Bitfury: "16nm... sales to public start shortly"

[philipma1957]

<snip>
Btw bitmaintech is taking money for presales as I do not think they have shipped since February

Technically you could call it pre-sale but the only reason is due to their time off for CNY. I see no reason to think the gear won't ship on time. Must be nice to have a gov mandated 15 days off...  Convenient for them no doubt but the units for batch 10 are supposed to be shipping now. I expect to get an email from Bitmain about my 2 batch 10's any day now.

I am glad you bring this up.  So far I have yet to hear or read any info on batch ten shipping.

[1715033417]

1715033417

[1715033417]

1715033417

[RoadStress]

Not quite sure why everyone is surprised that a Manufacturer does not deal with individual customers? It's the same with most everything else we buy from Cars to Televisions. I assume that Dealers & Retailers will emerge for us to buy from?


Rich

When the Manufacturer claims that he is trying to help with decentralization, I do expect that he will deal with individual customers. Considering the variety of members I don't think that you can get better decentralization than here. But when the Manufacturer prefers to deal only with large customers I can only say that they have LIED.

[NotFuzzyWarm]

Not quite sure why everyone is surprised that a Manufacturer does not deal with individual customers? It's the same with most everything else we buy from Cars to Televisions. I assume that Dealers & Retailers will emerge for us to buy from?
Rich

When the Manufacturer claims that he is trying to help with decentralization, I do expect that he will deal with individual customers. Considering the variety of members I don't think that you can get better decentralization than here. But when the Manufacturer prefers to deal only with large customers I can only say that they have LIED.

Won't argue with that in the least.
If they don't want to deal with individual sales then setup vetted distributors to take the hassle of dealing with The Public. There are reputable folks here itchin' to do that with the BTC in-hand to do some decent size outright chip buys or-- AFTER proof of them working outside of a lab -- handle group-financed buys.

Not to mention, um privately let a few folks here who know what they are doing have a few engineering/sales samples along with chip data sheets and any other useful tips/info that would come in handy. Enough so they can initially at least make a full-up POC 2-4TH/s miner for the Community. As in most likely those involved eventually releasing all final design data or at least enough for DIY's to play with after fair recompense for time/effort/blood, sweat and tears involved in the process is garnered though sales..

[baazju]

Everyone seems to talk about decentralization and how much they care about bitcoin, but then do the opposite. One of the main reasons I mine with a certain pool and not others.

[lightfoot]

The new chip/miner design is so simple, that even a DIY at home is very easy! That's why Bitfury are so sure, there will be miners available at the end of March, just because there is not much to do.
Do your kids like LEGO? Do you want a LEGO type miner? Just put together the parts and start mining - even a kid could do it!

Hm. What are the chip pinouts, protocols, and voltages required to fire it? I may have just the board to put 8 of these chips on............

[dropt]

Everyone seems to talk about decentralization and how much they care about bitcoin, but then do the opposite. One of the main reasons I mine with a certain pool and not others.

Few people in this sub care about Bitcoin, most just care about making money.

[sidehack]

Most, but definitely not all.

[NotFuzzyWarm]

Hm. What are the chip pinouts, protocols, and voltages required to fire it? I may have just the board to put 8 of these chips on............

As far as I know BitFury has not even released said info for their 28nm chip. (If I'm wrong, someone please to point where it is...) Even as a preliminary-spec sure as hell aren't going to release info on the 16nm chip...

[Mitak]

Everyone seems to talk about decentralization and how much they care about bitcoin, but then do the opposite. One of the main reasons I mine with a certain pool and not others.

Few people in this sub care about Bitcoin, most just care about making money.

Very well said.
And it is that way for let's say last 3 years.

[Mitak]

Hm. What are the chip pinouts, protocols, and voltages required to fire it? I may have just the board to put 8 of these chips on............

As far as I know BitFury has not even released said info for their 28nm chip. (If I'm wrong, someone please to point where it is...) Even as a preliminary-spec sure as hell aren't going to release info on the 16nm chip...

They are not going for 28 nm, it is officially stated, as they are not going to sell it to general public.
For the 16 nm , they have docs and they provide it to 1M MOQ customers AFIK

[NotFuzzyWarm]

Hm. What are the chip pinouts, protocols, and voltages required to fire it? I may have just the board to put 8 of these chips on............

As far as I know BitFury has not even released said info for their 28nm chip. (If I'm wrong, someone please to point where it is...) Even as a preliminary-spec sure as hell aren't going to release info on the 16nm chip...

