Topic: Simple argument for FPGA (Read 2384 times)

'Freshly Minted' BTC (or even other Cryptocurrencies) might actually prove to be a way to get contemporary finance in the game.

If you remove the traceability of their assets, it might decreases their relative risk.

Thus - any mining might be greater than no mining.

Thus - mining with FPGAs, regardless of relative profitability to ASICs, might generate more value than input cost under the right market conditions.

-Johnny
www.americandatafarms.com

Is it possible to pass newly mined BTC through "anonymizer filter," a digital machine that perhaps breaks up the bitcoins into hundreds of branches, sends them all through several wallets and 2-3 days later they all return to your wallet through various channels?

New BTC are only anonymous if you cannot be linked in any way to a particular IP address when signing up for a pool account with an email address (another vector), logging in to check your mining stats, doing the mining or collecting the mining reward.  It can be done but it requires careful planning and no mistakes.  Solo mining when done right is safer but far more costly in terms of equipment required.

Mining always has and always will be premised on the fact that BTC will be worth more in the long term future than it is today.

Dont forget, in 2009-2010 people mined before BTC was worth anything.

Finally, BTC that you mine are 100% untraceable and anonymous. Its very hard to buy BTC anonymously these days in the USA.

Its a grinders game, not for people looking to make a quick buck. For those people, I suggest day trading or playing poker on Seals With Clubs.

One thing I'd like to personally toss at you is also this.  The intangible allure of having some mining capability that is measurable, as opposed to none; yesterday I convinced a backer to front me $3000 to buy real gear.  Couldn't have done that without the block erupter.  Backers want to see the money in their face.  So I showed them what I can do now, compared to what I could do with some money.  Done.

Good luck.

Understood, but this is also in a framework where literally everything else is kept constant - most notably, the difficulty is assumed to be constant or rise - and it occasionally falls in fact.  There is no telling if suddenly half the users on the network will peace out for precisely 6.4 days, stuff like that.  Nobody seems to factor in resale of the hardware either, which is madness, because resale is, for the moment, often done at totally ridiculous prices that folks infatuated with Satoshi's circle would pay.  I will stand by, as I feel I DO understand the mathematics damn near perfectly - it is good for someone who cannot currently produce bitcoins to pay a bit by making a small monetary investment (a break even or loss under typical simplified analysis) and have the ability to produce them, enabling power in the bitcoin culture, even if only a small amount, so long as the margins are close enough...to catch up to a level where you ARE producing measurable fiat currency every day rather than NOT - it has power.

This discussion has split into two topics, I'll now address the other - inability to justify buying from an ASIC and/or mining unit manufacturer.  Yeah, I still cannot justify it.  There is no way buying from them can possibly make sense to me unless we can presume they are quite literally giving you money for you to sit on your rear while they work very hard.  Mysterious broken packages and unending delays are the result of this economic mechanism.  They do not want us to have the good yet, so they aren't delivered...surprise surprise?

This is a mathematical trap that many of the BFL supporters fell into (and some GPU miners as BTC prices dropped.) If the only way you can get a positive ROI is if the value of BTC rises, then you made a bad investment and should have just bought BTC instead. You'll end up with more in the end. (I'll need to mine my Jally for 100 days at the current difficulty to make a positive ROI over just buying BTC back when I ordered it).




Other than that, I think people are forgetting the secondary market for ASICs. They can be sold on auction sites, and that market is a good measure of their current "value". At some point, when backorder logs have been cleared, the market should theoretically stabilize to where the value of the coin will be around the marginal cost of coin production. Why buy a new miner from a company for $3000 when I can get the same one on ebay used for $200. Prices will all change and the evil mining manufacturers will have little control over it.

I have two Jalapenos here hashing away, generating about $25/day income for me.  By the above assertion, why did BFL ship these highly profitable units to me and not keep them until mining wasn't profitable?  BFL has stated they do not mine with the units they produce before shipping them.  Whether people choose to believe this or not is up to them.

BFL and other ASIC producers are not stupid by selling highly profitable units instead of keeping them for themselves.  Have a look at BFL's preorder queue.  It would not have built up to such an extent if people didn't believe they could get a profitable unit.  Whether BFL can ship in a timely manner is another question, but anyone is free to use a difficulty calculator to estimate whether their unit will be profitable by the time it ships.

