DIY FPGA Mining rig for any algorithm with fast ROI

[hashestohashes]

More info on FPGA's from BBT & OhGodAGirl.

https://www.youtube.com/watch?v=Dddc9r-xyN4

Thanks for posting this, I "probably" would of found it later but I found it here first!

[ShadowBladeX]

I think the scam card was overplayed with the 2x performance AND half the cost. It might've sounded a bit less scammy if they had just gone with "similar performance for half the cost."

Anyway... back to topic;

Did anyone actually conduct any tests? What happened to all those people that claimed they already had the card on hand?

The price is way too high on those FPGA cards for me to invest in right now, but any updates on the cheaper FPGA boards that was also supposed to get support?

A $1k FPGA card is easier to manage then a $4k card.

Hopefully if it does get support, and get 1/4th the performance. It's an easier entry into the field, cost wise.

Your not going to find a $1k FPGA with enough different chips and memory to be feasible for a bunch of algos. THe nice thing about the BCU is it is so versatile pretty much any algo is in the ballpark.

Well I was referring to the Avnet KU040 FPGA board ($975) and the Nexys Video FPGA board ($479) that OP mentioned he'd be releasing bitstreams for.

I wanted some kind of update on the status on the development of those boards to see if I should order one (Avnet KU040) as that would be easier to start a trial with.

[GPUHoarder]

I think the scam card was overplayed with the 2x performance AND half the cost. It might've sounded a bit less scammy if they had just gone with "similar performance for half the cost."

Anyway... back to topic;

Did anyone actually conduct any tests? What happened to all those people that claimed they already had the card on hand?

The price is way too high on those FPGA cards for me to invest in right now, but any updates on the cheaper FPGA boards that was also supposed to get support?

A $1k FPGA card is easier to manage then a $4k card.

Hopefully if it does get support, and get 1/4th the performance. It's an easier entry into the field, cost wise.

Your not going to find a $1k FPGA with enough different chips and memory to be feasible for a bunch of algos. THe nice thing about the BCU is it is so versatile pretty much any algo is in the ballpark.

Well I was referring to the Avnet KU040 FPGA board ($975) and the Nexys Video FPGA board ($479) that OP mentioned he'd be releasing bitstreams for.

I wanted some kind of update on the status on the development of those boards to see if I should order one (Avnet KU040) as that would be easier to start a trial with.

The issue is both of those boards are drastically weaker than the 1525. The $975 board is less than 25% of the logic on a -1 (slowest) speed chip, with no ultraram and built on 20nm instead of 16. That means that while it is 3x the cost the 1525 is more like 6x the performance. Similarly the Nexus Video Board is < 10% of performance on the algorithms it can handle (not many limited bram) and closer to 15% of the cost.

This coupled with only a few bitstreams and no developer focus generally means those boards are going to be a waste vs. the “standard” of the 1525

[ShadowBladeX]

I think the scam card was overplayed with the 2x performance AND half the cost. It might've sounded a bit less scammy if they had just gone with "similar performance for half the cost."

Anyway... back to topic;

Did anyone actually conduct any tests? What happened to all those people that claimed they already had the card on hand?

The price is way too high on those FPGA cards for me to invest in right now, but any updates on the cheaper FPGA boards that was also supposed to get support?

A $1k FPGA card is easier to manage then a $4k card.

Hopefully if it does get support, and get 1/4th the performance. It's an easier entry into the field, cost wise.

Your not going to find a $1k FPGA with enough different chips and memory to be feasible for a bunch of algos. THe nice thing about the BCU is it is so versatile pretty much any algo is in the ballpark.

Well I was referring to the Avnet KU040 FPGA board ($975) and the Nexys Video FPGA board ($479) that OP mentioned he'd be releasing bitstreams for.

I wanted some kind of update on the status on the development of those boards to see if I should order one (Avnet KU040) as that would be easier to start a trial with.

The issue is both of those boards are drastically weaker than the 1525. The $975 board is less than 25% of the logic on a -1 (slowest) speed chip, with no ultraram and built on 20nm instead of 16. That means that while it is 3x the cost the 1525 is more like 6x the performance. Similarly the Nexus Video Board is < 10% of performance on the algorithms it can handle (not many limited bram) and closer to 15% of the cost.

This coupled with only a few bitstreams and no developer focus generally means those boards are going to be a waste vs. the “standard” of the 1525

I understand what you're saying, but at least in my case, I would be more comfortable investing the $1K, under stand things better hands on, and then jump into the $4k+ items.