They are not going for 28 nm, it is officially stated, as they are not going to sell it to general public.
For the 16 nm , they have docs and they provide it to 1M MOQ customers AFIK

Of course they aren't selling 28nm anymore. But the website on their 16nm chips talks of using the same communications as their prev gen (28nm) chip. So... why not at least release that info for folks to chew on until the 16nm chip is seen running in quantity outside of a lab or trade show demo.

And therein lies the problem.
a. How many chips does that buy and what is the delivery schedule?
That right there is main thing that matters to anyone considering designing/building a miner.

b. Very few if any in the Community here at least  -note the big 'C' - have that kind of disposable coin to fill the MOQ. Those that do most likely will not run it as an eventual open-source venture.

[QuintLeo]

Avalon uses a third-party reseller (at least for orders outside China, not sure about inside) with a MOQ of 10 units. Not openly small-customer hostile, but not exactly friendly either.

I know the A2 is an Innosilicon chip, but is the Terminator an Innosilicon product or built by a third party? Also, pretty sure that's a scrypt miner so not bitcoin (just covering technicalities).

 The A2 Terminator and the Farm Boy were sold by Innosilicon, though some of the design of the Terminator may have originated from Lketc.
 Yes, the A2 chip is Scrypt - but nobody specified "Bitcoin miners" in the "only Bifmain" statement I was responding to.





 Based on hashrate the last couple days, it looks like Bitmain appears to be shipping SOMETHING. Might be someone else, but I've not seen any reports at all yet of B-Elevens from BW.com for sale yet, and Avalon isn't shipping enough units to affect the hashrate noticeably.

[KNK]

Hm. What are the chip pinouts, protocols, and voltages required to fire it? I may have just the board to put 8 of these chips on............

There is no official info yet, but most of your questions can be answered from the videos posted on youtube and the PDFs

[kilo17]

I think it may be a bit more complicate then legos . I cannot say much about board design but would like to see what one of the guys like SideHack, NotFuzzyWarm etc say about the layout for the new chips, in regards to what appears to be extremely fine tracing routes and how to deal with the heat generated on an air cooled PCB

[NotFuzzyWarm]

I think it may be a bit more complicate then legos . I cannot say much about board design but would like to see what one of the guys like SideHack, NotFuzzyWarm etc say about the layout for the new chips, in regards to what appears to be extremely fine tracing routes and how to deal with the heat generated on an air cooled PCB

Er, SideHack and I were just discussing them early this morning... Get out of our minds!!!
the pic of the  BitFury 48 chip board is pretty but considering all those traces are address and data coms lines that is just begging for trouble from crosstalk and depending on the actual speed the coms to each chip are, signal reflection. Problematic but usable with careful termination at both ends of the traces but shorter is better and easier to push to higher speeds.

Think it is the pic of the 12v boards that looks far better. Coms multiplexers in the middle of chip groups keeping the data lines short to/from the ASICS.

As for air cooling.. so far pics from like the B-11 are showing just backside cooling. History repeating itself? We've see how long that lasts. Soon direct contact topside cooling of the chips is showing up.... Now if they can get away with it and the chips are actually only slightly warm to the touch with just cooling through a big thermal resistor called the circuit board -- then I'm impressed. Until then...

Oh, KNK: Thanks for repointing me to the PDF's (again) http://dl.bitfury.com/28nm/pdf/ dunna know how I missed them. Leave a lot out to really do it such as coms protocols but also answered a few question I had about the BitFury boards - mainly WTF was that big loop structure? Is RFID tag to identify boards in the tanks. or at any other stage when board ID is needed.

[2112]

Er, SideHack and I were just discussing them early this morning... Get out of our minds!!!
the pic of the  BitFury 48 chip board is pretty but considering all those traces are address and data coms lines that is just begging for trouble from crosstalk and depending on the actual speed the coms to each chip are, signal reflection. Problematic but usable with careful termination at both ends of the traces but shorter is better and easier to push to higher speeds.

Yeah, better be safe than sorry. To my eyes the board looks only mildly complicated with respect to transmission line problems.

Compare it with a rather low-end FPGA kit adapter from Xilinx; look at the snaking traces between JX1 and JX2.



Can anybody here make an educated guess on how many revisions it took Bitfury to come up with their published design? It says "V3.1", what could that mean?