Here is the ASIC business model:  sell ASIC miners to the general public.  Once everyone has an ASIC they'll naturally want another, and then another, to keep profitability up as everyone else has caught up.  It's like the situation with the transition from CPU to GPU mining.  Once everyone was sitting with a GPU no one was better off than in the CPU mining days.  People went out and bought more GPUs, and more again.  That moment will come with ASICs, where 5Gh/s will be producing 50c a day and the only way to get more money is to... buy another miner.

Possibly, because I do not see how it is economically possible that a company that produces bitcoin mining hardware would conceivably do anything other than sell their units to us at the precise worst time possible for consumers to buy it.  Let me explain.

The mining companies would of course keep any mining hardware that is profitable via mining to mine with it - why on earth would they ever sell any of it?  Well, they HAVE to sell some of it to keep the network healthy because they need bitcoin to survive in order to sell us chips.

BUT the network can be healthy another way too, that doesn't involve consumer profits above absolutely bare minimum or perhaps zero - no single mining hardware producer has 51% of computing power at any given time, and viola, everyone feels ok.  

Simply put - mining hardware producers will produce hardware and if that hardware can profit by mining, they will mine; if it cannot, they will sell it to us.  Because to keep the network alive, they only have to split up into subsidiaries and other mechanisms that make it appear as if they are doing anything else.

This is not a cry to demonize them or create spread propaganda against these companies.  This is Econ 101; these companies care about one thing, the only thing they rightfully should care about, and that's profit.

Then what can we do?  We have to find a way to make the creation of hardware more accessible - IC's are hopelessly proprietary, which is why I was aiming to reduce the task of bitcoin hardware production to a programming task alone.  This way, information can be shared.  It is very apparent to me that these chips are the only thing about bitcoin that is NOT for everyone.

No, it was just a convenient and symbolic benchmark. There was plenty of FPGA development in 2012. After the various designs improved and evolved it just turned out that the Spartan 6 LX150 with this particular bitstream ended up being one of the popular and cost effective combinations. It is already heavily optimized. It is not like you could just buy a different type of FPGA and quickly write something for it that would equal this in terms of cost and performance.

Edit: As far as possible improvements, I believe that it is still possible. But is it worth it to get 20% or maybe even 50% with all the new technology appearing at this moment?

As to the https://www.cardreaderfactory.com/shop/lancelot.html reference, I see that we have equated clock rate with Mh/s.

I am not a programmer really but when I looked at the code for an open source FPGA miner I noticed that the code also aimed for a hash per clock cycle.  However FPGAs are capable of so many interconnects that clock cycles cannot always be directly compared.  I wonder if there is possibility for improvement there.  Was a 1 hash/cycle benchmark always the goal in FPGA designs?

Indeed, and you can see people on ebay doing this.

As for where to find FPGAs, I often look at places like this http://www.higheredsurplus.com/sms/ufl,fl/list/current?orgid=1239

I'm not ignoring your call for a calculation - I will make one, and show it here, regardless if I look like a donkey or not.  Deal?

I'm not trying to be an asshole or anything but you talk a big game of strategy like the other player entrants into this before you didn't think that maybe there was an opportunity there.  If you look at Blackarrow for instance (https://www.cardreaderfactory.com/shop/lancelot.html), this shows you that their boards will cost about 300 dollars and maybe have 80 more hashes above the Block Eruptor you bought. 

The FPGA market has two things working in its favor

1 is the fact that they are predominately available whereas a majority of Asics are not - Vendors realize this and increase the price as the market bears it.

2 is the fact that FPGAs UNLIKE ASICS CAN and ARE re-purposed for other projects/reasons.  They have a life after bitcoin use "in the right hands".

I'm on the same page with Crazy Yates - Anyone can talk big, but numbers solidify things.  Without them, you're not so concrete.  Not trying to seem like a naysayer/basher here; I mean its good to be thinking about how we can be innovating and increase hashpower, but looking at the other thread where people showed pictures of their so called "mining rigs" made up of 20+ hub chained USB sticks and I'm just like WTF were "YOU" thinking?  I see those and think of the people who just saw "available now" and thought this was their only shot to break into their own personal btc harvesting business. 

The guy who said he used his previous mining proceeds to buy those block eruptors was hella smart if he got in at the .95 BTC price point.  He would be even smarter if he decides to sell them 1 by 1 on ebay each week while waiting for his KNC "preorder"...Probably could have placed that on preorder WHILE also buying maybe a 3rd of the Block eruptors and while he wouldn't see the same amt up front, he could see a profitable roi by slowly selling off the Block eruptors while waiting for the "big machine"

Bingo.  This is what I am basically seeking here.  I am new to bitcoin, but not new to engineering.  Your comment is useful - I will be in touch with you about what we have in mind and I'll see if I am full of shit or not.

