DIY FPGA Mining rig for any algorithm with fast ROI

[R0land]

Do you think there will be many multi-card (or too-big-for-BCU) bitstreams? It seems like it's gonna be hard for those 2-card combos to beat 2 individual cards hashing solo. Plow funds into BCUs first as long as they're available - with CVP's as a backup?

Multi-card bitstreams are difficult to develop, and to be honest I'm not even sure they are scaleable to large farms.  You need to load a different bitstream into each card, then create a config file on the PC which tells the PC which cards are connected together, and obviously if one card goes down or fails, both cards are down because they cannot calculate a hash result on their own.  It does have potential for really high profits for the small-scale miner, but I doubt a 200 FPGA farm would bother with multi-card bitstreams, especially if you need 4 or 8 cards daisy chained together.  Again, with 8 cards daisy chained, if something goes wrong with 1 card, then the whole rig is down.

We make the NR104 which is quad fpga into an rack mount 100% super modified which beats hands down the sqlabs bcu1525 which we call NR1525M, here is the thread https://bitcointalk.org/index.php?topic=4957557.0

Over-priced, with outlandish claims of value and ROI to boot. Try to answer the real questions in your thread rather than deleting them. Until then, scam alert.

...and video of a box running fans is joke right?

Sorry you feel that way, in due time as we release more details we hope to change your mind

One fpga board hash´s 1.4 G eth ... this requires 11.2 TB (TeraByte) bandwith ... I don´t belief it ! 

[1714893071]

1714893071

[1714893071]

1714893071

[whitefire990]

Do you think there will be many multi-card (or too-big-for-BCU) bitstreams? It seems like it's gonna be hard for those 2-card combos to beat 2 individual cards hashing solo. Plow funds into BCUs first as long as they're available - with CVP's as a backup?

Multi-card bitstreams are difficult to develop, and to be honest I'm not even sure they are scaleable to large farms.  You need to load a different bitstream into each card, then create a config file on the PC which tells the PC which cards are connected together, and obviously if one card goes down or fails, both cards are down because they cannot calculate a hash result on their own.  It does have potential for really high profits for the small-scale miner, but I doubt a 200 FPGA farm would bother with multi-card bitstreams, especially if you need 4 or 8 cards daisy chained together.  Again, with 8 cards daisy chained, if something goes wrong with 1 card, then the whole rig is down.

We make the NR104 which is quad fpga into an rack mount 100% super modified which beats hands down the sqlabs bcu1525 which we call NR1525M, here is the thread https://bitcointalk.org/index.php?topic=4957557.0

Over-priced, with outlandish claims of value and ROI to boot. Try to answer the real questions in your thread rather than deleting them. Until then, scam alert.

...and video of a box running fans is joke right?

Sorry you feel that way, in due time as we release more details we hope to change your mind

One fpga board hash´s 1.4 G eth ... this requires 11.2 TB (TeraByte) bandwith ... I don´t belief it ! 

The VU9 FPGA maxxes out at 100 MH/s on Ethereum, making it useless to mine from an ROI perspective.  These FPGA's are fast at a lot of algorithms, but Ethash isn't one of them.  Honestly I can't believe how many groups are advertising ridiculously fake hash rates for Ethereum.

[Iamtutut]

One fpga board hash´s 1.4 G eth ... this requires 11.2 TB (TeraByte) bandwith ... I don´t belief it ! 

That's 10 times the upcoming AMD Vega 20 (1,2TB/s) bandwith...Don't believe neither.
Vega 20 is said to be released with new gen of HBM2 and not DDR-4 or GDDR-4.

[dragonmike]

Do you think there will be many multi-card (or too-big-for-BCU) bitstreams? It seems like it's gonna be hard for those 2-card combos to beat 2 individual cards hashing solo. Plow funds into BCUs first as long as they're available - with CVP's as a backup?

Multi-card bitstreams are difficult to develop, and to be honest I'm not even sure they are scaleable to large farms.  You need to load a different bitstream into each card, then create a config file on the PC which tells the PC which cards are connected together, and obviously if one card goes down or fails, both cards are down because they cannot calculate a hash result on their own.  It does have potential for really high profits for the small-scale miner, but I doubt a 200 FPGA farm would bother with multi-card bitstreams, especially if you need 4 or 8 cards daisy chained together.  Again, with 8 cards daisy chained, if something goes wrong with 1 card, then the whole rig is down.

We make the NR104 which is quad fpga into an rack mount 100% super modified which beats hands down the sqlabs bcu1525 which we call NR1525M, here is the thread https://bitcointalk.org/index.php?topic=4957557.0

Over-priced, with outlandish claims of value and ROI to boot. Try to answer the real questions in your thread rather than deleting them. Until then, scam alert.

