The -2 and -3N are similar prices. If you don't need memory controller the main advantage of the -3N is that it offers slight better I/O speeds. However the logic delay which is a bit part of Bitcoin speed is the same between the -3N and -2. So for some applications the -3N might be slightly better but I don't think so for Bitcoin. Not anything worth talking about at least.

Our choice is based on the volume we use -2 at in other products. We would pay a lot more for the -3N and our board would be dearer if we used that for little or no performance improvement. When customers get hold of our board I am sure results will be on the forum within hours and we will see how close my experience is. There is far more to be gained in areas other than speed grades that I would think about first.

It's true boards could be used for some other purpose probably a processing task. However as this boardis cost optimised this will restrict usage to very similar processing tasks.

As to value that is only what someone else will pay. As yet there isn't any significant trading in second user FPGA boards so it's hard to say what value there might be. Also whilst FPGAs have a long production life there tends to be a replacement family every 2-3 years that is faster and bigger and that usually devalues the previous family to some degree. This is probably not as extreme as a GPU devaluation but it's a fact of life.

You might ask the same question of something like the BFL processors where there is almost no chance of reuse unless some technical details appear aboput the insides of them.

The speed that a FPGA will run at highly depends on the designer of the logic and the tools that synthesis, map and place and route the design. As such the only absolute thing you can say until a design is built and tested are the guaranteed specs from Xilinx. Those won't tell you how fast a design will run only speciified timing guarantees of various elements. Now most Bitcoins already push FPGAs, GPUs etc. beyond specification already so you all know that specification doesn't tell you everything.

For not so light reading look at http://www.xilinx.com/support/documentation/data_sheets/ds162.pdf and you can follow some of the following. if you head starts hurting do feel free to stop reading.I will mention a few things that might be of interest but bear in mind the paragraph above.

Starting with the clock tree specification in -2 is 375MHz max, -3 is 400MHz, -3N is 400MHz. That is the limit specification most logic won't run as fast as that as there routing delay, maybe multiple lut delays, all between registers. So what is actual the difference between the grades. It's not the limit spec but actually a range that various with a bunch of batch related things. Xilinx don't make a -2 or -3 as such they make a XC6SLX150 die which is graded into speed grades. Die performance has a statistical link to where on a wafer (usually large round thing with hundreds or thousands or dies on it) a die is, the processing (slight variations), and even the starting raw wafer quality. Out of all that you get a pile of dies with performance that typically follows a statistical curve that looks like an inverted bathtub. Somewhere along that curve Xilinx has drawn some lines that create bands that are speed grades. Remember the profile shape most die will be close together on the big lumpy part of the curve and somewhere at the top of that lump is the line that seperates -2 from -3. Most of the -3 yield will be close to the top of the -2 yield area. It's then a matter of luck what end of the -2 grade you are in but statistically near the middle is likely.The -3 is very likely to be close to the top of -2. So out of this your -2 is highly likely to mid point between the the -2 limit and -3 limit and -3 chips is likelyto be on the -3 limit.

I will mention the -3N grade. It is a runt grade, and very misleading, that Xilinx created because they had a die yield issue on memory controllers in S6. It should have been called -2N because most of the guaranteed specs are the same as -2. I think the clock tree is virtually the only one the same as -3. So our competitors selling -3N are probably not any faster than the -2 we are currently using.

Ok now that your head is spining lets make that slight simplier. I will take one of the timing parameters Tilo from the datasheet which is the propagation through a LUT your basic building block and very important to Bitcoin logic speed. A -3 has 0.21nS max, -3N has 0.26nS max and -2 has 0.26nS max. Notice the -3N time. So from this an average -2 is likely to be 0.235nS and -3 0.26nS from what I said above. In reality they are probably closer but lets work with these numbers. That says a -2 is 90% of the speed of a -3 and yes the -3 is quicker. However for that 10% more you will pay 25-50% more for the chip typically. Now for Bitcoiners there is something more of relevance to say on this and it's from knowledge gained over many years and not a spec. Generally faster grade chips will burn more static power than the slow chip. In temperature limited Bitcoining that will either reduce the benefit of -3 speed or mean you spend more on electricity and cooling.

Merrick3 sells for $10K. It's a complete different product level than needed for Bitcoin so there lot of work in this design. There 15K via holes in this board and roughly 7.5K connections that are all hand routed for quality of result. Just try imaging how you would arrange chips and make this number of wire connections. It's not easy.

The FT4232 interface is physically a single USB but appears as 4 com ports on a PC. That's a com port per FPGA. There are a few other ways this interface can be used by com port way is simple.

We build FPGA boards in fairly large numbers so we get silicon cheaper than most and hence the price. The Spartan6 for this design has been used in designs here for over 2 years now and we use loads of them in other applications. Have a look at what we do in our Merrick family boards. Anything up 101 FPGAs on a single board.