The other idea is of course buying the ICs and doing a straight up board.  I do not feel confident or even know how I could feasibly produce my own IC's...that is why I liked the FPGA idea from the get go.  But in truth if the ICs themselves are affordable then yes that is an option.

But my thought now is they will not be affordable and worth doing this as long as the ICs are made by the same companies that produce the  complete mining units.

What kind of math would I like to perform? The numbers kind.

I'm assuming you're mining for a profit, or at least you'd like to be. Profits can be had by doing 2 things: increasing income, and decreasing outgoing expenses. The income is based on hashrate and price, neither of which we can really change. Expenses are pretty much 2 main areas: 1) initial cost of hardware, and 2) power costs to run that hardware.

Lets ignore the network for a second. Let's ignore any difficulty projections, or any BTC/USD price projections. We're going to focus on keeping costs down as low as they can go, while still mining at a decent hashrate. Let's just choose a hashrate that we want to be at. Doesn't matter what the price or the difficulty is, we want 1,000 GH/s.

1,000 GH/s from "Hardware A" will earn the same BTC as 1,000 GH/s from "Hardware B' which will earn the same BTC as 1,000GH/s from "Hardware C", am I right? Well what happens when "Hardware B" costs 1/3 as much to buy, but uses more power? What happens when "Hardware C" costs the same amount, but uses 1/3 the power? What is the cheapest option available? That's pretty much the mindset of my THREAD where I look at projected costs of ASIC hardware over a long period of time.

Now in order for you to convince me that FPGAs are viable, I want you to convince me that you can obtain a FPGA with acceptable hashrates, low initial investment, and low power costs. If you can find a way to buy enough FPGAs to mine at 1.0 TH/s for under $50,000, and use less 6kW (The initial cost and power draw of buying 20 SC Singles, which total to 1TH/s), then I'll take it all back, and I'll prolly even buy one from ya!

No worries, this is a debate no offense taken, and I understand I am a n00b, but an arrogant one and not afraid to say I know something about tech and mathematics.

Ok well yes I think we are quite in agreement now.  As for the ROI of a block erupter...for a n00b like me who was getting 10 MH/s before buying one, it is positive I think...having SOME bitcoins to play with as steady income at the moment rather than virtually NOTHING is good as a learning tool and has value in a divide by zero kind of way...my net output went up like infinity %.

And another thing about the ROI of block erupters...well, if BTC spikes in price, that is all different isnt it.  Also, when people in this community use the term ROI, they tend to confine its meaning to something like 3 months, or a time period forseen to correspond to a shift in technological strength. 

However from a long term bitcoin theory standpoint, if bitcoins will one way be worth $10,000, or $100,000 each...the ROI of an erupter is quite clear.  You see?

Anyway I love the fucker, but yes am certainly enjoying these bitcents while they last and at the same time looking for the next step.

I feel for me it is FPGA because I am confident I can obtain them cheaply.  In fact I know I can, from the local university surplus.  I would be happy to disclose fully anything in this regards because I have some loose ideas for FPGA rigs from surplus auctions.

One key feature I am looking to get is ease of expansion.

Yea, the USB miners have a terrible ROI window.  But they're cheap, so people can buy one or two for fun and not care about it.  Anyone serious about turning a profit from mining probably isn't buying the USB sticks. (don't get me wrong, I think they're pretty cool and am considering getting one just for fun, but I'd consider the purchase price a loss rather than an investment)

I don't know much about FPGA design but I know that they're complex chips and that the per-chip cost is very high compared to other ICs.  For example, the x6500 FPGA miner uses two Spartan6 LX150-3FGG484C FPGAs and gets 350-400 MH/s total.  Each one of these FPGA chips costs US $175 on Digikey: http://www.digikey.com/product-search/en?mpart=XC6SLX150-3FGG484C&vendor=122
A Block Eruptor USB (ASIC chip + all supporting circuitry) was about US $100 in btc before they went out of stock.  And that's the super-inflated-due-to-demand price of right now.
I know there are probably more efficient FPGAs out there now but you're going to need to get the price down a couple orders of magnitude and also reduce power consumption to make them competitive with ASICs.

