Bitcoin Forum

Hey all, this is just to let you know that me and a friend are working on making an FPGA miner using 2 Artix-7s. These are 50-65% more power efficient than the Spartan-6s, and faster as well. I'll do my best to keep this thread updated on where we are in the process. We will be using the open source FPGA miner (verilog port). Will try to optimize performance for the two chips
 
Neither of us have much experience with this, so it is kind of a learn as we go type of thing.
 Have no fear, before releasing anything I am having professionals from Xilinx look at both hardware and software to make sure it is optimized.

Special thanks to iidx for helping out with the verilog.

April 14
I have the Xilinx software dev kit and an ML605, not much modification should be required for the Atrix-7 (there were no spare AC701s laying around so I can't be sure)
April 15
- Compiling the open source fpga miner is giving me some sort of translate error (it was made for spartan-6), Working on sorting that out.
- Thanks to fpgaminer I can safely estimate that this should put out about 1GH/s, probably a bit more.
- I (with help from iidx) have successfully compiled on the ML605, didn't have time to check out the actual hash it gets, but you have a good estimate now of what this FPGA will be able to get.
April 17
-Got a few of the hardware specs down
-Estimate a 5v supply will be used (still working on the power estimate)
-Estimated price is currently at ~250-300 USD per unit
October 6
-Project on hold
-So far mining software/hardware works (sort of)
-No idea when I will start this up again (school/work etc in the way)

Reserved

interested!

please post pics + docs when you can!

Will do :)
I can't really provide much info yet as I haven't got the actual PCB layout yet, but I will run a miner on the AC701 (Artix-7 Eval kit) and report back how many MH/s it gets (Should be about 300-400 per chip)

Hey man, very interested in this!  Post more info when you have it!  I hope you guys can make them very price-competitive.  ;D

That's what I'm going for. :)
thanks for the interest

Seems there are no spare Atrix-7 boards laying around anymore, so I will have to wait until I can get one. This should not cause much of a delay at all as I can still work on the hardware part of it. I should have a working PCB design/layout by next week sometime. I do have a full license for Xilinx Software, so I can still make power estimates and things like that. I also have an ML605 and whatever runs on that SHOULD work on the Atrix-7 with a bit of modification.

Working on compiling source now.
Will update with Virtex-6 Hash rate (this will be about 30%-40% higher than artix-7 hash rate) so you will have some sort of estimate on how fast it will be.

I just did a scan over the 7 series datasheets to see their relative performance.  Here is a summary of the important figures; I figure they might be helpful for this and any other 7-series based projects:

Everything is for the -2 speed grade

The FIFO Fmax (block ram) has been a fairly good measure of the absolute maximum frequency we can expect to see out of hashing cores.  My rough estimates show that Artix 7 is likely to have a better MH/s/$ based on these figures alone, and single unit prices.  However, it is difficult to tell for sure, because I suspect that the Artix is crippled in some other way.  I have not check each chip's routing and CLB configurations.

I'm suprised that the kintex and vortex are equal, I would expect the virtex to be quiet a bit faster xD
but thanks, for this, this means the FPGA should put out about 1GH/s

The Kintex and Virtex FPGAs are basically the same.  The Virtex just comes in bigger sizes and allows for higher speed serial transceivers (up to 28 Gb/s).

As a reference, I've been using ML605s at 375 MH/s since 2011.  That's just using 3 copies of the verilog port, with some of the adders replaced by DSPs (Not pipelined).  It's also using some space because I use PCIe for the connectivity to the PC.  So, you could use 375 MH/s as the benchmark for your ML605.

It's possible that there is more speed in there using more pipeline stages, but the size of the device limits what can be done.  I haven't tried to get more out of it so far. 

I've been trying to get 300+ MH/s out of V6LX130Ts, but only have it running at 200 Mhz right now (using PCIe as the connectivity method again).  I've been trying to fully pipeline the DSPs (as FPGAminer has done with the KC707), but the problem is there aren't enough registers/dmem to delay the rest of the pipeline in that device (plus it does not have enough DSP48s to replace all the adders).

Forgot to mention that the ML605s and the boards with the LX130s are both speed grade -1 (DSP48 max frequency 450MHz).

I do have access to boards with K7325T and K7480Ts, but neither has PCIe (they aren't xilinx dev boards) access to a normal PC so I never bothered trying on those boards.

Good luck with the project, it was time for someone to put to work in this theme, you know you can do once you have a good performance in a Virtex7 for example? Use it as a prototype for manufacturing an ASIC with your bitstream and you will have a new generation of ASICs.
But it's still a bit early for that, I do no want to make you feel pressured.

