520) which isn't that strange since it's automatically chosen by Ubuntu (as far as I know). But they're also both logged in as the same user (z61p) which is a stranger coincidence.

My guess is there is a reasonable explanation for this, and we will probably get it soon.

[1714038545]

1714038545

[sadpandatech]

hmmm

[Inaba]

That is BFL's laptop, not mine.  The one I plan on using for the test is an HP that BFL has not seen/had access to yet.

[makomk]

That is BFL's laptop, not mine.  The one I plan on using for the test is an HP that BFL has not seen/had access to yet.

So let me get this straight... you wrote all that stuff about compiling the code from source and about how the board and test machine didn't have any kind of internet connection, and then you neglected to mention that you'd used a laptop set up and provided by BFL which could've had a hidden wireless connection to anywhere.  Did you even make sure that the binaries you were running actually corresponded to the source code you received, or did you use binaries that BFL provided too?

[Turbor]

We should be happy that Inaba does this whole test stuff.

[bulanula]

Maybe we can get somebody else from Kansas to confirm Inaba and BFL Labs that clearly has no affiliation whatsoever and maybe does not even know about BTC. Another MyBitcoin ? Maybe, maybe not.

[xzion]

That is BFL's laptop, not mine.  The one I plan on using for the test is an HP that BFL has not seen/had access to yet.

So let me get this straight... you wrote all that stuff about compiling the code from source and about how the board and test machine didn't have any kind of internet connection, and then you neglected to mention that you'd used a laptop set up and provided by BFL which could've had a hidden wireless connection to anywhere.  Did you even make sure that the binaries you were running actually corresponded to the source code you received, or did you use binaries that BFL provided too?

take it easy man, he's doing this for our benefit remember.
i'm sure when inaba has his own laptop he'll be able to give us some solid evidence.
in the mean time, chill out.

[Inaba]

That is BFL's laptop, not mine.  The one I plan on using for the test is an HP that BFL has not seen/had access to yet.

So let me get this straight... you wrote all that stuff about compiling the code from source and about how the board and test machine didn't have any kind of internet connection, and then you neglected to mention that you'd used a laptop set up and provided by BFL which could've had a hidden wireless connection to anywhere.  Did you even make sure that the binaries you were running actually corresponded to the source code you received, or did you use binaries that BFL provided too?

Neglected to mention it for what?  We haven't done the test yet...  How can I not mention something that hasn't even happened?

November 25th: https://bitcointalk.org/index.php?topic=48863.msg631047#msg631047

I just got home after meeting with Sonny in person from BFL and their engineer via Skype.  We had a fairly long conversation and I live over 30 miles away, so that's why it's taken me this long.

We did some simulated testing, but no live testing tonight.  We elected to put that off until tomorrow or possibly Sunday to allow for proper testing instead of just throwing some tests together.  I have, however, seen several units (and I will post pictures of them) and seen them power up.  Beyond that, I have nothing definitive to say, other than so far, so good.  I will be meeting with them again this weekend to conclude the testing on the live pool on an isolated network with no net access.

While I was not able to inspect the chips under the two main heat sinks, the other chips are plainly visible and inspectable.  There is nothing on that board to indicate any sort of cellular connection, for those of you suspecting that sort of nefarious scheme.  It was hard to take pictures at the location we were in and I will have better pictures once they let me take a unit over night - I will put it in my light box and get some properly lighted pictures.

November 25th: https://bitcointalk.org/index.php?topic=48863.msg631069#msg631069

I don't know why it says 0.  It wasn't a live test, so I wasn't too concerned, I was more interested in putting into a test loop to see what the power consumption was (which was 17.6W) - I took pictures of the Kill-A-Watt, but my flash blanked out the LCD screen of the Kill-A-Watt - I will get some non-flash pictures of it tomorrow or Sunday.

November 26th: https://bitcointalk.org/index.php?topic=48863.msg631620#msg631620

The demo plan is that I will:

A) connect it to a non-routable development side of the pool, so that the box is unable to communicate with the internet.  I will then let it submit shares to the pool and I will have one of the getwork servers in debug mode and I will see what is sent out and what's sent back.  As I found no evidence of any wireless communications on the board, and since the computer it's connected to will not be on the internet, it won't have any way of falsifying the shares submitted.

B) I will take a unit home that evening, disassemble and take more robust pictures.  I will NOT be removing any heat sinks, however.  

C) I will do further testing that evening on my own with a packet analyzer to see  what packets are being transmitted, when and where they are going and coming from.

