What exactly is the point of everyone's hatred of ASIC's?

[Thylacine]

A point was made earlier in the thread that I wish to dispute. Somewhat.

I don't recall which of you said it, but what was said was that ASIC chips are relatively cheap to produce and sold at an absolute premium. So far, so good. But then the point was made that anyone developing ASICs would be insane to not follow that paradigm.

This is what I would dispute. The opportunity exists right now for someone who has the resources to follow a rather different model, and it's not only not insane, it would generate them more business than they could handle AND more goodwill than any other manufacturer.

That model is this:

Make a small batch that's expensive enough to cover NRE and a rather large chip run. Use the profits from that batch to build a huge amount of machines, and sell them just over cost. The profits per machine would be minimal, but selling IN HAND at low cost would gather up so freakin' many sales that said manufacturer would probably have to expand their operation to keep up with demand.

Now if you believe that Bitcoin is a passing fancy, this model will not appeal. If you're only in it for the short run, it won't appeal. If you want to be top dog when the dust settles, it will have a great deal of appeal.

This is the way I would do it. I am crazy, but I'm not insane. I understand the consequences and morals of my actions. (reference M'Naughton, the legal standard for insanity at common law). The only thing stopping me is lack of funds, as with money I could hire the experts.

Personally, I love ASICs. They're making the network so strong that mainstream adoption of bitcoin is within reach, as even the true paranoid can do the calculations and figure out that the network can't be easily broken. Would it have gotten there without the ASIC revolution? Probably, but it wouldn't be there now.

This same "arms race" happened with PC's. In the long run it was a good thing, as you can now go to Walmart, plop down 400 dollars, and have what was defined as a supercomputer when I was a pup.

This.

Given all the ill will towards pre-orders and so on, I can't understand how someone hasn't quietly manufactured a mid-range mining rig, no pre-orders, no website, nothing. Boxed up every single rig, ready to ship, then opened their webstore and customer service centre and watched the money and kudos roll in. Hell, let's go all out and say they might even have owner's names, a physical address, and a phone number on their website.

[Biomech]

A point was made earlier in the thread that I wish to dispute. Somewhat.

I don't recall which of you said it, but what was said was that ASIC chips are relatively cheap to produce and sold at an absolute premium. So far, so good. But then the point was made that anyone developing ASICs would be insane to not follow that paradigm.

This is what I would dispute. The opportunity exists right now for someone who has the resources to follow a rather different model, and it's not only not insane, it would generate them more business than they could handle AND more goodwill than any other manufacturer.

That model is this:

Make a small batch that's expensive enough to cover NRE and a rather large chip run. Use the profits from that batch to build a huge amount of machines, and sell them just over cost. The profits per machine would be minimal, but selling IN HAND at low cost would gather up so freakin' many sales that said manufacturer would probably have to expand their operation to keep up with demand.

Now if you believe that Bitcoin is a passing fancy, this model will not appeal. If you're only in it for the short run, it won't appeal. If you want to be top dog when the dust settles, it will have a great deal of appeal.

This is the way I would do it. I am crazy, but I'm not insane. I understand the consequences and morals of my actions. (reference M'Naughton, the legal standard for insanity at common law). The only thing stopping me is lack of funds, as with money I could hire the experts.

Personally, I love ASICs. They're making the network so strong that mainstream adoption of bitcoin is within reach, as even the true paranoid can do the calculations and figure out that the network can't be easily broken. Would it have gotten there without the ASIC revolution? Probably, but it wouldn't be there now.

This same "arms race" happened with PC's. In the long run it was a good thing, as you can now go to Walmart, plop down 400 dollars, and have what was defined as a supercomputer when I was a pup.

This.

Given all the ill will towards pre-orders and so on, I can't understand how someone hasn't quietly manufactured a mid-range mining rig, no pre-orders, no website, nothing. Boxed up every single rig, ready to ship, then opened their webstore and customer service centre and watched the money and kudos roll in. Hell, let's go all out and say they might even have owner's names, a physical address, and a phone number on their website.

Bitmain has come the closest, and if LightningAsic pans out, they seem to be playing the same way. But they do it in small batches, rather than continuously in hand. Still, far better than ANY of their competition.

