Topic: Modular FPGA Miner Hardware Design Development - page 3. (Read 119320 times)

My situation has not changed since my PM. What timeline were you thinking of?

Back in June I decided I was not interested in buying mining hardware this year: getting a stable Internet connection with IPv6 support is more important for me. I don't have any bitcoins to my name, and am not mining, so the current low prices are not really bothering me. However, I want this project to succeed, and I know enough to to know you can't ship the prototypes without the software for the microcontroller.

I also know the software will take time to write. It is tedious and safety-critical: just the type of programming I feel I need to learn how to do (Insert rant about inherently insecure computers here). The code is safety critical, because even if it does not burn your house down when a temperature sensor fails, apparently flashing the wrong image can damage the expensive FPGA.

TL;DR: I think I can commit about 3 hours a week almost immediately. I suppose a good first step would be to manually review all of the pin-outs on the proposed PCB.

Hello everybody.

So just to give word. this thread isn't obsolete afther all. But we are desperatly seeking  software and hardware developers willing to contribute again.

The main goal before getting to a prototype stage is to posses a code for the MSP 430 chip in order to access our fpga's.

So anybody interested and still supporting us  please post so or give me a pm.

regards

Jens 

This thread maybe took some of the air out of the room:
https://bitcointalksearch.org/topic/btcminer-open-source-bitcoin-miner-for-ztex-fpga-boards-215-mhs-on-lx150-40047

Thanks! Snatched one up

Hi,
Just wanted to let you know that http://tideals.com/ offers today the MSP-EXP430F5529 USB Experimenter’s Board for half the price with free international shipping while supplies last. Could maybe be useful.

twinpeaks

Hope this is still alive. It got very quiet in here.

WoW boards!   


http://www.bitcoinminer.com/post/9268040136

http://fillwithcoolblogname.blogspot.com/2011/08/heres-where-its-at-now.html

fizzisist: Yeah, I doubt anything I write now will ultimately be used as-is, but it will likely make good ground-work, and heck it shouldn't take me that long. The Python side of things will probably be the most useful

If the code doesn't end up getting used, add it to my pile of incomplete hobby projects

Regarding heat dissipation, I was going to recommend these until I realized they are still quite a bit smaller than the FPGA itself. Still, newegg has some others that might work well...



http://www.newegg.com/Product/Product.aspx?Item=N82E16835108071

fpgaminer, you're right, we are set up to use the Slave Serial Mode as on page 26. I believe Olaf.Mandel did this work, but he seems to MIA lately.

The thing about the F55xx series, is that they have hardware serial interfaces. You can bit-bang the outputs of the Launchpad, but that might be more work than it's really worth... Since we don't have a F55xx board, we're kinda out of luck until we get this real board made. Also, not having access to the pins you need on your dev board, it gets even worse.

In that respect, I think switching to a JTAG interface is a great idea. Why not? Hopefully Olaf will come back and answer some of these questions better than me before you get too far on the wrong track (in case I'm missing some reason not to do it this way).

Another option is to add a flash IC to the board, and use the MCU to reprogram that when necessary. Would this be easier/better? As ngzhang has pointed out, these chips are cheap.

To clarify, I think you mean page 26, where they show an MCU hooked up to the FPGA for Slave Serial Mode Configuration. Correct?

And then the MCU will also be hooked up to a SPI bus going to both FPGAs (on general IO pins), with the MCU as master. Correct?

What I can do is hook up my Launchpad to 4 GPIO pins on the Spartan-6 LX150T dev kit, and develop all of the SPI communication portion of the code. However, I don't believe I have access to the pins on this board necessary for Slave Serial Mode Configuration. So, I wouldn't be able to code the MSP430 to load the bitstream into the FPGA.

Besides that, I will probably experiment with having the MSP430 talk to the JTAG port of my devkit. That isn't directly applicable to your design but it will provide the ground work for getting the bitstream over USB to the MCU, and the general flow of loading it into an FPGA.

Unless someone else has some way of testing Slave Serial Mode Configuration, you may have to actually build the board before anyone can write and test that side of the code.

Just throwing this out there, but if I can get my MSP430 to load a bitstream into the FPGA through JTAG, would it perhaps make sense to switch the programming interface on your design to that?

fpgaminer, your understanding of the design is correct. The plan is to use the MCU in a configuration similar to that described on page 38 of ug380.pdf to both program the FPGAs and for communication. I'm guessing you have an MSP430 Launchpad? I think that'd be a great test platform. Yeah, we'll have to tweak the code later to work with the 'F55xx series chips, but your work will be a huge step forward. If you'd like to see the current design in detail, PM me and I'll invite you to the shared Dropbox folder. Also note the currently empty "Software" folder in there

The actual functionality of the MCU can grow and change as time goes on. The plan for the hardware was to design it in a flexible way, so that it can be easily incorporated into bigger, more complicated systems (i.e., through the DIMM connector to the motherboard). For the bare minimum, it will need to be able to load bitstreams and handle the data transfer to the FPGAs. Incorporating this with the python interface won't be an easy task, I'm sure, but it seems like you are the right person to do this.

