Bitcoin Forum

.

I'm really not convinced that "many" people will be able to do this, it will be cost prohibitive. I think a better solution would do be deny MAC address of the "few" FPGA clusters.

IP rationing was considered early on, and rejected for other reasons.  There is no way to limit the miners based on their choice of tech, and even the FPGA's will be undercut by the first ASIC based PCI mining card to be released.  And if Bitcoin is even moderately successful, we will see such a card being sold publicly within three years.

The network doesn't track mac addresses.  Even if it did, pretty much every one of the concerns with IP based distribution applies equally to mac addresses.

oh, didnt you know? even in windows you can change your mac address with only a handful of clicks :p even aunt tilly could do it. so how would you enforce these clusters?

Who's to decide what level of processing power deserves denial? What if I buy a bunch of GPUs? Should I be denied?

Bitcoin is a full on capitalistic implementation of a cryptocurrency. Because of the need to mine and because of its anonymity, trying to enforce any kind of regulation will be hopeless. The only option we have is to compete, compete, compete.

Even if we could actually deterimine this, we shouldn't limit mining based on tech.  The goal is to make the blockchain as secure from brute force attack as is possible.  If we limit ourselves to certain tech, then there will be tech out there that is fundamentally faster at the task, and an attacker isnt' going to be so inclined to limit themselves to follow our rules.

Beyond that, a fpga card in a computer is a very useful thing.  If bitcoin mining encourages the wide adoption of fpga's into desktops then other programs are going to start making use of those same cards. 

I remember people complaining about how closed source GPU miners are so unfair to those mining with CPUs.
Now everyone mines with a GPU we end up with a much stronger network.

So true, i guess the problem is, 1 i don't have enough cash to buy a decent fpga, 2 i would struggle to set it up, but your right perhaps someone will be kind enough to produce a step by step guide on how to build one.

Yeah.  It's the "these new advancements in technology are going to destroy the weavers!" issue all over again.

Wait long enough, and someone will be selling plug&play PCI expansion cards and you won't have to understand anything about the tech at all.

well i still think we are forgetting that these things are pretty specialist, and sure there are most likely a couple of clusters out there already but they were not built specifically for bitcoin, buidling a uber rig from fpga specifically for bitcoin? i wonder would you ever pay it back?

Why not have an even *strongerer* network then.

Please, read and think about this thread -> http://forum.bitcoin.org/index.php?topic=1332.0



What are you talking about ?

But if there is a limit, then you can never know if it's the strongest.  Logicly speaking, any limit on the choice of tech will effectively result in a network somewhat less secure than otherwise.

well bitcoin is actually the worlds most powerfull supercomputer right? so how strong does it need to be.  :)

It needs to continue to be the worlds most powerful supercomputer.  New tech moves the bar a little higher.  Bitcoin must keep up with that tech.

If you buy a Cyclone 4 dev kit right now ($600USD), I can give you software to have it up and mining with just a few clicks.

Seriously.

The only reason I haven't made that software public is because I don't think anyone is crazy enough to spend $600 for a measly 50 or 80 MH/s :P

It's like average CPU minining people complained that people with cash for expensive GPU cards kicked them from mining market.
And no more coins for people mining on laptops. Now it's your turn :) I don't get why would that be a bad thing.

If that's the only reason then you should make it public anyway, it might always help someone.

But I don't really get what all that FGPA thing is all about, AFAIU you don't get much Hashes/$ but you do get lots of Hashes/Watt amirite ?

Please stop whining, if you cannot compete on mining anymore, you may just buy your bitcoins at the market.

"FPGA cluster miners will kill the Bitcoin!" that reminds me Luddites protesting against industrialisation in XIX century...

Bitcoin does not enforce anything else than tx out collection. For the rest it's auto-adjusting.

If new technology comes around that hashes faster than anything else, the Bitcoin network must embrace it. The new technology can be used to either strengthen the network or it will be used to attack the network. I heartily suggest that we promote the former.

20,999,999.9769 BTC actually.

Strength is to be measured relatively to the current technical possibilities, there's no such thing as "sufficient global hashrate".

Please feel welcome here, help yourself to some lurking, let it rest a bit, realize that wanting to limit the hashrate of certain clients is pointless and its technical feasability is doubtful.


Implying you're...