[nlsupernova]

I don't get those people who complain that ASICs won't ROI, and then they go ahead and build giant GPU farms that won't profit unless they can count on reselling their used hardware on eBay for >75% cost after 1 year.

if you buy overpriced 290x cards for 900 dollars a pop, you wont roi, true.
i build a new rig with 12 7950`s over christmas, i think 5 of them i bought where used, the rest brand new, also the cpus and ram where used as well.. in terms of btc it has allready has made roi.
i got the btc back that i put in allready. its just gravy from now on. and because difficulty doesn`t move insanely fast, it wil bring goed money for at least 6 months. maybe a year, and than ik can probably sell everything off for maybe 25% of what i payd for it. the rig wil pay for itself at least 3 times. probably more like 5 times.
you just have to be smart about it.

good luck doing that with a 10k neptune.

[nlsupernova]

Wrong. The answer is that an ASIC protected network prevents malicious entities from building ASIC devices and 51% attacking a network protected solely by CPU/GPU.

Are you on the grass this evening sir?  Since ASICs are already developed, the "malicious entity" with money to burn that you talk about, would simply go and buy a bunch of ASIC to accomplish the 51% attack.  They could also do the same thing with GPUs, thus, it is neither a positive or negative to either argument.

I would argue though, that a worldwide network secured by both GPU and CPU would be the hardest to attack.  Take a coin like Darkcoin for instance.  I believe GPU and CPU will live together on this coin somewhat, and CPU won't be entirely crowded out.  If Darkcoin fails in this regard, you can always design a coin strictly for this purpose, and leverage memory requirements + hashing functions high enough that GPU always has the advantage, but CPU miners will still be around in high enough numbers.

The obvious solution to all these problems is to just have more than one block chain, with vastly different hashing functions and memory requirements for all of them.  Obviously toss some Proof of Stake systems into the mix too.  Wait...that's what we already have now...it only upsets people that are highly invested in Bitcoin...so they call anything that isn't Bitcoin a trash coin.

We don't need 200 coins to exist, but we do need at least 10 with highly different attributes.  I figure many proof of stake coins will also survive as virtual bank accounts.  I think PPC will definitely be dethroned as top PoS coin.  Cashcoin in particular has a better economic model.

we do need the fastest hardware available to protect the network, if btc network still ran on cpus. and one gouverment started to develop an asic on their own in secret, 5 antminers probably where enough to kill the network. i dont think the cpu hashrate would be much higher then a terrahash.

if you want to that now you need much more then a few smart people and a few bucks to design an asic that is capable of taking over the network.

[safeminer]

because they missed out on getting rich with zero or little investment.
Suckt it up, invest take risk and profit, dont be a cry baby.
The sun already gets up every day for free !

[nlsupernova]

A point was made earlier in the thread that I wish to dispute. Somewhat.

I don't recall which of you said it, but what was said was that ASIC chips are relatively cheap to produce and sold at an absolute premium. So far, so good. But then the point was made that anyone developing ASICs would be insane to not follow that paradigm.

This is what I would dispute. The opportunity exists right now for someone who has the resources to follow a rather different model, and it's not only not insane, it would generate them more business than they could handle AND more goodwill than any other manufacturer.

That model is this:

Make a small batch that's expensive enough to cover NRE and a rather large chip run. Use the profits from that batch to build a huge amount of machines, and sell them just over cost. The profits per machine would be minimal, but selling IN HAND at low cost would gather up so freakin' many sales that said manufacturer would probably have to expand their operation to keep up with demand.

Now if you believe that Bitcoin is a passing fancy, this model will not appeal. If you're only in it for the short run, it won't appeal. If you want to be top dog when the dust settles, it will have a great deal of appeal.

This is the way I would do it. I am crazy, but I'm not insane. I understand the consequences and morals of my actions. (reference M'Naughton, the legal standard for insanity at common law). The only thing stopping me is lack of funds, as with money I could hire the experts.

Personally, I love ASICs. They're making the network so strong that mainstream adoption of bitcoin is within reach, as even the true paranoid can do the calculations and figure out that the network can't be easily broken. Would it have gotten there without the ASIC revolution? Probably, but it wouldn't be there now.