Right now I just have a small setup of 4 1070ti's and 2 1080ti's running on NiceHash. So I'm willing to dip my toes in a bit in the FPGA ring with the Avnet KU040 board without things getting too advanced.

[matale0]

very nice, how much ?

[freshcoinsfarmer]

What happened to the miner code?

http://zetheron.com/index.php/downloads/

It says:

Then, download a special version of the Zetheron PC miner software, FXMiner-Test.exe here:
<coming in the next 2-3 days>

Last update was 13 days ago.

Any light on this?

[igotek]

there is no mining software so far.....

[Xavien]

Thank you to OP for starting this post and providing the needed info to move this initiative forward.


I live in NY and have a direct personal contact (a close friend, not a random sales rep) that works at Avnet. I would like to organize a group purchase as to hopefully benefit myself and the community and put us in the position to possible negotiate a lower price. My friend says he can work on expediting the delivery of the Xilinx VCU1525 (maybe by a week or 2) just for me. He said the original price of $4000 is off the table and the price is now roughly $5000 till end of June, and in about 2 weeks its going up to $8,000 due to the high/increasing demand.

I would also like some direct answer as to the project status if its ready to go and if I get the FPGA today, or is it still being worked on? I think most people would be happy to pay any dev a fee for a final working solution, right? As I understand it the Zethron website does NOT have a production version of what is needed to run this card only a test version? If someone could please clarify or summarize because there is 66+ pages of conversation and its hard to get a solid answer as to whats going on as of today.

Anyone want to jump in on a pool purchase of the VCU1525 please email me directly xavien.crypto@gmail.com I have a brick and mortar store and my friend is local so I can certainly arrange to mitigate the orders as long as we have enough people onboard to give us the buying power needed to get some discounts before the price skyrockets out of control.

[ZedZedNova]

Thank you to OP for starting this post and providing the needed info to move this initiative forward.


I live in NY and have a direct personal contact (a close friend, not a random sales rep) that works at Avnet. I would like to organize a group purchase as to hopefully benefit myself and the community and put us in the position to possible negotiate a lower price. My friend says he can work on expediting the delivery of the Xilinx VCU1525 (maybe by a week or 2) just for me. He said the original price of $4000 is off the table and the price is now roughly $5000 till end of June, and in about 2 weeks its going up to $8,000 due to the high/increasing demand.

I would also like some direct answer as to the project status if its ready to go and if I get the FPGA today, or is it still being worked on? I think most people would be happy to pay any dev a fee for a final working solution, right? As I understand it the Zethron website does NOT have a production version of what is needed to run this card only a test version? If someone could please clarify or summarize because there is 66+ pages of conversation and its hard to get a solid answer as to whats going on as of today.

Anyone want to jump in on a pool purchase of the VCU1525 please email me directly xavien.crypto@gmail.com I have a brick and mortar store and my friend is local so I can certainly arrange to mitigate the orders as long as we have enough people onboard to give us the buying power needed to get some discounts before the price skyrockets out of control.

There are several discord channels regarding the xCU1525 ("x" means choose a letter, and the modifications are being made to the VCU1525 board specifically for crypto mining). You can buy these from mineority.io for $3600 and they will ship the card to you. There are some shipping restrictions based on your country.

According to their website ( http://store.mineority.io/product/fpga/ ) they have 1224 in stock (yes, I logged in and checked, no I have not bought one, no I am not a shill for minority.io). mineority.io is part of OhGodAGirl's company, and they are working together with GPUHoarder and a number of other FPGA knowledgeable people to make these cards and bitstreams for a variety of algorithms available, especially for mere mortals.

Fair warning, mineority.io only accepts crypto as payment.

mineority.io discord: https://discordapp.com/invite/HnJSDf2

When you join the discord, don't be the typical noob and just blurt out your questions. Read the FAQ, read the pinned messages, scroll backwards in the channels and see if your question has been asked (trust me IT HAS) and answered. Reliable answers in that discord come from people with their names displayed with colors other than white (pink, red, blue).

[Xavien]

thank u for the clarification, I will reach out that way, sounds much easier and cheaper!

[Miguelgreen]

Hello people. I am sorry to ask This here, but its anibody
Sell FPGA ready for mining?
I have zero experience on This subject.
Also i read Somewhere that FPGA and asic already crack
Criptonight v7. Is This true?

