DIY FPGA Mining rig for any algorithm with fast ROI

[ciciteng]

Hi whitefire990,

You are said in the different threads from the beginning that you're also gonna support this type of board (Xilinx Artix-7): https://store.digilentinc.com/nexys-video-artix-7-fpga-trainer-board-for-multimedia-applications/
you gonna release the bitstreams for this (Artix-7) board also instead of VCU board only, right?

Awesome work! We need more people like you here 

************** UPDATE JUNE 2/2018 ********************
The website and initial downloads are now available at:
www.zetheron.com

PLEASE REALIZE THAT TO ACHIEVE THE FULL ADVERTISED HASH RATES REQUIRES EXTENSIVE MODIFICATIONS TO THE STOCK VCU1525 BOARD, as described here:
http://zetheron.com/index.php/hardware-modifications/

If you are interested in exploring FPGA mining, it is strongly recommended that you reach each page on the website in detail.

******************************************************


People asked me to make a new thread on this topic, so here it is.

Here are some pics and video of my 8 x Xilinx VCU1525 rig.  Each VCU1525 card has one Xilinx VU9P Virtex Ultrascale+ FPGA.  Hash rate for the whole rig combined is:

Keccak (Smartcash, Maxcoin): 136GH/s (17GH/s per card x eight) ($160/day at Apr-30 prices)
Tribus (Denarius, Virtus): 16.8GH/s (2.1GH/s per card x eight) ($304/day at Apr-30 prices)
Phi1612 (Luxcoin, Folm): 5.2GH/s (650MH/s per card x eight) ($456/day at Apr-30 prices)
Skunhash (Various coins): 10.4GH/s (1.3GH/s per card x eight) ($261/day at Apr-30 prices)

Those yield around US$20-$57 per card per day ($160-$456 per day for the rig).  Each VCU1525 card costs $4000, or $32K for the whole rig.  At $160-$456 per day, ROI is 70-200 days depending on the algorithm.  I'm not the only one mining with these cards.  Apparently some guy in Germany is getting 64KH/s with Cryptonight-V7 on the same VCU1525's, earning him over $100 per day per card or $800+ per day for a whole rig.  
*********** UPDATE JUNE 2/2018:  The 64KH/s hash rate is mostly likely fake.  GPU_Hoarder has achieved 22KH/s which is the true hash rate for CN7*************

Next up on my implementation list is SHA-224 and Neoscrypt.  I am also developing for the Bittware XUPP3R-VU9P which is an almost identical board as the VCU1525.  I'm planning on releasing the first bitstreams (FPGA config files) to the public May 30 with an embedded 4% development fee.  If you are interested in acquiring hardware, contact jason.harvey@avnet.com for the VCU1525, or Christian Robichaud of Bittware, for the Bittware XUPP3R-VU9P (crobichaud@bittware.com).  Tell them you were referred by Zetheron Technology and you want the cards for crypto-mining and they can expedite the lead time, which is currently around 4 weeks.  The intro price (at Avnet) on the VCU1525 is $3995 USD, but it will be going up to around $5K in July.  The Bittware XUPP3R-VU9P crypto version is $5895 USD, and has two advantages over the VCU1525: (1) it has four QSFP28 100G ports so you can daisy chain 4 FPGA's together to mine Xevan at 162MH/s, and (2) it has flexible memory options, so you can install either DDR4 or QDRII+ SRAM; the QDR memory gives way faster hash rates on Equihash vs. DDR4.   I've been in communication with many members of this forum who are already organizing a group buy.
***************** JUNE 2/2018: The above information is superseded by information posted on the Hardware page at www.zetheron.com *****

Pics:
https://drive.google.com/open?id=1Oh8VV0CDi-R6ls4Up9n7uJk5_gVp5OK2
https://drive.google.com/open?id=11NQoU-R08u9kJkJIMTleqe2wTaTj2jw2

5 second video to hear the noise level:
https://drive.google.com/open?id=1PYJ1Ht7r3s9zmMisEIpsngbW7C9EIENl