We are looking at the bitstream aspect right now whilst the layout is completing but we should be able either directly use a bitstream or alternatively rebuild for our package/pinout.

On doing our own core that is certainly a possibility as is just improving existing ones. By past experience the team here are capable of reducing logic size, reducing power or even increasing performance over most "average" user designs. In extreme cases we have seen designs shrick by as much as 30-50% when we have worked seriously on a design. I won't say that can always be done, or it's easy, but we can often do something in this respect.

We are going to support the holes for heatsinks with fans by having suitable holes for these. Our shipping standard will be the 12cm fan with individual heatsinks. The 12cm fan should be a lot less noise and more reliable than the small heatsink mounted ones. For stacks we are planning a push-pull fan arrangement. Between all the holes planned in the PCB there should be a possibility to do a water cooled or single piece heatsink or even heatpipe. There will be 4 fan connectors to take power from and use those any sensible way you like.

We will look at mtgox or even Bitcoins directly but we are newbies in this market so we start with what we are already good at i.e. making boards and using payment systems we that have used for years. We are aware these payment systems might be an issue to some people in Bitcoin community so it will get looked at. It is a learning process for us and it would not be good for the business, or our customers, if we get it wrong. All of our cost base is in traditional money so we do need take care on conversion rates and so on. So please have patience in this respect and let us find our footing. it will get better in this respect. Our first target is to deliver what we have promised and ship customers their boards from late May.

Even with the large number of pre-orders we have received in the last 8 hours we are still well within our May and June capacity. As of this minute any pre-orders place are likely to be fulfilled by about mid-June if not sooner. We will have a better idea on timelines when we hit the milestone of ordering in the bare PCBs some time next week.

This isn't a PCIe board. We may do one more specific PCIe board for Bitcoin and better price than the Merrick6 option currently available. I won't promise this at the moment but anyone has a serious interest, i.e. some numbers, do come and talk to us.

We have stock of the XC6SLX150 at our Xilinx distributor on normal call off that would cover maybe 500 boards. It's a part we already use in reasonable numbers on other things. That is why we can do these timescales. After that silicon is currently on about 6 weeks lead and a manufactured board would have been about 10 weeks.

The design side is easy for our very able team. To give you an idea the design, and first proto was built, of Merrick1 in 8 weeks. That's 100X the complexity of what we are doing on this first Bitcoin product. I could give a very long list of similar examples of other projects. This is a professional design and manufacture team that's at the top of profession teams.

We will look at FPGA optimisations and software aspects but I won't promise any timelines. That's part of the reason for the current price. You will have to do a bit of work yourself so this may not be for everyone in this raw form. The timelines on this will depend on what else the design team end up doing. The team is rarely underloaded so this sort thing can take time.

In addition to this product see my new post on quad XC6SLX150 board.

****PRE-ORDER IS COMPLETED****



Pricing for new orders:





Features

FPGAs

4 x Spartan(TM)-6 XC6SLX150 FPGAs can be run as singles, duals or quad combos depending on Controller version/bitstream in use.

Array Power

12A, 1.2V power supply for each array FPGA. Input is 12V nominal from jack, disk drive connector (Molex), PCIe graphics power connector (6 way) and Pheonix connector.

Array Controller  

Manages array can shutdown power to each FPGA on individual basis.
Controls clock generator and clocks to each FPGA.
Can read temperature sensors (attached to and next to each array FPGA).
Can drive JTAG chain (needs development work).
Controller powers up on USB supply only for debug.
Can be used to pass data between FPGAs.
Supports up and down interfaces for data flow within board stacking.
Fan monitoring.

USB

FT4232 USB controller can be used in multiple ways including 4 COM ports. Can also support JTAG engine. Product is supplied with 1M USB cable.

DIP Switches

For mode setting etc..

Fan Connectors

4 x 3 pin fan connectors supporting 12V fans. Fan monitoring hardware.

Fan + Heatsink

Fitted with Arctic F12 fan and high performance heatsink on each array FPGA.



Prices are based on current exchange rates or staying within 5% of current rates.

We will accept pre-orders on the basis that no money will be charged until your board is ready to ship. We reserve the right not to sell to any person or organisation and/or to ask for more information in respect of and for export control. To pre order send an email to bitcoin AT enterpoint DOT co DOT uk with address and contact details.

Local duties and any applying taxes will be charged direct to you by courier. For UK and EEC we will charge UK VAT unless you are VAT registered in your country (not UK).

We can actually do a USB interface slightly indirectly on this board. There is a debug port on the board that we have FT232 adaptor for and that can plug into a PC as a serial port. We were going to do it directly on the board but we ran out of board area for the chip + connector.

We have a Kintex design and silicon already and this silicon is unlikely to meet Bitcoin expectations pricing wise. We also have Arrria-V silicon and some parts there might be viable for Bitcoin but it needs a proper analysis to be sure. One Cyclone-V (150LE part) is supposed now to be available in small numbers and we might do something with that.