It's not like that at all.  It's not a condition of diminishing returns, it's a condition of the highest level of security that the collective users are willing to pay for.  If the security is too costly, then the miners will start losing money and some will drop out; thus lowering the difficulty.  The result would be a cheaper network, but a weaker one as well.  New tech just permits new miners to operate more cost effectively.  If the new tech is efficient, but too costly, the mass of GPU miners with paid for "sunk" costs still have an advantage.  GPU mining isn't going to go away for the gamer who does part time mining while not gaming.  FPGA's may take over professional mining, but professional mining companies have to be able to undercut the amatuers with paid for GPU's in addition to pay for their electric and facilities costs and turn a profit.  I consider the idea that professional miners will take over the network to be rediculous, because there is no possibility that a mining company can be more cost effective than the gamer who mines when not at his computer and has to heat his tiny flat with baseboard electric anyway.  No possible way.  If 5% of the world's gamers turned their GPU's to this end while at work or asleep, the difficulty would be much higher than it is today even if every professional mining pool were to stop immediately.  Co-generation (the act of mining for profit while also utilizing the heat to contribute to space heating needs) will likely be a major part of the baseload mining into the foreseeable future.  Professional mining is still very profitable because the growth curve of Bitcoin economy is still so steep, so the exchange price stays well ahead of the difficulty.  When Bitcoin's market matures, this growth rate will slacken significantly, resulting in a much tighter profit margin that will permit only new miners with alternative sources of income (such as a professional miner with service contracts with major retail chains or banks) or individuals that do not need strict profitability (said gamer in an electricly heated flat) will be able to compete with those miners that have already paid for capital costs.

Not unless the rest of the network was willing to change their clients to permit it, and that isn't likely.  Consensus is hard even when it's actually a good idea.

Perhaps the next cryptocurrency to be invented will use a highly recursive and memory-intensive algorithm to give the democratic power back to the CPU.

google/search around the forums for various suggestions for "btc2."

My own includes replacing ECDSA with RSASSA-PSS, for example.  Others suggest something like sha1(hdr)+sha256(hdr)+md5(hdr)...

Be careful about the "memory-intensive" route, you do not want to give miners an incentive to mine empty blocks.

creighto makes imminent sense.

A formula like sha1(hdr)+sha256(hdr)+md5(hdr) will still scale well on a GPU and/or ASIC, it would just run about half as fast as it does now, and would still blow CPU's out of the water by orders of magnitude.  The idea would be to come up with something as proof-of-work that GPU's wouldn't be good at.

GPU's are great for doing parallel repetitive calculations wherever the flowchart for each calculation is the same and where most of the work is done by its numerous ALU's...  But if I understand them correctly, they would be poorly suited for a recursive algorithm containing difficult-to-predict branches.  Someone correct me if I'm wrong, but even doing something like the "pow-mod" function (as used in RSA) might be the sort of algorithm that a GPU would run no faster than a CPU because the only feasible way to do it for large numbers (if I understand correctly) requires lots of recursion - something ALU's can't do by themselves.

So even just sha256(powmod(sha256(hdr))) where powmod used two arbitrarily large pre-chosen constants selected using a method satisfactory to the community.

btw, powmod is m^n mod d, where n and d are carefully chosen very large constants in the hundreds or thousands of bits, and m is the input to the function (and is less than n).

Even as I write this, I am thinking of a way even powmod can be made to scale on a GPU.  There's got to be something that CPU's can do better.

It's hard to predict what processors (CPU/GPU/xPU) will look like in the future but perhaps something can be done that is based on physics, like the speed of light.  If the speed of calculation is limited by latency between distant nodes, there is no speeding it up.  Issues like pipelining would have to be considered, etc.

I don't think you have grasped the fundamental point that if Bitcoin doesn't use the best hardware then someone else will use it against bitcoin.

The network needs to made up of a bigger proportion of honest nodes, regardless of hardware ... it is the hardware belonging to honest nodes that backs the bitcoin value, literally (technically) and economically.

contras:
1.today FPGA's is quite expensive 4 such purpose.
2. no FPGA produced enough ;)
3. its can only reinforce bitcoin
pro:
GPU miners suffer
wrong. you probably over-educated[by schoolbooks] about nature/ideology of Luddites: they protested not "industrialization" itself, but [social]impact/results, ie, turning people into machines, which is now common[employers threat workers as tools and inteact correspondingly]

http://bitcoin.sipa.be/speed-lin-10k.png

Somome brought about 1000 gpu mining going by that chart.

Don't panic about it, I just just came to realisation today that FPGA might seem like a killer now, but we are already half way through 2011, AMD have the 7000 series already lined up they are quite tight lipped about the desktop cards but rumour is its a re-archtecture which just means more power as opposed to the optimised 6 series.

So based on the growth of power of gpus.... just look at the ghash charts we are talking big jumps in architecture generations. almost moores law  ;)

It might just go the other way here, i.e the amount of people able to afford 1 GHASH cards from AMD will far out way the number of people with FPGA's.

I'm completely speculating here, but i have actually gained more faith in the GPU again for price power ratio in the next 18 months will be far superior.

Am i wrong?

Depends if AMD notice the demand for Integer ops from the miners and beef that up some more or go the Nvidia flops direction ... commodity boards are hard to beat with the mass production amortisation of dev. costs.

I wonder if anyone in the GPU industry has actually noticed Bitcoin yet.

Some must have.  Powerful ati cards are sold out in many places.  Maybe a good time for and stock.

Yeah, hard to know ... couple or three thousand cards equivalent ... maybe not ... ~ 2-3 million in sales though.

any AMD shares pricing/trading movements on NYSE  ? :)
at lest Nikkei should respond :P