Topic: DIY FPGA Mining rig for any algorithm with fast ROI - page 7. (Read 99474 times)

Bye Bye squirrels...

https://www.hashaltcoin.com/en/batches/1

Way cheaper and easier than this op. Bitmain wins again and won't charge 4%.

A dropdown menu rather than waiting for a bitstream.

Check out - ''The Technicals'' on Youtube.

Yes, but you can already see a looming problem: developers are always going to be tempted to go with the newest/fastest card when rolling out new bitstreams.  I am still waiting for my BCU-1525 and I am reading about the CVP-13, which may ship right after I finally get my BCU-1525.  I don't see developers working very hard on last year's model of FPGA to add new bitstreams.  I'm taking what is posted on the Zetheron site more or less at face value, i.e., you can't easily port your bitstreams between models and different brands.
[/quote]

Technically the most attractive FPGA to develop for is the one in biggest circulation which is BCU1525.  However, if an algorithm 'fits' into a different FPGA's of the same brand (xilinx or intel) it is actually quite easy to port over.  RB1 only took about 3 days to port from BCU1525 to CVP-13.  In some cases bitstreams can be ported from one card to another in a few hours.  Porting 0xBTC to the low cost cards only takes a day or two each.  The bigger problem is porting from Xilinx to Intel or vice-versa. 

Yes, but you can already see a looming problem: developers are always going to be tempted to go with the newest/fastest card when rolling out new bitstreams.  I am still waiting for my BCU-1525 and I am reading about the CVP-13, which may ship right after I finally get my BCU-1525.  I don't see developers working very hard on last year's model of FPGA to add new bitstreams.  I'm taking what is posted on the Zetheron site more or less at face value, i.e., you can't easily port your bitstreams between models and different brands.

I will be porting my code to the Bridgelink FPGA and the Cairnsmore5.  However, be warned that the lower-cost FPGA's have some pros and cons:
PRO -- the low cost FPGA's can sometimes mine core-heavy algorithms at better ROI than the high end cards
CON -- the low cost FPGA's can't mine memory intensive algorithms (no equihash variants or cryptonight variants)

Right now in terms of availability
CVP-13 $6000 available now
Krunchlabs MA-X1 $3650 available Dec 10
TUL BTU9P $3599 available now

Full list:
http://zetheron.com/index.php/supported-hardware/

The current dip in the market is pushing more GPU's offline and GPU farms out of business, resulting in global drop in nethash.  This will eventually be good for FPGA's especially when the market rises.

The latest profit table was update a few days ago:
http://zetheron.com/index.php/fpga-performance-profit/

Keep in mind also that several big coins are seriously considering new algorithms specifically designed for FPGA's.  These would likely run very well even on the low cost cards. 

I'm just now diving into the FPGAs, frankly mining in general after reading up on it for quiet some time this year...it appears the FPGAs are the most versatile and hottest commodity right now. No matter the price, which one do you guys see available to purchase and have in hand the quickest? A possible top 3 to 5 list would be great?

Thanks in advance!

What kind of performance can be expected out of these Cairnsmore5 and Bridgelink units please?
As they are noticeably cheaper than the BCUs one would expect them to be weaker... but with progress in technology etc...?

Whitefire,

Will your algos be available for these Cairnsmore5 and Bridgelink units coming out soon? If so, do you have a procedure in place for us to apply for them once we have our cards? Thanks in advance.

How much each one?

If anybody interested I have 4 Bcu1525 in hand in Hungary, PM.

Soon the $1599 Cairnsmore5 and the $549 Bridgelink FPGA cards will be available, and no matter a person's finances they will have access to home FPGA mining.

In case people haven't already realized it, I am not doing this for the money anyway.  My RB1 bitstream was released with no dev fee and people volunteer to send in their 4%.  In many cases in the last 8 months I turned down huge financial offers from private farms that wanted private bitstreams, but I am trying to help crypto stay in the hands of home miners.  So, even if it were true that releasing stuff publicly wasn't the best financial choice, finances have little to do with it.  Idealism does.  And I think I speak for the other groups as well; you can bet that up until now, Squirrels, Allmine, Mineority and Zetheron have all operated at a loss trying to get FPGA's into the hands of home miners.

On the bright side I scored a nice cheap GPU for gaming.

Can you show me these numbers? How many FPGA at what speed would a developer need to have to compare against 5% fee of 5000 BCUs, 1000+ CVP-13s, few 100 BTU9Ps, few 100 VCUs, and that's not even mentioning some of the other projects that will be coming to market. What would these equivalent FPGA cost a developer? Do you know a lot of engineers? All the ones I've met are ridiculously risk adverse.

In addition, I can't speak for everyone but... I'm not trying to "keep minimum profitability". I'm trying to roll out support for as many algos that we can as fast as we can. There are 3 community developed algorithms on the way. Allmine will be releasing an additional one going into December. I'm hoping things will calm down at SQRL and David will have time to shell his CNv7 and release a CNv8. We're releasing some of our private code to select community developers to give them a jump start on some of the larger and more profitable algorithms. I believe we will be seeing X-N variations coming toward the new year, neoscrypt, new equihash variants, and more

$15-$25/day is still on the table.. Even in down hill market.

the bitstream economics just dont work.  There is no incentive to release a public bitstream, there is only an incentive to mine in secret.  Now with the big group buy they are pushing for a *minimum* of profitability so that sales dont dry up I suppose.  "Cheating" is very profitable.  This assumes secret mining on GPU coins at huge profitability.  Selling the bitstream to others kills the golden egg.

edit:  which is why, FPGA mining will always be marginal.  At least I hope.  Whitefire is wrong on a monumental level.  GPU mining is the most profitable when its a home miner.  You can not beat a home miner at his game if he has decent power costs.  His mining cost is paid in part by actually living in the house.  Like a subsidy.  Greedy people don't like that, and they push the "big farm" argument, which is bollocks.  A GPU is a GPU.  Its not more profitable if you have more.  The lower your fixed costs (0 for a home miner) and variable cost (only electricity for a home miner) the more profitable you are.  Plus most miners live in nordic countries where winter gives mining a negative cost, it actually pays to run at a loss because of heat.

Dont go on the discord it is heavily censored and it is mainly a place for a business to manage customer relations.  I felt I could trust mineority but thats only because of W0lf who was very active here.  Now Wolf has sanitized his account and is nowhere to be seen.  Oh god a company has made great things but they were all sold to private (big) farms I think.  If anyone is partly responsible for big GPU farms, it would be those that sell FPGAs now.

Are you still accepting refunds for the fpga orders?

https://shop.fpga.guide/collections/all/products/btu9p-by-tul
https://www.krunchlabsminers.com/
https://store.mineority.io/sqrl/cvp13/

You still need water cooling radiator/hoses/pump, I recommend Koolance products.

Is there some shop whitch offer whole FPGA solution ?

Yup, these guys have waited long enough so it's time they got some love.

I'm here waiting for the coming day when I can post an announcement about the bitstreams community developers have created and are releasing

Thanks for keeping this thread updated.