When I say disk connector that's hard disk i.e. Molex. Not always manufactured by Molex they are usually rated at 5 to 7A but can deliver a lot more under stress. We have see wires melt before these connectors fail and that's 10's if not 100's of amps.

On Merrick6 we usually fit either a 7A or 10A fuse on the disk drive connector which is way more than needed for the Bitcoin version. Similarly the 12V from the PCIe has a 2.6A fuse which is a reasonable limit for PCIe operation. If customers want this latter fuse can simply not fitted when we build boards forcing all the current through the DD connector.

I wasn't suggesting Merrick1 for Bitcoin for those reasons. The cut down Merrick6 is a better board initially to talk about. Doing a derivate design for Bitcoin it could be possible to up the power supplies to support all devices by simply say removing DDR chips to make space. The 12A regulator circuit we use could fit into the space taken by 2 DDR chips and we could end up with 6/7 FPGAs each with 12A of core voltage.

That's all viable to do and that sort of design change could be done in a few days maybe a week. A prototype could be turned in about another week if we pushed although that fast gets expensive.

We are based in the UK but we already ship more or less worldwide with a few exceptions. Principally the exceptions are places where we are not allowed to sell due to export restrictions. For most of Europe and N America delivery is usually 24-48 hrs for anything in stock. Other places can take a bit longer.

In terms of deliveries we running this as a professional business and it's rare for us not to meet our delivery promises. Usually if we are out it's because one of our suppliers let us down and beyond our control. We have been in business for 23 years and building FPGA boards for the last 9 years so we have this pretty hooked. We are  building boards every week and this is a normal task for us so I hope we know what we doing by now.

Do have a look at our 101 FPGA board Merrick1. Current FPGAs might be a little small for Bitcoiners but they could be Spartan-6 LX150 if we did the design again. Current Merrick3 is our biggest S6 board launched with a total of 26 S6 on board. It's only got 72 amps for those and would need a haircut, or more power supply, to make a better balance for Bitcoin.

On the power Merrick6 uses a disk drive connector for higher power applications so it won't burn out the PCIe connector.

At the moment the full product is a build to order and the reason for that is cost of silicon sitting on the shelf. That's already being considered to become a stocking board and a cut down would be easier to justify stocking especially if we get steady orders. The 4-6 weeks is a standard time we quote and usually if we have parts, which is usual on Spartan-6 based boards, it can be a lot less. It's more a case of not disrupting our assembly line schedule for say 1 board. If we have a 6 week window we can make a nice slot and maybe build a proper batch size at one time. These are all to do with the economics of building boards.

Merrick6 can run stand alone if needed and there are some communication options here if you are not using the PCIe.

If there is a serious interest we could do designs based on our Raggedstone2 and Lamachan2 boards to offer a PCIe board with 4 or 8 XC6SLX150 FPGAs that would come in cheaper per FPGA. That would need some numbers to make this a worthwhile project.

At the moment the configuration is 2 + 1 although the native configuration is 6 + 1. Various people have looked at our boards aimed general applications and commented that for Bitcoin we have too small a power supply. By cutting the array we get the current per FPGA goes up. Until we get some real performance measurements it migh t be argued that it is more expensive than competing solutions but this a highly engineered board which will give back something in performance. That I don't want to promise until there are real numbers.

It terms of getting dense have a look at http://www.enterpoint.co.uk/merrick/merrick1.html.

Merrick6 is a 10 layer board and that is part of the cost. It is possible for us designs more targetted at Bitcoin and reduce cost but to do that it needs enough demand to justify the cost of doing that design. There is a cost trade between just using a standard-ish board and doing a fully targeted designs.

Merrick6 is indeed a Spartan-6 board. Not really for Bitcoin our Broaddown3 board has 5 Virtex-6 if that takes your interest.

For the industrial motherboards with no fan directly above the board it would be possible to use every slot. That does need very good case fans then. Our solution in development will also allow every slot to be used.

We still have work to do with a Bitcoin load but the 2 FPGAs in the main array should match Icarus performance if not better. If the third FPGA is used as well it can add to that especially if we make the modification to retask the third regulator. Running absolutely flat out in the current Bitcoin configuration it will take about 40W. Doing nothing about 1/10 of that. If we bring in the third regulator add 14W to that.

We will supply heatsinks and a fan board that mounts over the array and heatsinks. That will have a 8cm fan in it and we can trade heatsink size for height above board as necessary. It does do better than heatsink/fans combos and a lot less annoying. More exotic things can be done as well but that all adds to the cost.

There is also another cooling system that we are developing for general use and that could have a significant improvement over or with the standard approach and offers low noise as well. We are looking to add this to all Merrick family boards. I can say much more on that until we do some more work and officially announce it. I don't think that will add a lot to the cost but if it does as well as we hope it will be interesting to see what we can push these boards to in performance terms.