We will be lucky to have your help here (I doubt anyone but you could get it done right, based on your work so far)!

I'd be happy to write firmware for the MSP430, and supporting PC software. Heck, I even have three MSP430 kits  I couldn't resist their $4+free shipping price.

Correct me if I am mistaken: The MSP430 you have selected will be present on each DIMM board. It will be an MSP430 with built-in USB support. The MSP430 will be connected by SPI to pins on both of the FPGAs.

If that is correct I can:

1) Write the SPI module and test it on my Spartan-6 dev kit.
2) Use my MSP430 dev kit to write the MSP430 code and talk to my S6 dev kit.
3) Write a Python interface, console/UI, for the PC software to talk to the MSP430.

Caveat: My MSP430 dev kit has the lowest form of life MSP430 on it. USB is supported through extra chips, so I don't have an actual MSP430 with built-in USB support. So I, or someone else, will have to eventually get one of those chips and tweak up the code to adjust for any differences.

Please correct my understanding of the current design I quoted above, and if the steps I listed seem correct. If all is well, I'll go off and get your firmware and software all ready 

Perhaps you should break down exactly what this software would be doing. I am envisioning some sort of C or assembly giving direct instructions to various hardwares, which seems sort of scary for someone who has mainly done warm and fuzzy C# and some web stuff (CSS and HTML) in the last five years

My problem is that I am so unfamiliar with this sort of project I wouldn't know where to begin.

It certainly will contiune, its just a question of speed.

Such people as you are ones we are currently lacking.

We desperatly need a firmware and software part for the MSP 430 or any other simmilar chip providing most of its funcitons so it you be great if you could focus on the software part.

While I have done a lot of development in the past, it was 100% software development. I have done absolutely nothing hardware related in my life, and wouldn't know where to start. I hope this project continues though!

Thanks a lot for your feedback li-gangyi. I also gave you word at your thread.

Im currently learning to programm the msp 430. Anybody interested is invited to join me as this is the biggest obstacle for our project to continue.

As li_gangyi's attention is reasonably now bound elsewere and we are missing other colleagues like Olaf.Mandel and Bahnfire seemto be not present currently,
this will slow down the development considerably but i aim to complete it anyway.

So maybe just to have a quick roundup of our active personell:
Who is still here to contribute development ?

 

Hi,
Yes I've been working on a separate project, and yes it does work (100Mhash/s at this moment). I can share what I've learnt so far though.
I didn't post for awhile because I've nothing constructive to add (being unfamiliar with the MSP430).

Decoupling: whatever the Xilinx datasheet has spec'd is adequate, I've not had noise issues with the minimum number, never tried going below though.

Power Supply: At 100Mhash the entire board was consuming 6+W of power, real life measurements indicate at least 80% of this is through VccInt. A 5A supply for VccIO and VccInt is more then enough. I stuck with the resistor values previously calculated, and came up with 1.18V on VccINT (measured) and 2.48V on VccIO/AUX (measured).

Heatsinking: In an effort to improve decoupling/capacitor performance, I placed alot of the capacitors on the topside right up to the edge of the FPGAs, the smaller 4.7uF capacitors are fine, they are lower then the chip itself, however the larger 100uF caps can be situated away or on an edge where the heatsink will not 'cross'. In the end with my layout I was forced to use a small heatsink (2.5x2.5cm foot print) with a small fan, couldn't find a heatsink that was super tall to get away with a passive design. The heat output itself at 100Mhash/s is significant, you can actually measure an increase in current consumption as the part heats up (possibly due to leakage increasing), and drop when u cool it.

FPGA itself: I've trouble sourcing for the LX150-N3 part in the Fgg484 package in single quantities, forcing me to use a -3 part and thus raising costs. I'm not sure if it's the same situation where you guys are though.

My help and whatnot: I'm still willing to solder the first few prototypes for free. However right now I don't see a very clear direction and idea of what's to be done next (still no MSP430 guys in the fray), especially with the USB to FPGA interface hardware+software side of things, so I can't say for sure if I'll fund the initial boards.

Ah, interesting that someone took newMeat1 up on his offer. That explains why li_gangyi's been so quiet lately! It is very cool that something is out there and mining, though.

I hope this doesn't mean that we lose li_gangyi's help with design and soldering work!

Also, there is clearly a lot of room for improvement on their design. Their cost is $440 + the cost of the Xilinx platform cable ($200 new, but eBay has some for $50... are they knock offs?). We're looking at about $600 for our prototype which does twice the hash rate. Future designs and bulk orders can bring our price down to probably close to $440.

Looks like li_gangyi has been working on this other board: https://bitcointalksearch.org/topic/m.465501

Which it's actually pretty cool something is out there.