You'll notice the blue USB cables from the riser cards are *not* plugged into the motherboard.  The risers are only used to power the FPGA cards.  The black USB cables plug directly into the FPGA cards and directly into motherboard USB slots.  To get enough full bandwidth USB slots I have added an extra 4xUSB3-to-PCIe adapter card.  The whole rig is powered by a single Rosewill Hercules 1600W supply, each card burns 100-160W depending on the algorithm.  Any decent mining motherboard will work, the load on the CPU is very low.  Any GPU rig can be immediately converted to FPGA by swapping out the GPU cards and replacing them with FPGA cards and performing the USB cable adjustment as I described.  Personally I am hoping to sell my 48 GPU's and replace them all with FPGA's but I have been so busy with the FPGA programming I haven't had time yet.

I am also releasing bitstreams for the Avnet KU040 FPGA board and the Nexys Video FPGA board, although due to the small size of those FPGA's they can only mine a couple of algorithms profitably, and if you are serious I suggest a VU9P board.  If you are skeptical then I suggest just waiting until someone else you know has an FPGA rig up and running and you can make a decision then.  Keep in mind these chips are fully reprogrammable and can mine any algorithm.  A lot people are mining with them in secret.  I hope others follow my lead and start to release their software/bitstreams publicly.  I also suggest searching non-English forums and trying to convince the others who are using these cards to publish their software.

Once the community has switched over to high end FPGA's, crypto as a whole will be far more ASIC resistant, because developing an ASIC will then have a poor ROI.  Consider that this 8 x VCU1525 can mine Skein and Myriad-Groestl and make about $7 per card per day ($56/day for the rig); not a great profit, but that is amazing considering you are competing directly against Baikal X10's which are ASIC machines.  The fact that you can still make a profit (with FPGA's) when mining directly against an ASIC shows how this technology can change the crypto landscape.  

Please forgive me if I am slow to respond to PM's, I am quite busy.

[1715153700]

1715153700

[senseless]

Lux just released a statement of their new hard fork.

"Phi2 is also designed to be scalable and modular so that its FPGA resistance can be maintained over time. As the possibility for an industrial, commercial or otherwise mass-produced FPGA for Phi2 becomes clear, the algorithm can be easily modified and updated to Phi3, thus immediately rendering those devices obsolete. This cat-and-mouse game can continue indefinitely until FPGA producers realize they are artlessly wasting their time and their money trying to build FPGAs for Luxcore’s hashing algorithm."

Godspeed.

[Karlitosss]

I want to join the list to get two of these FPGAs. How can I add my name?

[keksov]

I want to join the list to get two of these FPGAs. How can I add my name?

Send a PM to GPUHoarder

[senseless]

I want to join the list to get two of these FPGAs. How can I add my name?

Send a PM to GPUHoarder

No need, see the thread in my sig.

[suchmoon]

This is awesome.  The fewer people that get involved, the better.  I hope the FPGA community stays small, so profits don't get diluted.
...
Still, despite this information, hopefully people don't buy; the fewer people that get involved, the better.

Shooting yourself in devfee legs? Two times?

devfee is a percentage of the pie that is not infinite. It stands to reason that flooding the market isn't in dev's best interest. At least until competitors flood the market.

[powerload79]

This is awesome.  The fewer people that get involved, the better.  I hope the FPGA community stays small, so profits don't get diluted.
...
Still, despite this information, hopefully people don't buy; the fewer people that get involved, the better.

Shooting yourself in devfee legs? Two times?

devfee is a percentage of the pie that is not infinite. It stands to reason that flooding the market isn't in dev's best interest. At least until competitors flood the market.

That pie is a lot less finite than the amount you can scrape together to invest into your own hardware, though. The genius of it is that you effectively own the devfee % of the total hash rate.

[greyday]

Bitmain and Baikal will have their move, i hope all of you realize that FPGA miner developing won't "destroy" these companies.... And it is not aid decentralization either.
Big mining farms will buy a thousand of FPGA's if it is going to be stable and profitable.

This always confuses me. For coin decentralization the hardware type is irrelevant, you can help keep them decentralized by mining solo or on smaller pools.