...and video of a box running fans is joke right?

Ha! I'm glad somebody noticed! He keeps deleting my posts in his thread when I'm calling him out, hilarious.

[senseless]

Do you think there will be many multi-card (or too-big-for-BCU) bitstreams? It seems like it's gonna be hard for those 2-card combos to beat 2 individual cards hashing solo. Plow funds into BCUs first as long as they're available - with CVP's as a backup?

Multi-card bitstreams are difficult to develop, and to be honest I'm not even sure they are scaleable to large farms.  You need to load a different bitstream into each card, then create a config file on the PC which tells the PC which cards are connected together, and obviously if one card goes down or fails, both cards are down because they cannot calculate a hash result on their own.  It does have potential for really high profits for the small-scale miner, but I doubt a 200 FPGA farm would bother with multi-card bitstreams, especially if you need 4 or 8 cards daisy chained together.  Again, with 8 cards daisy chained, if something goes wrong with 1 card, then the whole rig is down.

We make the NR104 which is quad fpga into an rack mount 100% super modified which beats hands down the sqlabs bcu1525 which we call NR1525M, here is the thread https://bitcointalk.org/index.php?topic=4957557.0

Over-priced, with outlandish claims of value and ROI to boot. Try to answer the real questions in your thread rather than deleting them. Until then, scam alert.

...and video of a box running fans is joke right?

Ha! I'm glad somebody noticed! He keeps deleting my posts in his thread when I'm calling him out, hilarious.

While they spam my thread and other's threads with their sales.

[whitefire990]

I posted a video on the modded KU040 board:
https://www.youtube.com/watch?v=cim1-ubz4oE

It does work remarkably well, and with over-volting the hash rate could be quite amazing (compared to the low price of the board), but as it stands it requires a few further tweaks in order to be stable.  I need to add a potentiometer to the voltage control to raise the voltage AFTER the bitstream has started hashing and I also need to implement the gradual-power up feature.  Without these, the board is extremely sketchy/unstable and prone to freezing and requires reprogramming and several tries to get it to start up properly.  Not viable for large scale mining (at all), currently.

Running at 600MHz it only starts getting errors around 0.86V (default core voltage 0.95V since it is a 20nm FPGA).  With my heat sink holding the FPGA at a remarkable 40C, there is massive room for hash increases.
I believe it can run at more than 750MHz at higher voltage and maybe even 800MHz.  In the video it is running at 600MHz (x4 cores = 2.4GH/s).

[mineroc]

I posted a video on the modded KU040 board:
https://www.youtube.com/watch?v=cim1-ubz4oE

It does work remarkably well, and with over-volting the hash rate could be quite amazing (compared to the low price of the board), but as it stands it requires a few further tweaks in order to be stable.  I need to add a potentiometer to the voltage control to raise the voltage AFTER the bitstream has started hashing and I also need to implement the gradual-power up feature.  Without these, the board is extremely sketchy/unstable and prone to freezing and requires reprogramming and several tries to get it to start up properly.  Not viable for large scale mining (at all), currently.

Running at 600MHz it only starts getting errors around 0.86V (default core voltage 0.95V since it is a 20nm FPGA).  With my heat sink holding the FPGA at a remarkable 40C, there is massive room for hash increases.
I believe it can run at more than 750MHz at higher voltage and maybe even 800MHz.  In the video it is running at 600MHz (x4 cores = 2.4GH/s).

#teamNexysVideo?

are the ku04 sold out everywhere?  Decently sized for immersion cooling

[TheronB]

Using best case scenario with most profitable bitstream available, what is the current daily profit for this batch of FPGA?

[OurManInHavana]

Using best case scenario with most profitable bitstream available, what is the current daily profit for this batch of FPGA?

Neither of the group-buy FPGAs have shipped yet, and the most profitable bitstreams are under-wraps until the BCU comes out in a month, so it's better to think of conservative estimates if you're considering a purchase from store.mineority.io. I'm using $15+/day for the BCU1525, and $20+/day for the CVP13. If they make more than that (like the numbers on the first post of this thread), great!

[Oakey22]

Anyone received any of the pre orders yet?

[adi1972]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

[Oakey22]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

Not sure about that but one thing makes me hesitant is the fact you are relying on a few to make the miners work with this. At the moment with GPU mining it has a sell case unlike these FPGA which is a very limited market.

[Metroid]

fpga too costly and very low warranty period, if anything happens to it then gameover hehe, you will have to sell a car to cover the loss hehe

[melpheos]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

At the current rate my ROI for my rig is 3 years.
A FPGA even with the current available not profitable bistream is 1 year but indeed, the warranty period is ridiculous.