This isn't a function of the ASIC or FPGA hardware but of the support circuitry.  Bitcoin mining is embarrassingly parallel and doesn't require any communication between individual mining chips/devices/whatever.  Each mining device just needs to communicate with a host capable of supplying work and this can be via USB, Ethernet, CAN Bus (linking multiple units to one host), or any other communication protocol.  I personally like the direction that Avalon, ASICMINER blades, and hopefully KNC are going with standalone networked miners.  This removes a central point of failure and provides for more reliable mining.  

You may be a more efficient (i.e., more profitable) miner by doing more work in terms of building your own rigs but you're never going to be able to compete with people using current technology if you're using last-generation technology unless you have unusual circumstances like a pile of free FPGAs and/or free electricity.  "Jimmy rigging your own shit" is still definitely a reality but it's most likely doing something like ordering Avalon or BFL (lol) chips and assembling them onto a board yourself.  I'm not arguing with the premises of this point; doing more work yourself and thinking outside of the box exactly is how you can be competitive since lots of people aren't willing to go that route.  I'm just saying that this mentality shouldn't mean restricting yourself to subpar technology.

No problem, sorry if I'm coming off like an ass. :p

For one, I know of the ROI expected for the USB stick...thing is I went from "jack shit" bitcoin income to "some" so as my first little bitcoin investment there was some existential/sentimental value in that for me, but lets get serious.

As to your saying FPGA is more expensive...depends.  That is my current area of research.  Why do you say this for sure - designing integrated circuits is very expensive.  Laser cutting transistors is not easy and is a trial and error type thing.

FPGA is essentially a programming task.  I also think FPGAs can be made to "scale" better.  That is, adding continually more units to the whole system to heighten power without diminishing returns (incidentally the USB sticks kind of have this property which is another reason I was attracted to them despite shitty ROI on first sight).

Well my thesis is simply this:  hardware companies like BFL or ASICMiner will always take care of themselves (cannot blame them of course).  But that means that still, as it was in the era of GPU dominated mining, the miners who jimmy rig their own shit will benefit from doing so. By avoiding the hardware developer middleman, a miner can control his/her own destiny, take advantage of timing to get deals (e.g.surplus auctions for FPGAs), and circumvent the "timing dilemma" faced by anyone buying new ASIC hardware which amounts to trying to predict the future to decide if buying or preordering is good; further FPGAs can be designed to scale up without end.

Thank you

Right now buying an in-hand ASIC miner will yield you questionable ROI because the demand for the hardware is so high.
The price will drop as more ASIC-based miners are produced and as the difficulty increases.  Eventually the ratio of hardware price to difficulty will stabilize such that mining will once again be marginally profitable for most people.  People will only pay for overpriced hardware for so long before they realize they're wasting their money, i.e., hardware prices will drop so manufacturers can keep selling, and this will cause the difficulty to rise and hardware prices to drop again until an equilibrium is reached.

FPGAs will most likely not favorably compare to ASICs with respect to mining because FPGAs cost significantly more to develop and produce and because their hashrate/power performance can't compare to an ASIC.  ASIC miners are only so expensive right now because of the demand. (I'm going to make a safe bet that ASICMINER, Avalon, and BFL have all already paid off their their initial investment costs)

I didn't buy anything, except 1 block erupter just to get some BTC so I can gamble it away senselessly.

As for the math, what kind of math would you like to perform to compute bitcoin hardware projected outcome?  This currently cannot be done without some knowledge of the difficulty increase and what the price of BTC will be within reason during the time period over which you calculate...which cannot be done.

But don't we BOTH know that block erupters are probably NOT worth the investment bitcoin to dollar within > 150 days; butterfly labs a fine example of the preorders taken at a time at which a machine would have been tremendously profitable...now jalapenos shipped to the precious few who received them will profit briefly while the majority will arrive at or after a 5 GH/s break even point...

The math is economics.  Why would ANYONE or ANY company really want to sell a machine that is better off sitting and earning bitcoins?  It only makes sense economically to ditch a piece of hardware when one cannot foresee a better outcome keeping the device and setting it up to mine 24/7... 

Show me the math or STFU.

It is clear to me that manufactured ASIC's will always sell for more than they are worth (in returns). 

It is also clear (to me) that specialized IC's (I did not use the term ASIC) are "what is up"

==> if professionally developed and marketed gear always costs too much, and integrated circuits are dominating...

enter FPGAs to the forefront of real profit for miners (the other profiteer being the manufacturers)