I think you are wrong, sir.
everybody can buy GPU but not everybody can buy asics (regarding price and scarcity or manipulation of distribution)

True, but they are for sale. But either way what I'm saying is that it's where they are pointed, not their existence, that determines coin centralization. Speaking mostly of the larger coins here, a large enough farm can of course control 51% of a smaller coin.

[greyday]

Bitmain and Baikal will have their move, i hope all of you realize that FPGA miner developing won't "destroy" these companies.... And it is not aid decentralization either.
Big mining farms will buy a thousand of FPGA's if it is going to be stable and profitable.

This always confuses me. For coin decentralization the hardware type is irrelevant, you can help keep them decentralized by mining solo or on smaller pools.

Mining solo ? You'll need huge $$$$$ to obtain the hashrate that can find blocks, unless you mine much smaller coins with a very low network hashrate.

Depends on the coin. You can buy enough cpu power to solo mine Verium relatively effectively for less than a new asic or two...

[nonny12]

Is my interpretation of the recent hardware mod discussions correct? The passive heatsink VCU1525 with a high CFM external fan is the easiest off the shelf cooling solution, right?

[whitefire990]

Is my interpretation of the recent hardware mod discussions correct? The passive heatsink VCU1525 with a high CFM external fan is the easiest off the shelf cooling solution, right?

This is correct.  Better yet, use cardboard and duct tape to channel the fan directly into the heat sink more effectively.

[Etherion]

Is my interpretation of the recent hardware mod discussions correct? The passive heatsink VCU1525 with a high CFM external fan is the easiest off the shelf cooling solution, right?

This is correct.  Better yet, use cardboard and duct tape to channel the fan directly into the heat sink more effectively.

or 3d print an air duct.

[melpheos]

Is my interpretation of the recent hardware mod discussions correct? The passive heatsink VCU1525 with a high CFM external fan is the easiest off the shelf cooling solution, right?

This is correct.  Better yet, use cardboard and duct tape to channel the fan directly into the heat sink more effectively.

Do anyone have a picture of the heatsink without the cover ?

[Aleksandervanb]

iam almost finish with my project on emission cooling with 3m on novec, i may cool your fpga. contact me )) maybe we can build something, what best temperature is for fpga ?

@whitefire990
Could immersive cooling  help  to avoid at least some of the modifications, like soldering a $4K FPGA board?
People have achieved great results with transformer oil cooling, using ASICs and GPUs.

Based on current experiments, immersion cooling an unmodified VCU1525 will get you to about 80-85% of peak performance on power-hungry algorithms without other modifications. 

[m.vina]

no, gpus will always have something to mine, as some algos will almost certainly be created with anti asic/fpga properties. wont stop fpga/asics but will make it much more expensive to develop.

OhGodaCompany has its ProgPOW plans that will leverage so much of a gpus capabilities (not just CUDA cores or whatever) that its just not worth emulating with fpga/asic, as to do it one would basically have to emulate an entire gpu to do it. easier just to use a real gpu.

of course it remains to be seen how that goes, but fpga/asic is not the end of pow on gpus. at least not yet.

This is a bit of good news to read. Hopefully, it is true.

I am pretty worried about my GPU investment and can't help but wonder how i'll be selling all my GPUs incase this turns south. I would have already started selling them by now except that i love taking big risks. Hopefully, it pays off.

[senseless]

The last thing I want to do is see this come out and totally annihilate GPU mining.  And that is exactly what it has the potential to do.  Profitability for everything else will be completely destroyed and we'll be stuck fire selling everything we have in the hope we can get in line to acquire this hardware and software.

no, gpus will always have something to mine, as some algos will almost certainly be created with anti asic/fpga properties. wont stop fpga/asics but will make it much more expensive to develop.

OhGodaCompany has its ProgPOW plans that will leverage so much of a gpus capabilities (not just CUDA cores or whatever) that its just not worth emulating with fpga/asic, as to do it one would basically have to emulate an entire gpu to do it. easier just to use a real gpu.

of course it remains to be seen how that goes, but fpga/asic is not the end of pow on gpus. at least not yet.

ProgPow is a bit of marketing. The hardest part of it is still the memory bandwidth. Nothing really changes.