Pretty cool -- I'd love to see a walkthrough/narrative of what you've done (that is to say, once you've arrived where you want to) to get fpgaminer up and running on the ML605.

I'd be keen on picking up one of those boards to muck around with, but would be a little more confident with a primer.

I had iidx explain to me what was required, but I've been doing some reading and as soon as I get it to mine properly I'll post exactly how :)

I'll disregard power efficiency, since that may not matter to some.  However, aren't these models of FPGA's just too expensive to compete with ASIC chips?

The Spartan's typically used were $150 each. Even if bought for $100, that's only maybe ~270MH/$100. I'm not knowledgeable on all the pricing (bulk or otherwise), but wouldn't an Artix be proportionately as costly - if not more?

If I'm not mistaken, an Avalon is about ~280MH/BTC.08 (today's rate that equates to ~280MH/$5.50). So, unless you can purchase an FPGA with a performance/cost that beats that, then I don't know how this can work. Perhaps, this is just an academic pursuit for you.

I would expect them to be substantially _less_ costly per MH.  E.g. fitting potentially 400-900MH/s in an $100 part. Beyond the basic process improvements the Artix/Kintex parts have a substantially improved architecture.  Of course, the S6LX150 designs have had a lot of effort applied to optimizing them.

A batch 3 Avalon cost 75 BTC and delivers 68GH/s. At the current market price that is  $75/GHs.   So maybe a 28nm FPGA ends up being twice the price per MH initially.   But: It would have a smaller basic unit size (probably a 2 or 4 chip board) which would create a larger customer base and it could potentially have better supply.  What does it matter that _in theory_ an Avalon is $75/GHs now when you can't actually buy one??

Plus, once there is a ASIC miner on a better process in widespread use the avalons will either need to head to the landfill— or at least Alaska for use as heating units, while a 28nm FPGA may have better reusablity.

I imagine that if someone had a board like the x6500 based on artix _right now_ they would be selling out. In a month? who knows.

So I don't think this is interesting just as a novelty project... though if it's just an "okay idea that won't make much money" vs an amazing one depends a lot on how the asic miner supply plays out, how the bitcoin market price plays out, etc.

Also, FPGAs have the ability to be re-purposed and resold for a different task, ASICs however, do not have that.
However ||bit is correct, this is mainly an academic pursuit for me, if people end up wanting to buy it great, if not, that's fine aswell and I'll just make and keep some for myself xD

That would be a five times improvement. Sounds much better so far. To be on par with the Avalon, it's going to need another four times improvement.

One of my assumptions is that the supporting PCB & electronics components will be roughly similar. So, the comparison I'm doing is the cost for the chips alone. Avalon is selling chips in bulk at much cheaper prices.

It seems the only benefit of FPGA, for now, is that it can be re-used for other applications, and thus have resale value (whereas ASIC's will be trash if faced with some kind of catastrophic failure of bitcoin). Also, FPGA's delivery time might be better - shouldn't be too difficult to beat any of the ASIC developers at this time.

Well I don't have any sort of specialized production equipment, so IF I end up selling anything, depending on volume shipping might take awhile anyway :p

I can most likely have 1000 pcs of artix on my doorstep tomorrow morning. The same is not true for Avalon— AFAIK they've not given much in the way of dates, but it looks like their pipeline is a couple months long (though hopefully shortening).  Re-usability is a benefit. It's not the only one.

What I think will be great is if chip makers were to make FPGA's super cheap. Perhaps, there are some technologies on the horizon that will enable them to do that. I'd love to see a paradigm shift back to FPGA. Paradigm shifts are entertaining to watch unfold.

well, in theory, I can get xilinx chips at about half price (free if only a few)
So I could make this fairly cheap, probably still not cheap enough to beat ASIC however.
Also, on a related note, what sort of software is out there to design a case?

epic is lucky cause he has the fpga hook up.

fpgas normally aren't in cases. harder to cool.

If memory serves me I believe Xilinx offers an 'Easy Path' program that would hardwire the bitstream and lower per chip cost.  Turn around time is 2 months.

But re-usability is lost.

I realize you're doing it for fun, but better to devote your energies elsewhere.
Just buy a standard case.  Mouser has a pretty good selection.

@Epicblood I'm selling my gpu rig, I feel its obsolete and inefficient. If you are producing a cost-effective fpga unit, I'm interested.

EDIT:how can asics be more cost affective then fpga? I'd rather spend ~$400 on a 1Gh/s fpga miner than 5K asic.

