DIY FPGA Mining rig for any algorithm with fast ROI - page 84.

arielbit

legendary

Activity: 3444

Merit: 1061

Quote from: suchmoon on May 04, 2018, 11:21:50 AM

Quote from: Sandal_Hat on May 04, 2018, 11:01:26 AM

Quote from: suchmoon on May 04, 2018, 10:59:28 AM

Quote from: Sandal_Hat on May 04, 2018, 09:50:04 AM

No, it is u that dont get it. U ARE COMPETING with asics because asics will push difficulty to the coins u are mining. Profit margins for mining are thin these days. It is abit higher right now compared to 1-2 months ago due to recent price spike but that will correct soon as more hashrate enters market. U can always have an alt coin to mine but it is gonna make low or negative returns for u because of difficulty increase, which are pushed by asics from other coins.

If GPU miners are pushed from ETH to another coin that is FPGA-mineable then:

1) diff goes up but an FPGA is still more profitable than a GPU;
2) diff goes way up, a GPU is not profitable anymore, still profitable for an FPGA;

There are many other variables in this and many other reasons why this might or might not work, but what you're saying doesn't really make much sense. I think you're misunderstanding how difficulty adjustment works. It will go down if profitability turns negative.

No because u can buy an FPGA, so can everyone else. FPGA is the new GPU. It just costs money which everyone has. Mining is easy, everyone can do it.
U make such an FPGA first. U are first and it has 6 months break even. Really quickly, everyone else will have an FPGA and break even will go to 11-15 months or so, the rough average.
See post 268 for a simpler illustration. I leave that as the last explanation of the conceptual flaw.

How fast can u start buying the items and do this. Lets see this go forward yes.

I don't even know what you're talking about or how it relates to what I posted. "Everyone" is not going to have FPGAs overnight, the whole thing is still in its infancy (scale-wise; I'm sure some smart folks have been mining with FPGAs for a while). GPU mining ramp-up for big farms took months even during the massive ETH boom, and that was with an off-the-shelf product available in huge quantities.

Break-even of 15 months is great for those who don't exhibit the ASIC-induced attention span of a toddler. I still have some GPUs I bought in 2015. Paid off many times over. I don't have ASICs from 2015. Can you guess why?

let me guess....

mine the next big thing and/or mine the next big pump Grin

yrk1957

member

Activity: 531

Merit: 29

Quote from: philipma1957 on May 04, 2018, 11:47:30 PM

Quote from: flankiee on May 04, 2018, 11:05:03 PM

This is a very interesting discussion, everyone stands on their own interests, and op just wants to add a way to mine. Let time prove everything.

I am very interested in the economic structure of mining.

Millions of rigs can mine with gpus. So this reduces that or does the market shift? To adjust to this?

I can’t see crypto coins succeeding without gpus.

5000 dollar FPGA cards don’t cut it if you need 10 x 10 card rigs. 100 x 5000 = 500000 dollars

Where is the driving force to get more of them? If the coin developers dont resist FPGA why should they resist asics? So FPGA will simply be all replaced by asics.

So once again how fast does the op upgrade your FPGA? If a coin switches every month? If it takes 2 weeks your gear costs 2x what you thought it would.

And if the developer say fuck it let asics take over FPGA is not fast enough.

So for this FPGA to work you need just a bit of forks not too many.

This is why I wish the op luck earlier in the thread.
He has to hope both amd and nvidia give up on Mining . Do nothing .

This could happen.

He also needs some resistance from developers ie some forks. But not too many forks.

This could happen.

I can wait and see. My 18x nvidia 1080ti earn more then 40 a day after power . They are paid off.

I could get lucky if forking becomes the norm. As I do see that too many forks make FPGA hard to profit.

Why do you think AMD/Nvidia are even in the mining game?

What does Nvidia get when a retailer price gouges while selling a card? Zero. NVidia would make money by selling chips to board manufacturers or selling cards directly. Do you know if NVidia has marked up its chip prices due to mining?

It’s does not seem interested in selling cards directly as it’s store is almost always out of stock and always at sane prices (e.g. 1080ti at $699).

If it was interested in mining, it would stock up its store and mark up its prices due to high demand. Clearly that has not happened.

And would a public listed company risk investing heavily in crypto market, which faces regulatory challenges worldwide? It would an extremely adventurous move to say the least.

wtfonly16

full member

Activity: 285

Merit: 105

how to find miner for this ?
i have bought 30 fpga card but cant mine anything theres no miner ! how to program them ?

b9ron

jr. member

Activity: 111

Merit: 1

Bitmain has already FPGA board with a xilinx kintex chip on it : sophon sc1, sold at $600.

I think Bitmain mine secretly with them (HPC card with FPGA) ETH since end of 2015.

pikachuy

sr. member

Activity: 544

Merit: 250

Quote from: stash2coin on May 05, 2018, 12:33:58 AM

The moment forking becomes a norm ASIC manufacturers will change the tactics that will make changing the algo pointless. Not many realized it still but TSMC 7nm node is opening a huge room for improvisations for the hardware engineers and designers compared to 16nm nodes. So the hardware limitations will became less relevant for the engineers and if a company is looking after the profit they can achieve it. Crypto market is still blooming so the profits in long term will be astronomical. I'm not pro ASIC but we cant rule out the progress these companies will make, just a reminder.