[nonny12]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

At the current rate my ROI for my rig is 3 years.
A FPGA even with the current available not profitable bistream is 1 year but indeed, the warranty period is ridiculous.

While the warranty periods are pretty short, the longevity of FPGAs run within specs are not short at all. If the FPGA is run within spec then it is rated to run for 10 years of continuous operation. Failure of other components on the board is much more likely than a failing FPGA.

[colomine]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

At the current rate my ROI for my rig is 3 years.
A FPGA even with the current available not profitable bistream is 1 year but indeed, the warranty period is ridiculous.

While the warranty periods are pretty short, the longevity of FPGAs run within specs are not short at all. If the FPGA is run within spec then it is rated to run for 10 years of continuous operation. Failure of other components on the board is much more likely than a failing FPGA.

@nonny12,
 What is the source of your  information about FPGA longevity? And what do you mean by "other components on the board"?  RAM, drives etc? Or components of an FPGA card?

[adi1972]

Miners care a lot about ROI. FPGA seems too over priced, 4-6K per unit.

Mining with FPGA=Snakeoil?

At the current rate my ROI for my rig is 3 years.
A FPGA even with the current available not profitable bistream is 1 year but indeed, the warranty period is ridiculous.

You based on which coin or algo? We need to understand when it comes to crypto investment or mining, it pretty much the old wild wild west.
Full of scam from every corner, be it dodgy ico, mining software or hardware.
IMHO, FPGA is just a subset of gpu mining.

[COINKING7]

Just got an email and I am speechless!

''Our long-awaited FPGA board for cryptocurrency mining is available to order, with expected November delivery for the first batch. Sales are through SQRL.'' From BITTWARE.

This Squirrel company can't even get their 1st batch out from Xilinx! They flat out lied to me/us and said that I/we would be refunded for the 1525. Guess what? NO refund, and now they are putting out a more expensive pre-order together to screw more people over??? Nice Ponzi huh?

Wake up! They took in 15 million from ''batch 1'', and still nobody has a working FPGA. They said I would get my refund this week, and LIED.

Don't even get me started on bitstreams and deadlines. Where are they?

The people behind this should receive jail-time, and very well might. Any lawyers on here?

You can't just take people's money for months and then not deliver.........

 

[stash2coin]

Back in the days some may still remember Butterfly Lab , this guys was taking peoples preorders and no refund policy so people who prerecorded from them was forced to wait many months for delivery, Butterfly Lab doesn't exist anymore but history repeat itself . People have to be careful ordering FPGAs the demand is high and the waiting time in some cases could be months.  

[senseless]

Just got an email and I am speechless!

''Our long-awaited FPGA board for cryptocurrency mining is available to order, with expected November delivery for the first batch. Sales are through SQRL.'' From BITTWARE.

This Squirrel company can't even get their 1st batch out from Xilinx! They flat out lied to me/us and said that I/we would be refunded for the 1525. Guess what? NO refund, and now they are putting out a more expensive pre-order together to screw more people over??? Nice Ponzi huh?

Wake up! They took in 15 million from ''batch 1'', and still nobody has a working FPGA. They said I would get my refund this week, and LIED.

Don't even get me started on bitstreams and deadlines. Where are they?

The people behind this should receive jail-time, and very well might. Any lawyers on here?

You can't just take people's money for months and then not deliver.........

#1) Many people have working FPGA, including me.

#2) This was a group buy with Xilinx supplying the products. The delay in shipping was due to component acquisition problems on Xilinx's end.

#3) Refunds are being processed. It will take time.

#4) The sales for the CVP-13 board was another group buy for bittware hardware. (bittware is doing the production / fabrication there)

#5) The lyra2z bitstream we've developed will be released when boards are shipped. We're actively mining on our (ONE) prototype / development unit.


http://zergpool.com/?address=1JWWCkVKHtDrxQicFkLS2JrwCMbb4nWuFs

^ 1 FPGA at 40Mh/s lyra2z. I'm expecting even more from the fully modified BCU-1525. 60Mh/s or more seems plausible.

Back in the days some may still remember Butterfly Lab , this guys was taking peoples preorders and no refund policy so people who prerecorded from them was forced to wait many months for delivery, Butterfly Lab doesn't exist anymore but history repeat itself . People have to be careful ordering FPGAs the demand is high and the waiting time in some cases could be months.  

BFL was designing their own asic. These FPGA chips are already designed and in mass production. The delay is a problem getting the PCB built, not any problem with the chips as was the case with BFL.