I've taken a look at compiling the Hashvoodoo and similar bitstreams on Kintex/Artix platforms.  I'm relatively less known on this forum so take from it what you will.  Even if you assume a 3-4X speed-up which I haven't been able to obtain with some knowledgable effort, it would be difficult to equal the Avalon effort.

For quotes with the spartan-6 largest part, a quote for ~100 parts would obtain a roughly ~1/2 - 2/3 price discount from digikey/avnet(150-160/unit) prices.  For 1000, you can expect a roughly 1/3 price discount.  So despite a 1/3 discount you are still looking at a price roughly 10X the price of a similar Avalon chip (even from a bulk order with resale.)  Obtaining a ~1-2 million dollar quote price may drop you into the teens, but at that point developing an asic yourself is a better idea.

Easypath devices require a 300k upfront investment, which is relatively unlikely / unprofitable for an FPGA based solution at this point.

Assuming similar bulk pricing for a kintex/artix, you are still looking at a ~50$/part sale for a device providing ~500-750 Mh/s, not exactly competitive.  This also accompanies higher board cost for a higher pincount part, as well as higher power consumption necessitating increased power supply on the PCB.

Sadly it is difficult to say FPGAs will be profitable in the long-term for new buyers.  Short-term however, it all depends on the ability of Avalon / BFL to deliver on said chips.

I completely agree that FPGAs will never rival ASICs in terms of profitability.  Even when FPGA miners were being sold, their up-front costs were higher than GPU rigs.  However, the most important role Bitcoin miners play in this world, above all else, is securing the network.  The entire Bitcoin ecosystem will collapse without a hardy foundation of mining equipment and miners.  Personally, I don't want the future of Bitcoin to be held only in the hands of a few companies that manufacturer mining ASICs, regardless of their intentions.  Having FPGA-based mining software and hardware available is, as I see it, at least a "plan B."  Diversity is a good thing.

I completely agree with you, especially regards to diversifying the network, but this is a similar paradigm shift.  At the current point you would be hard pressed to find someone encouraging CPU mining to secure the network.

Assuming you can take a Kintex-7 from Xilinx.
http://www.digikey.com/product-detail/en/XC7K160T-2FBG484C/122-1842-ND/3911021 - current price 227.50.

Let's assume you can get it for a 1/10 price discount in volume of many 10's of 1000.00s
~23$

Let's say you dedicate 300,000 dollars to develop an easy path device and eliminate 30% of costs.

~15$

Let's then assume you can tweak the bitstream to get 500 Mh/s (highly optimistic)

You are still looking at 30$/Ghash compared to ~5$/Ghash from a similar ASIC(avalon) system for the chip only.  Again, board costs will be higher.

You'll notice the bulk spartan 6 orders never went below ~200$/chip, even from established volume providers like Enterpoint.

I wish FPGAs were a good long-term option, I have many and love them!

Yeah :'(

Hmm? At $70USD/BTC, Avalon's ASICs currently cost $19.85USD per GH/s, and require 3.63 chips per GH/s.  And to be honest, with some effort, I think we can squeeze 1GH/s out of an Artix-7 200.  Though Artix-7 aren't EasyPath chips ...

On the other hand, FPGAs will always use more power than ASICs, so they will lose that war regardless.

Seeing as I want to let people have the ability to recode/re-purpose their FPGAs I won't be using easypath.
But I appreciate all the input.

Absolutely right!  You are looking at closer to 20$/Gh/s , my math error, instead of 5$/chip.

I would love to see 1Gh/sec out of the Artix 200K fpga!  Optimistically, from our experiences with the proprietary fastest bitstreams with isolated multiple hashers, I would say 600-700 Mhash/sec is the likely fastest we will see.  In the months it would likely take to develop a 1Ghash/sec bitstream we will likely continue to see ongoing asic shipments, the availability of Avalon chip only asics, and the likely shipment of a BFL device, and other competing hardware.

Tough war to win.  I still love my Spartan 3E-500 dev kit bitcoin aside ;)  My cairnsmores are still happily churning away by the fireplace :)

the 1 GH/s is an estimate based on the numbers FPGAMiner posted (this FPGA is using 2 chips, not just one)

chrisp,

speaking of cairnsmores, did you succeed in running the tricone bitstream on them?

spiccioli

Updated with some estimates

+1 watching

+1 watching too

very interesting, another person watching here :D
can wait to see some prototypes :D

Double that!

@Epicblood:  Please post any relevant links to where we can get the parts needed to follow you on this project! Apicture or two of you current setup would also help.

At the moment I only have an ML605 that I am using to learn verilog. By monday-ish I should have a general PCB layout. I will post pics of everything I have when I get it :)

Definitely interested in this!