That's the plan as of right now, at any rate.

So my plan was published after the first meeting and we haven't done a second, live test yet.  So I'm a little confused as to what you are taking issue with?

[makomk]

Neglected to mention it for what?  We haven't done the test yet...  How can I not mention something that hasn't even happened?

This looks very much like a test to me, even if it isn't the final test:

Sorry it took so long to get this post up, but I have a few other things that had to be taken care last evening and I didn't get home till late and pretty much went straight to bed.  At any rate, we did a small demo of the hardware last night, here is the test data I used, which I pulled from one of the getwork servers of my pool:

Data: fd90c721557226679bfc01bc971be894ec08137d0f36fd923f822e4743f954da
Merkle: 9d0e5b394ed6ae311a0f61b1

Data: e4f4a3eb23855f185379d5833f0eabb9daee8483e43d39a6a9b3888882bfc0fa
Merkle: 29a9690f4ed6aecc1a0f61b1

Both of these were fresh out of my pool:

The first test should return one nonce of: 22D5485D  
The second should return no nonce value.  

I would ask that someone else validate my test data as well, to be sure what I have is accurate.  

In any case, the first test ran at a hashrate quite a bit lower than projected, but returned the expected nonce on the first round and zero nonces on the second, which was an accurate result.  After this, I did in fact witness a faster test at a speed close to the projected rate, but it was slightly unstable, often returning extra nonces that are invalid.  The explanation I was given for this was plausible and equally plausible that they could work the issues out.  Power consumption for the development unit was substantially higher than projected, but still well within a reasonable amount (as in, it wasn't taking 200 or 300w; it was much, much less).

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

There were some technical difficulties that prevented us from conducting a more thorough test given the time constraints I was under - though BFL was more than willing to take whatever time was needed, even offering to come to my house this evening once they fixed the technical issue that prevented us from doing a comprehensive mining operation - though I am told it's fixed now and ready to go.  But in either case, I feel that the test we conducted showed a POC that adequately demonstrates that at least the hardware does what it's designed to do, if not at the speeds or power consumption stated at the moment.

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.

[Inaba]

There were some technical difficulties that prevented us from conducting a more thorough test given the time constraints I was under - though BFL was more than willing to take whatever time was needed, even offering to come to my house this evening once they fixed the technical issue that prevented us from doing a comprehensive mining operation

There we go, I bolded the relevant part of the quote you just quoted.  Some people are never happy. 

[ngzhang]

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.

it means that if i change Icarus's price to less than 350$ than BFL's life is over? 

ADD@2011/12/10

there must be soming misunderstanding here, i mean some other product(my me or other) reach <1$/M, not Icarus. sorry.

[kronosvl]

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.

it means that if i change Icarus's price to less than 350$ than BFL's life is over?  

more like 250$ if you want almost the same amount of $/MH

[rjk]

more like 250$ if you want almost the same amount of $/MH

But Icarus is open and has more ports and usefulness, so yeah...

[kronosvl]

But Icarus is open and has more ports and usefulness, so yeah...

Depends how many people buy the boards for mining vs people that buy for mining and/or using them for other purposes. Some people will not pay 100$ more for other features. just $/MH and MH/w will be taken into consideration. I could be wrong though

[ngzhang]

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.

it means that if i change Icarus's price to less than 350$ than BFL's life is over?  

more like 250$ if you want almost the same amount of $/MH

you mean 1.5MH/$?

yeah, it's extremely possible in the next 6-8week. 

[makomk]

There we go, I bolded the relevant part of the quote you just quoted.  Some people are never happy.

Personally I tend to think that the part just after that which you didn't bother to quote is kind of relevant too:

But in either case, I feel that the test we conducted showed a POC that adequately demonstrates that at least the hardware does what it's designed to do, if not at the speeds or power consumption stated at the moment.

Because it doesn't adequately demonstrate that after all - it doesn't even prove that the BitForce boards can calculate a single hash - and the only way we found out about this huge and unexpected flaw in the only third-party test so far was by accident. Bear in mind that for every day that passes more pre-order customers hit the 45 day limit after which Paypal won't give refunds if BFL don't deliver.

[BFL]

There we go, I bolded the relevant part of the quote you just quoted.  Some people are never happy.

Personally I tend to think that the part just after that which you didn't bother to quote is kind of relevant too:

But in either case, I feel that the test we conducted showed a POC that adequately demonstrates that at least the hardware does what it's designed to do, if not at the speeds or power consumption stated at the moment.

