[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance

View post   
File: 773 KB, 972x506, Screenshot 2018-12-16 at 22.44.21.png [View same] [iqdb] [saucenao] [google]
12147779 No.12147779 [Reply] [Original]

how does one go about creating an FPGA miner? I can program but how do i actually create an FPGA miner? It seems like for the smaller coins with custom algos, you could just create a miner and essentially print free money.

Whats the catch here? Is it really so hard that no one does it? Also is there an FPGA miner for monero yet? Please anons. I need help. this crash has left me with only $5k left...

somebody tell me what i need to learn in order to make FPGA miner's.

>> No.12147813

>>12147779
FPGAs parallelize algorithms allowing them to be completed faster. Its not like computing your algorithms on weak ass 600 Mhz FPGAs would make then run faster than on a PC.

>Also is there an FPGA miner for monero yet?
Get back to me when you can parallelize that shit and I'll write up the VHDL for you. We'll split the profits 50-50

>> No.12147830

>>12147779
>thinking a pleb like you can just 'create' some hardware
>only $5k left
just buy the coins dumbass

>> No.12147847

>>12147813
yes i understand sort of how they work. The programming ive done would be considered by most as "low level" - using rust and c++ for things like game engines etc. but not low level enough as FPGA's.

So correct if i am being a retard but basically you have the circuit on the board, and you can control all of the gates and how data is passed through them. is this right? then through one operation you can do multiple bitwise operations?

Why is it so hard to do with monero?

>> No.12147861

>>12147830
FPGA is just a programmable chip. you can program it how you want. no need to create the hardware.

>> No.12147955

>>12147861
yeah my bad, thought you wanted to create hardware. Just buy the fpga's then and toy with them. You need to really understand the algo then just program the logic. The profitability will depend on how efficient you make this.
If there are no FPGA miners for certain algos, there's probably a good reason

>> No.12147975

>>12147779
what's happening here?? is it going up the pooper? need to know for business and finance

>> No.12147987

>>12147779
im gonna need sauce

>> No.12148000

>>12147847

Aren't fpga's written in verilog? Either way the first step would be find an electrical engineer desu senpai

>> No.12148041

>>12147847
>So correct if i am being a retard but basically you have the circuit on the board, and you can control all of the gates and how data is passed through them. is this right? then through one operation you can do multiple bitwise operations?
nah usually you program in a c like language, you write the parallel modules that basically all do sha256 on the same upper bitmask, and you feed the counter to the fpga and it calculates multiple hashes at once. for this you also need some sort of communication gateway code setting the block hash and nounce mask using the fpga legs and a circuit board that parallelize the fpga-s, the fpga will signal if any off the parallel computations found the 0 hash then the circuit board pic sends a message to the control unit which is a raspberry or a pc. something like that.

>> No.12148053

>>12147779
Most FPGA mining bitstreams utilize Xilinix chips. Buy a Xilinix FPGA card, find an existing bitstream, and start reverse engineering.

>> No.12148073

>>12148000
well i am on a tight budget so thats not possible. i am more wondering if this is something that can be learned in a few months or if its something that needs years of experience/PhD level of knowledge to do.

for reference i have no formal training in electrical engineering but know a lot about cryptography and have a v. good understand of how mining works. this anon seems to know what hes doing >>12147813

>> No.12148107

>>12148073

Nah you could pick this up in a few months if you have a strong cs background. My buddy does this type of stuff and we've talked about it at length.


>>12148041

That's pretty cool

>> No.12148119

>>12148041
ok so the bottleneck here looks like it would be sending the all the block data into the fpga and then the fpga having to fetch the entire blockdata from the RAM each time it increments the nonce. actually hashing the data should be very fast right since that can be parallelised?

>> No.12148324

>>12148119
>ok so the bottleneck here looks like it would be sending the all the block data into the fpga
no because you only need the block hash and the nounce. it's just less than 512 bytes i think with various message frames and control codes still less than 1kb.

>> No.12148786

>>12147987
>im gonna need sauce
Me too. I know it's Riley Reid but I want the title. I like those earrings