Also, would it be possible to adapt for scrypt mining if there is sales interest? - ie litecoin

I'm not really familiar with litecoin, but I'll read up on what needs to be done and get back to ya

Interested since i wanted to do this myself.

I purchased a xilinx LX9 Microboard (Spartan LX9) to get into programming for these things. Just haven't found the time to get it started:)

The hardest part seems to be the software xD

Also, will post pics of progress on Monday to all those interested

Hey epic how far away do you think you are with completing this? And what would be the turn around time once you place an order for units to arrive?

Honestly I have no idea, like I said in OP, this is mostly just because I want to learn about this stuff.
I should have a rudimentary idea of the PCB layout by monday and a better Idea of what power consumption will be.
Still trying to get a hang of Verilog (will be taking a class over summer)
Currently just working on getting he hang of Verilog using an ML605 and the Xilinx ISE design suite

Any updates on that PCB?

If you have the FPGA powered, configured, clocked and able to talk to the world via one of those (FTDI) USB-Serial converters, I would have thought you're mostly there - the details of the FPGA's insides can come later, as you don't really have any IO planning to worry about...

Enterpoint was never the most cost effective solution. The closest they came was their initial pre-order offering, and once they jacked their prices they lost market share. I love my CM1's, and I'm very glad I chose Enterpoint over other manufacturers, but I only purchased them to hedge risk, knowing I could resell the boards. I knew I wasn't going to get BTC ROI out of them. Thankfully I've made a considerable amount of usd, but it pales in comparison to what I would have made buying coins.

DIY fpga boards have the benefit of being relatively easy to produce (or have limited batches made), can be made in smaller quantities, and can be resold for a decent price, especially if they were bought in relative bulk and sold piecemeal.

I would like to see more open source development of PCB designs and software so a wider userbase can get into the game with low entry costs, and am willing to fund it once my money frees up.

good luck epicblood! been cheering for you since you got the board :D!

My friend who knows the hardware side of things has been sick for awhile, so I have been unable to get a PCB design. I have been meeting with someone from Xilinx who knows Verilog very well and is also interested in BTCBTC. We are working on the miner software and that is coming along nicely.

You've got me watching this one too and on a side note you've got me thinking about the Artix7 FPGA. I'm wondering how well something like this http://www.enclustra.com/en/products/fpga-modules/mars-ax3/ might perform with Scrypt and how well it may scale. For $650 for the starter kit it might be worth me a fiddle.

Again, slightly off topic with the Scrypt but the modules provide an interesting different approach to FPGA PCB's.

I wish them a speedy recovery!

Enclustra have some great stuff, but that board is only using an xc7a100t - which is smaller than the spartan 6-lx150's used on the current FPGA boards. I don't think you'd fit a whole unrolled miner on there, and the PSU on that board is probably not that beefy either.

 

I want to do my FPGA development. Where can I get Xilinx developer IDE from?

Directly from the Xilinx website.

There's a free web edition which includes FPGAs up to, and including, the Artix7-200, and there's also a 30 day free trial of the full edition. But keep in mind, creating a placed & routed & FAST miner within a mere 30 days would put you somewhere in the vicinity of Stephen Hawking with respect to IQ.

I recommend using the free web edition - I think they call it Web pack.

Okay thanks, let me download the IDE and try to compile the open-source hardware code.

If you end up making something like this, make sure to make an pcb that exposes IO etc, so the board can be used as an devkit or similar later, I'm thinking about something like tihs:

http://embeddedmicro.com/products/the-mojo
http://papilio.cc/index.php?n=Papilio.Hardware
http://pipistrello.saanlima.com/index.php?title=Welcome_to_Pipistrello
http://www.zedboard.org/

Then, it can be used for other things when you convert to asic or whatever is most profitable in the future.
Having an badass big fpga for running linux or similar is way cool ;)

If you really want to run Linux on an FPGA, you'd be better off buying a Zedboard - you can actually mine on that too, just not very fast!

Adding low-speed I/O's is probably doable, but to run Linux, you're going to want some decent amount of RAM, and flash, and both of these would drive up the cost of a BTC-mining board - you'd need other voltages for the RAM for example...

So what's up with the project? Are you using a ready made board, like the Icarus project, or you're rebuilding a board from scratch? Also, have you calculated the power of these fpgas? It would be nice to achive 1GHash with 3-4 chips..

same story, we only need to now 3 things

cost
hashrate
delivered date

the rest is useless

Hello,

I have recently been thinking about this route, especially given the impending availability of a 20nm fpga.

So, did you get anywhere with this project? are you still going or given up?

Update ?