[Iamtutut]

Hello people. I am sorry to ask This here, but its anibody
Sell FPGA ready for mining?
I have zero experience on This subject.
Also i read Somewhere that FPGA and asic already crack
Criptonight v7. Is This true?

FPGA yes, ASIC I don't think so yet. Difficulty is lower than just after the fork.
https://www.coinwarz.com/difficulty-charts/monero-difficulty-chart

[DevelopmentBank]

Your not going to find a $1k FPGA with enough different chips and memory to be feasible for a bunch of algos. THe nice thing about the BCU is it is so versatile pretty much any algo is in the ballpark.

True. All those claiming to have profitable FPGAs that cost under $1000 are either over-reporting their hashrates/profits or are lying. Just take a look at what happened with the Dwarfminer FPGA scam. So sad to see that those kinds of scams are able to operate even here in these forums.

Always do your research before buying! If its too good to be true, it's probably a scam.

[R0land]

The VCU1525 is a very nice dev-board.
But building a rig with them

I think for mining it has to much. It´s like buying a car, and only use the engine.
We don´t need PCI-E, because USB is enough (we need it anyway for configiration).
No QSFP28 Interfaces and no DDR4 dimms.
We need HBM-RAM and a lot of GPIO´s.

I would prefer a board without PCI-E, because it scaled bad.

(I hope you understand my german-english   )

[Ivo_07]

Hello everybody. I'm new to this forum like user but i read almost all 67 pages.
What i understand "we all trying to buy a car, cheep, but there is no fuel yet for that car"
Does anybody started minig with FPGA? If yes where is the proof that it works? everybody is talking how to buy but nobody have the mining software. Im located in NYC and willing to buy 2 card as soon as posible if they are ready to mine. Im willing to pay some extra just to see them installed and working.
(please excuse my english, still learning)

[mineroc]

OP, any updates?  Want to try your miner on one of these av cards.

[cheebeez]

About 25,000 fpga cards can work on all algos http://zetheron.com/index.php/fpga-performance-profit/
Several thousand fpga cards already sold by this thread guys + many bought directly.
At this moment in public not have profitable firmware and as soon devs will released it - thousands of cards will start mining!
I think ROI will be 5-10 years.

[Pennywis3]

"Several thousand fpga cards already sold"

"in public not have profitable firmware"

U can't make this stuff up, only in Crypto world 

[bandana121]

About 25,000 fpga cards can work on all algos http://zetheron.com/index.php/fpga-performance-profit/
Several thousand fpga cards already sold by this thread guys + many bought directly.
At this moment in public not have profitable firmware and as soon devs will released it - thousands of cards will start mining!
I think ROI will be 5-10 years.

Aside from the cost of the FPGA's the effects the on the network hashrate is also my biggest concern.

I've read whitefire990 take on this on his website.

"It is important to understand that based on the daily rewards that a coin offers, it can support a limited number of FPGA’s before the FPGA’s overwhelm the coin and the profits diminish, as all the miners split a fixed amount of available daily rewards. There is a simple method to calculate how many FPGA’s a coin can handle before profits start to diminish. Using Monero as an example, we go to a common mining calculator like coincalculators.io, and find that currently Monero (Cryptonight-7) has a network hash rate of 457 MH/s. We cannot increase that hash rate by more than 50% before profits drop too sharply, so we could theoretically increase the hash rate by 457/2 = 228.5 MH/s. With each VU9P FPGA yielding about 22KH/s, then Monero could support 228,500,000 / 22,000 = 10,386 x VU9P FPGA’s before profits drop significantly.  There are other coins on the Cryptonight-7 algorithm which sum total to a supported 12,376 FPGA’s."

http://zetheron.com/index.php/fpga-performance-profit/

So in theory if FPGA's don't dominate the network hashrates like ASICS have done the ROI should be reasonable. Lets just hope that how it works out practically

[fr4nkthetank]

"Several thousand fpga cards already sold"

"in public not have profitable firmware"

U can't make this stuff up, only in Crypto world 

its an fpga.  either you know fpga's or you are eager to learn and spend nights and nights and nights figuring stuff out and learning a lot of new things....but seriously they seem really dedicated to putting out public bitstreams and dev fee ones, so It'll probably be as easy as running a gpu for most people.  Mining some coins in their early days was also source of frustration and hours spent making it work and optimizing.  Now they literally have 1 click miners for n00bs.  so it can happen, and i dont see why its a bad thing.