Or Bitmain will just mass release their own FPGA, kind of like how they pumping out ASIC's. They got the money/research for it. If coins keep forking left to right and put ASIC out of business, you bet for SURE they will release their own FPGA to the masses.

stash2coin

jr. member

Activity: 108

Merit: 1

The moment forking becomes a norm ASIC manufacturers will change the tactics that will make changing the algo pointless. Not many realized it still but TSMC 7nm node is opening a huge room for improvisations for the hardware engineers and designers compared to 16nm nodes. So the hardware limitations will became less relevant for the engineers and if a company is looking after the profit they can achieve it. Crypto market is still blooming so the profits in long term will be astronomical. I'm not pro ASIC but we cant rule out the progress these companies will make, just a reminder.

philipma1957

legendary

Activity: 4256

Merit: 8551

'The right to privacy matters'

Quote from: flankiee on May 04, 2018, 11:05:03 PM

This is a very interesting discussion, everyone stands on their own interests, and op just wants to add a way to mine. Let time prove everything.

I am very interested in the economic structure of mining.

Millions of rigs can mine with gpus. So this reduces that or does the market shift? To adjust to this?

I can’t see crypto coins succeeding without gpus.

5000 dollar FPGA cards don’t cut it if you need 10 x 10 card rigs. 100 x 5000 = 500000 dollars

Where is the driving force to get more of them? If the coin developers dont resist FPGA why should they resist asics? So FPGA will simply be all replaced by asics.

So once again how fast does the op upgrade your FPGA? If a coin switches every month? If it takes 2 weeks your gear costs 2x what you thought it would.

And if the developer say fuck it let asics take over FPGA is not fast enough.

So for this FPGA to work you need just a bit of forks not too many.

This is why I wish the op luck earlier in the thread.
He has to hope both amd and nvidia give up on Mining . Do nothing .

This could happen.

He also needs some resistance from developers ie some forks. But not too many forks.

This could happen.

I can wait and see. My 18x nvidia 1080ti earn more then 40 a day after power . They are paid off.

I could get lucky if forking becomes the norm. As I do see that too many forks make FPGA hard to profit.

2112

legendary

Activity: 2128

Merit: 1073

Quote from: GPUHoarder on May 04, 2018, 06:33:12 PM

That’s a pair of 6xQSFP28 adapter that connects to the FPGA transceivers, with 100Gbps Direct attach cables.

Aka as 6x100Gbps worth of interconnect between two more typical Virtex development boards.

So, is it better or worse than connecting those cards through the transceivers connected to the PCIe and plugged into a PCIe passive backplane? What are the trade-offs? According to the Xilinx docs that card has twice as many xcvs hooked to the PCIe edge than to the both of QSFPs.

flankiee

newbie

Activity: 2

Merit: 0

This is a very interesting discussion, everyone stands on their own interests, and op just wants to add a way to mine. Let time prove everything.

philipma1957

legendary

Activity: 4256

Merit: 8551

'The right to privacy matters'

I ask again what do Nvidia and AmD do to counter this?

@ op if My coin does an algo every month and it is closed software because I am Nvidia what do you do to program your FPGA.

As many would say Nvidia made flashing its bios very hard.

So if Nvidia wants to continue in sales of gear for mining they can make their own coin. With a new algo every month.

How fast for you to reprogram your fpga in that case?

would you be able to not lose 2 weeks on each algo switch?

Or does nvida make its own fpga for 2k and sends you the buyer of that fpga a file for the new algo every month.

I can punch holes in your plans for days and I am not Nvidia or AMD. So basically what do the major card makers do here.

If FPGA and asics kill off gpus' they are millions of 1080ti's that will sell for 200 to 400 on eBay so amd and nvidia won't be able to sell newer cards . I am thinking there will be counterpunching by those two companies.

I will wait it out.

s1gs3gv

legendary

Activity: 1316

Merit: 1014

ex uno plures

Cool project and great thread with lots of interesting perspectives. Props to the OP for his/her/their work. I want one, just cause it would be fun to tinker with.

The future is unpredictable and especially unpredictable in crypto. All these same kinds of ROI arguments and GPU/FPGA/ASIC comparisons have been made many times in the past, plans laid, plans made and plans ruined by a curl of lady luck's lips. How many people regret not mining a coin because the ROI considerations were marginal for you at the time, only to watch it pump 10x or 100x and the coins you might have mined at a small loss suddenly represent a huge lost profit ?

Mine, have fun, keep learning, and keep contributing to crypto's bright future.

HardwareCollector

member

Activity: 144

Merit: 10

Quote from: GPUHoarder on May 04, 2018, 06:26:56 PM

Still working on it, it “runs” 22kH but not completely convinced a good portion of that isn’t spitting errors. I’m using a relaxed definition of correct to squeeze higher hashrate out while allowing some rare corner cases to be incorrect. Currently more hashes than expected are incorrect.