If you take a look at the paranoia in the monero community now..... Is that really what the coin devs want for their coins? All of their miners and holders to be in a constant state of paranoia about secret asics? Wouldn't even need an asic at that point to destroy a coin. Just a large GPU cluster causing diff spikes to make the community freak out.

I made an asic for monero that could survive their fork!!   I'm mining 100Kh/s   -- That, plus a little bit of falsified evidence and the community would be frothing. Seems like it's almost more profitable to put the fud algo in an asic so it can be more power efficient. Poloniex has options now, right?

[senseless]

Just so I get this straight, we need to install RAM on these boards to mine correct?  Specs of RAM requirements/compatibility?

At present, no.

[GPUHoarder]

No, you did not opened it up. It's @whitefire990 that opened FGPA to others. You just identified a business opportunity after you realized that secret FGPA mining will not be as profitable as before once @whitefire990 releases the bitstream in market. That's what @whitefire990 wanted to do and in my opinion he succeeded.

No, I opened up once I realized that companies were going to try charging $6000 and $7000 for these fpga. Go through my post history. If you think I'm going to make any appreciable amount of money from this, you're wrong. What little we do make will be dumped back into development to provide the community the ability to make, compile and distribute their own bitstreams securely.

Edit: Our money is going to come from the devfee in the bitstreams we'll release. And, no one has to use ours. Whitefire's bitstreams will work on the boards too.

We are all in this for money. You should make money if you are adding value to the community. You are adding value by lowering the FGPA cost.  I am also buying from you. It is none of my concern how much money you are making. In my opinion, credit to open up FGPA mining to newbie miners or to the people who are late in mining should be given to @whitefire990. Also @whitefire990 saved crypto community from Bitmain, Baikal and other ASIC manufactures, who were usurping majority of the value generated by cryptocurrency mining.

Bitmain and Baikal will have their move, i hope all of you realize that FPGA miner developing won't "destroy" these companies.... And it is not aid decentralization either.
Big mining farms will buy a thousand of FPGA's if it is going to be stable and profitable.

From technical perpective FPGA is very very exciting, and hope that i have a chance to play with it. I don't really care who was the " first" guy to open it to public. I'm really glad FPGA guys working together to success their mission. Fingers crossed!

There are not thousands and thousands for them to buy...

[percy_tc]

No, you did not opened it up. It's @whitefire990 that opened FGPA to others. You just identified a business opportunity after you realized that secret FGPA mining will not be as profitable as before once @whitefire990 releases the bitstream in market. That's what @whitefire990 wanted to do and in my opinion he succeeded.

No, I opened up once I realized that companies were going to try charging $6000 and $7000 for these fpga. Go through my post history. If you think I'm going to make any appreciable amount of money from this, you're wrong. What little we do make will be dumped back into development to provide the community the ability to make, compile and distribute their own bitstreams securely.

Edit: Our money is going to come from the devfee in the bitstreams we'll release. And, no one has to use ours. Whitefire's bitstreams will work on the boards too.

We are all in this for money. You should make money if you are adding value to the community. You are adding value by lowering the FGPA cost.  I am also buying from you. It is none of my concern how much money you are making. In my opinion, credit to open up FGPA mining to newbie miners or to the people who are late in mining should be given to @whitefire990. Also @whitefire990 saved crypto community from Bitmain, Baikal and other ASIC manufactures, who were usurping majority of the value generated by cryptocurrency mining.

Bitmain and Baikal will have their move, i hope all of you realize that FPGA miner developing won't "destroy" these companies.... And it is not aid decentralization either.
Big mining farms will buy a thousand of FPGA's if it is going to be stable and profitable.

From technical perpective FPGA is very very exciting, and hope that i have a chance to play with it. I don't really care who was the " first" guy to open it to public. I'm really glad FPGA guys working together to success their mission. Fingers crossed!

There are not thousands and thousands for them to buy...

not yet, but there will be. money talks.

[Ginzink]

There are not thousands and thousands for them to buy...

not yet, but there will be. money talks.
[/quote]

Indeed! Where there is money there is a solution usually. It will take some time, but if orders for FPGA rise at prices the sellers make good money, production increase will follow.