This same "arms race" happened with PC's. In the long run it was a good thing, as you can now go to Walmart, plop down 400 dollars, and have what was defined as a supercomputer when I was a pup.

i was the one who said that. the problem with your statement the way i see it is quit simple. asicsthat are released at the moment are becomming obsolete far to quickly.
chips that are made now are surpased with chips with double the power within months. (i`ll say it again, thats because current chips are based on manufacturing processes that are considerd old by the rest of the industry)
the people who are rushing these things to market are not out to build a brand or something, they wanna get rich quick.
plus they dont have enough r&D to sustain a long term model. they don`t have the money to design the new generation of chip technology, thats done by the big dogs. the chips architecture was developed in the fpga days, and for the last year they just have been shrinking the dye. that isn`t really developing anything new. because the r&d that was needed to figure how to do that was done years ago by people who didn`t have a clue what bitcoin was and was done probably for the aerospace industry.
once we hit 20 or 18nm,  the huge performance gains are a thing of the past.

i`m sure there are some devs out there who are allready developing real top of the line hardware, but that costs time and money. but when they are done they dont have to wory there will be anyone making vastly more efficient hardware for quite a while.then it becomes much more interesting to sell at normal prices, because you can keep selling it for at least a year.
in the pressent you have to take what you can get. your hardware will be outdated faster then you are able to develop a new product.

[Rampion]

Ok so what is so bad about the fact that Bitcoin now uses ASIC's that has everyone using Scrypt or other ASIC protected code?  Is it because they don't want to have to invest the cash for Bitcoin, or is there another reason that seriously disrupts the effectiveness of Bitcoin.  I understand they say it centralizes Bitcoin, but as the price rises the whales will get out earlier because they got in earlier so centralization doesn't seem to be a problem to me.  What is it?  Why use Scrypt?

ASIC's are a very important step towards mining centralization. That's it.

[nlsupernova]

Ok so what is so bad about the fact that Bitcoin now uses ASIC's that has everyone using Scrypt or other ASIC protected code?  Is it because they don't want to have to invest the cash for Bitcoin, or is there another reason that seriously disrupts the effectiveness of Bitcoin.  I understand they say it centralizes Bitcoin, but as the price rises the whales will get out earlier because they got in earlier so centralization doesn't seem to be a problem to me.  What is it?  Why use Scrypt?

ASIC's are a very important step towards mining centralization. That's it.

asics are a an important step to making cheap efficient hardware available for everyone. ultimatly fighting centralisation.

[Against Hunger]

i was the one who said that. the problem with your statement the way i see it is quit simple. asics are becomming obsolete far to quickly.
chips that are made now are surpased with chips with double the power within months. (i`ll say it again, thats because current chips are based on manufacturing processes that are considerd old by the rest of the industry)
the people who are rushing these things to market are not out to build a brand or something, they wanna get rich quick.
plus they dont have enough r&D to sustain that model. they don`t have the money to design the new generation of chip technology, thats done by the big dogs.
once we hit 20nm,  the huge performance gains are a thing of the past.

I will agree in the past ASIC's have been built on legacy technology so it was fairly easy to develop a faster model. The reason this occurred was it was cheaper to manufacture 135nm versus 110nm versus 65nm versus 28nm versus 20nm. The cycle has been proven and has continued to 28nm and we are already seeing 20nm announcements from KNC and others but I do not believe as you state that "huge performance gains are a thing of the past" for several reasons.

1) We are not at the edge of Moores Law. I agree we are close but not their yet. The majority of ASICs are coming from 2 foundries TSMC and Global Foundries. Both companies have the ability to go to 16nm today. At 16nm they are now getting close to Moores Law but they are not there yet.  A couple of companies (Intel and IBM) can go beyond 16nm to 14nm but at that point the industry hits a wall until the next jump to 10nm (2015/2016) but it will take a couple more years until mass production can occur at the 10nm size. To develop at 14nm is expensive so I do not believe we will see products at this level until we have exhausted all other options.

2) Current ASIC's have essentially been a re-targeting of FPGA's without any significant improvements in design.  I believe there is a significant amount of research underway in this space and I am certain current developers are optimizing their designs. This level of optimization should allow a few more iterations to increase the hash to power ratio's. Look at Bitfury's design extremely efficient ASICs but still on 65nm tech and GHASH.IO have shown that with a great design and large quantities you can still be relevant against new technology. Companies like 21e6 (ex samsung) certainly know a lot about ASIC design.

So yes we are getting close to the edge of Moores Law and I agree when we get there we should see a good 2 year run where we can move from Niche players with limited access to next gen ASIC to a proliferation of low cost competitive ASIC's but we are not there yet.

2015 will be an interesting year.