Wow, better than I expected, you are the man.

GPUHoarder

member

Activity: 154

Merit: 37

Quote from: saint78 on May 04, 2018, 06:29:22 PM

And also what is it that we are looking at in that pic?

That’s a pair of 6xQSFP28 adapter that connects to the FPGA transceivers, with 100Gbps Direct attach cables.

Aka as 6x100Gbps worth of interconnect between two more typical Virtex development boards.

saint78

newbie

Activity: 16

Merit: 0

And also what is it that we are looking at in that pic?

GPUHoarder

member

Activity: 154

Merit: 37

Quote from: HardwareCollector on May 04, 2018, 06:21:10 PM

Quote from: GPUHoarder on May 04, 2018, 05:50:48 PM

Keccak @ 11 GH, ~500Mhz 22 full pipelines, but I wasn’t strictly targeting Keccak directly so I didn’t push it to the limit.

For CryptoNightV7 I do the finalizer off FPGA because it’s hardly worth the area for the other SHA-3 candidates.

Thanks for the info, may I dare ask for your CryptoNightV7 hash rate?

Still working on it, it “runs” 22kH but not completely convinced a good portion of that isn’t spitting errors. I’m using a relaxed definition of correct to squeeze higher hashrate out while allowing some rare corner cases to be incorrect. Currently more hashes than expected are incorrect.

HardwareCollector

member

Activity: 144

Merit: 10

Quote from: GPUHoarder on May 04, 2018, 05:50:48 PM

Keccak @ 11 GH, ~500Mhz 22 full pipelines, but I wasn’t strictly targeting Keccak directly so I didn’t push it to the limit.

For CryptoNightV7 I do the finalizer off FPGA because it’s hardly worth the area for the other SHA-3 candidates.

Thanks for the info, may I dare ask for your CryptoNightV7 hash rate?

GPUHoarder

member

Activity: 154

Merit: 37

Quote from: HardwareCollector on May 04, 2018, 05:13:02 PM

@GPUHoarder

Great, have you ran any Secure Hash Algorithms on these devices, and if so, what was your throughput?

Keccak @ 11 GH, ~500Mhz 22 full pipelines, but I wasn’t strictly targeting Keccak directly so I didn’t push it to the limit.

For CryptoNightV7 I do the finalizer off FPGA because it’s hardly worth the area for the other SHA-3 candidates.

philipma1957

legendary

Activity: 4256

Merit: 8551

'The right to privacy matters'

Quote from: jstefanop on May 03, 2018, 03:03:08 PM

...

Nothing is resistant...it all comes down to economics. With FPGAs your limited with the number of programable gates the silicon has. The more complicated the algorithm the less "cores" you can fit on the die. If someone wanted to create a GPU algorithm that was not as economical on an FPGA, you would make it so each round requires a ridiculous amount of instructions, or have the coin use a large amount of different algorithms which take up precious FPGA space. You then take the advantage away from FPGAs because 1) it would require alot more effort and time to program the FPGA, and 2) you can't fit as many cores on it, so the instead of it being 10x as fast as a GPU, you can only get away with 2-3x with maybe a slight increase in efficiency.

So I am amd or nvidia.

I don’t want to lose sales.

Do I develop Nvidia algorithm of the month coin.

A coin that has 36 algorithms that will change once a month .

That would be asic proof and FPGA would be push to run it.

Or do I make my own FPGA cards to mine coins.

I think both amd and nvidia won’t surrender the Mining sales. I wonder what they will do to respond to this.

HardwareCollector

member

Activity: 144

Merit: 10

@GPUHoarder

Great, have you ran any Secure Hash Algorithms on these devices, and if so, what was your throughput?

GPUHoarder

member

Activity: 154

Merit: 37

Quote from: HardwareCollector on May 03, 2018, 09:34:51 AM

Don’t be afraid, GPUs are here to stay; even ASICs can be bricked as we’ve seen with Monero (CryptoNightV7).

Also, there are many misconceptions about FPGAs for mining, and in most cases they turn out to be worst dollar for dollar vs modern day GPUs. FPGA programing require very long development cycles, maybe Xilinx will change that with their optimizing compiler (SDAccel).

Kramble summarizes it here and you can read his last comment:
https://github.com/kramble/FPGA-Blakecoin-Miner/issues/1

With that said, OP may have a breakthrough and we will know in less than a month if no credible evidence precedes it. There’s no need to rush and purchase these cards now, MaxCoin’s and SmartCash’s (Keccak @17 GH/s even if true) profitability with these $4,500 cards dollar for dollar will match GPU profitability for some coins like Bismuth (SHA-224).

If the OP can pull this off, he will be my hero. Grin

Kramble was trying to milk real power out of very tiny FPGAs and use lots of them. This isn’t efficient any more, vs this which is using very high end FPGAs that are designed for this type of crunching. They work very well at it, I’ve been doing this for a while.

Topic: DIY FPGA Mining rig for any algorithm with fast ROI - page 84. (Read 99472 times)