Because it doesn't adequately demonstrate that after all - it doesn't even prove that the BitForce boards can calculate a single hash - and the only way we found out about this huge and unexpected flaw in the only third-party test so far was by accident. Bear in mind that for every day that passes more pre-order customers hit the 45 day limit after which Paypal won't give refunds if BFL don't deliver.

The hostility is unnecessary.  If you have a purchase and are uncomfortable, simply ask for a refund.  If you don't and you're just a third party jerk, then with all due respect, mind your own business.  The heated passion is juvenile and irresponsible.  Get a grip.

Regards,
BFL

[Inaba]

Because it doesn't adequately demonstrate that after all - it doesn't even prove that the BitForce boards can calculate a single hash - and the only way we found out about this huge and unexpected flaw in the only third-party test so far was by accident. Bear in mind that for every day that passes more pre-order customers hit the 45 day limit after which Paypal won't give refunds if BFL don't deliver.

That would be fine if there had actually been a test of the system.  All I've seen so far is a POC and as stated, what I've seen so far had lead me to the conclusion (hence the "I feel that the test we conducted showed a POC that adequately demonstartes that at least the hardware does what it's designed to do...") that this merits further investigations and is not an out and out fraud.

But seriously, come on down to KC here and you are welcome to sit in on the next test and conduct it in a manner you think is the right way.  You conspiracy nuts can go piss up a tree at this point. I'm tired of trying to cater to the nutbag leanings that have absolutely no basis in reality in terms of time frames, business practices, or technical development.

If someone has something to provide as far as something specific to test or some specific methodology, please speak up and I'll be glad to include it assuming it's feasible.  Otherwise, I really don't care. I'm evaluating this on behalf of the community and for my own edification and decision on whether or not to sink thousands of dollars into specialized bitcoin equipment or maintain my GPU farms.  If the community doesn't want my evaluation that's not really a detriment to me - I will still evaluate my options and what I see here and base my personal purchase decision accordingly. 

If the time comes and I recommend buying if it meets your needs and my credibility is not adequate, then by all means DO NOT BUY IT. I've never advocated preordering; I haven't preordered myself as of yet.  But should these devices prove to be everything they are shaping up to be so far that I have seen, I will likely be ordering at least a few to either supplement or replace some of my rigs.  Whether or not you choose to buy is, was and will be completely up to you.

[makomk]

But seriously, come on down to KC here and you are welcome to sit in on the next test and conduct it in a manner you think is the right way.  You conspiracy nuts can go piss up a tree at this point. I'm tired of trying to cater to the nutbag leanings that have absolutely no basis in reality in terms of time frames, business practices, or technical development.

This might possibly be a reasonable response if the "but the computer could be connecting to mining hardware elsewhere" problem was just something I came up with after the test that you didn't think of beforehand. It wasn't. It was important enough that your original test plan went to great lengths to explain why this would be impossible - obviously you knew people wouldn't believe a test that didn't block this kind of outside communication. Yet somehow you concluded that their box could actually mine shares based on a proof of concept test that didn't exclude this potential cheating, and failed to mention the change so that people could judge for themselves how reliable the test was, instead leading us all to assume that the test proved more than it actually did.

[Inaba]

I made no such conclusion.  I said the hardware does what it's designed to do - namely process SHA 256 hashes.

That said, I posted the test data which I provided from my pool.  The nonces were returned within the expected time frame.  The laptop could not have processed and produced the nonces on it's own (and therefore is not a candidate for cheating) and I find it highly unlikely that any remote connection could:

1) Setup a connection to process the data
2) Remote system process the data
3) Return / receive the processed nonces

All in the roughly exact same amount of time as it would take a box hashing 4.2 billion nones should take at a given hashrate.  I suppose they could have a giant GPU farm somewhere (Say a 4 GH/s farm as a conservative estimate) that is custom programmed to break up a single data block into multiple work units and farm all of those out to the multiple GPUs to hash, aggregate that data back (from a minimum of 5 separate GPUs) and return the results... but could they do it in under 6 seconds (and in some cases 4 seconds) over a wireless or cellular link?  You tell me, since your technical knowledge is superior to my own.

But for my money, I'm going to say it would have been far harder and far more costly to FAKE the results than it would have been to actually bring a piece of hardware that does what it's designed to do.  Not to mention, after sinking all that time and money into faking it, they are still going to have to hand over a unit to me for isolated testing at some point.

Additionally, this was not a reliable test by any stretch of the imagination.  It was a POC, as stated.  A reliable test has not been conducted as of yet, which has also been stated numerous times.