Is anyone here familiar with any commercial software for frequency-sensitive PCB design? I only used custom Fortran software. Bitfury himself posted some links years ago:

Edit:

http://home.educities.edu.tw/oldfriend/article/PI/Power_Plane_and_Decoupling_Optimization.pdf
http://home.educities.edu.tw/oldfriend/article/PI/PI%20and%20GND%20bounce%20sim.pdf
http://home.educities.edu.tw/oldfriend/Tutorials/Ansoft/Ansoft%20solution%20review.pdf

Edit2: At least those 3 links still used to work last year.

Edit3: Original bitfury's post from 2013:

https://bitcointalk.org/index.php?topic=183368.msg2266329#msg2266329

[NotFuzzyWarm]

Software no but I'll see what links to design articles re high-speed design and PDN considerations to watch for I can pull off of my computer at work manyana..

My guess is that the 'snakey' lines are parallel coms of some sort and the bendy bits are to make sure the traces are the same length so signal propagation times are the same on all lines. Also note the lack of 90-deg corners aside from in the bendy bits. With 90-deg corners a trace is wider across the diagonal of the corner and that creates an impedance disruption in the transmission line at those points.

EDIT: As promised
Damn good one on PCB routing and how it has changed http://electronicdesign.com/what-s-difference-between/what-s-difference-pcb-routing-then-and-now

Reflections and transmission lines http://www.eetimes.com/document.asp?doc_id=1280937

On EMI in general http://www.eetimes.com/document.asp?doc_id=1280798

On PDN design considerations http://powerelectronics.com/power-electronics-systems/five-things-every-engineer-should-know-about-pdn

Buck converter design http://powerelectronics.com/dc-dc-converters/buck-converter-design-demystified

Another on EMI in general http://machinedesign.com/motion-control/eliminating-emi-motion-systems

As for articles on stripline layout tips.. most of what I know is from decades of reading on it/working with it.

[Mitak]

Great resources
But I actually do not think the sha 256 miner signalling is so sensitive.
From my experience just following the general PCB signalling rules is ok

[jstefanop]

My guess is that the 'snakey' lines are parallel coms of some sort and the bendy bits are to make sure the traces are the same length so signal propagation times are the same on all lines. Also note the lack of 90-deg corners aside from in the bendy bits. With 90-deg corners a trace is wider across the diagonal of the corner and that creates an impedance disruption in the transmission line at those points.

Yes if you see any "snake" type traces its because they are parallel IO traces to some other longer traces, so they need to make the traces the same lengths hence the snakes. If you check out your PC motherboard you'll see them all over the place on the PCIe and RAM lanes. When your talking about throughputs in the muti Gigabits/s even the speed of light works against you

This does not matter on anything bitcoin related though, since these are very dumb chips IO rates coming to/from them are in the order of sub 1Mb/s, since nothing larger than a couple hundred bytes is ever sent to the chips( and return IO is even less...just an 8 byte nonce return and maybe a few bytes of chip info data).

[2112]

Yes if you see any "snake" type traces its because they are parallel IO traces to some other longer traces, so they need to make the traces the same lengths hence the snakes. If you check out your PC motherboard you'll see them all over the place on the PCIe and RAM lanes. When your talking about throughputs in the muti Gigabits/s even the speed of light works against you

This does not matter on anything bitcoin related though, since these are very dumb chips IO rates coming to/from them are in the order of sub 1Mb/s, since nothing larger than a couple hundred bytes is ever sent to the chips( and return IO is even less...just an 8 byte nonce return and maybe a few bytes of chip info data).

You completely forgot about SSN (Simultaneous Switching Noise).

The Enterpoint's Cairnsmore is the best example. It was designed by a professional FPGA board designer, yet it failed to properly distribute clock from the single Spartan 3 array controller to the four Spartan 6 mining chip. The guy who finally developed the working bitstreams had to do a lot of trial & error before he managed to squeeze a respectable output from that board. The competing 1.15y board from ZTEX did not exhibit those problems.

Even a simple, but regular and symmetric multichip design can produce hard to suppress resonances.

I applaud sidehack & friends for being careful. If they don't already have access to the appropriate analog models and software it is a very good decision to produce small and conservative design at first.

Edit: Being conservative is especially important with the attitude towards the integrator's designers exhibited by Bitfury and especially punin in the nearby thread. I presume that Bitfury doesn't even have proper IBIS models that would facilitate high-performance board design.

Edit2: grammar & spelling fixes