[ljudotina]

Ok so what is so bad about the fact that Bitcoin now uses ASIC's that has everyone using Scrypt or other ASIC protected code?  Is it because they don't want to have to invest the cash for Bitcoin, or is there another reason that seriously disrupts the effectiveness of Bitcoin.  I understand they say it centralizes Bitcoin, but as the price rises the whales will get out earlier because they got in earlier so centralization doesn't seem to be a problem to me.  What is it?  Why use Scrypt?

ASIC's are a very important step towards mining centralization. That's it.

asics are a an important step to making cheap efficient hardware available for everyone. ultimatly fighting centralisation.

And as there would not be centralization if there was no ASIC.
So what your sain is that we have ASIC to fight centralization which was created by ASIC....wouldnt it be easyer not to have ASIC in first place?

Imagine how much more miners would BTC have than it has now. How much harder would it be to centralize it and shut it down if there was no ASIC around...
More miners, more BTC users, more users wider adoption, wider adoption better BTC price....ASICs just ruined initial idea of BTC.

[induktor]

Original poster asked an invalid question.  The valid question is:

"What benefit does ASIC bring over CPU/GPU?"

The answer is none.

There's no energy consumption benefit.  Corporations simply upscale their hardware size, using the same amount of power, until all inventory on the entire internet is out of stock.

ASIC brings no benefits, problems too numerous to list, and also severely reduces market penetration.  

A currency with no market penetration has no liquidity.  With no liquidity, it's not a currency.

Why do you think Dogecoin has more daily transactions than Bitcoin?  Market penetration.

Wow, touche!
I was thinking exactly about doge coin phenomena, you are correct!.

[nlsupernova]

i was the one who said that. the problem with your statement the way i see it is quit simple. asics are becomming obsolete far to quickly.
chips that are made now are surpased with chips with double the power within months. (i`ll say it again, thats because current chips are based on manufacturing processes that are considerd old by the rest of the industry)
the people who are rushing these things to market are not out to build a brand or something, they wanna get rich quick.
plus they dont have enough r&D to sustain that model. they don`t have the money to design the new generation of chip technology, thats done by the big dogs.
once we hit 20nm,  the huge performance gains are a thing of the past.

I will agree in the past ASIC's have been built on legacy technology so it was fairly easy to develop a faster model. The reason this occurred was it was cheaper to manufacture 135nm versus 110nm versus 65nm versus 28nm versus 20nm. The cycle has been proven and has continued to 28nm and we are already seeing 20nm announcements from KNC and others but I do not believe as you state that "huge performance gains are a thing of the past" for several reasons.

1) We are not at the edge of Moores Law. I agree we are close but not their yet. The majority of ASICs are coming from 2 foundries TSMC and Global Foundries. Both companies have the ability to go to 16nm today. At 16nm they are now getting close to Moores Law but they are not there yet.  A couple of companies (Intel and IBM) can go beyond 16nm to 14nm but at that point the industry hits a wall until the next jump to 10nm (2015/2016) but it will take a couple more years until mass production can occur at the 10nm size. To develop at 14nm is expensive so I do not believe we will see products at this level until we have exhausted all other options.

2) Current ASIC's have essentially been a re-targeting of FPGA's without any significant improvements in design.  I believe there is a significant amount of research underway in this space and I am certain current developers are optimizing their designs. This level of optimization should allow a few more iterations to increase the hash to power ratio's. Look at Bitfury's design extremely efficient ASICs but still on 65nm tech and GHASH.IO have shown that with a great design and large quantities you can still be relevant against new technology. Companies like 21e6 (ex samsung) certainly know a lot about ASIC design.

So yes we are getting close to the edge of Moores Law and I agree when we get there we should see a good 2 year run where we can move from Niche players with limited access to next gen ASIC to a proliferation of low cost competitive ASIC's but we are not there yet.

2015 will be an interesting year.

yes but the steps from 120 to 28 nm took the industry maybe 2 decades. we went through the whole cybvle in one year.
and performance will improve. just like the new generation of gpus  are also 10 procent faster. and next years release wil be 10 percent faster again. same goes for intel chips. hell even cars are on the same cycle. steady small increments of eficienty with every new release.
btc asics have become a 1000 times faster in a year or something, thats just not going to last. if this where to continue we would have 100 petahash machines the size of an antminer by may 2015. that just isn`t going to happen, because the room for improvement to catch up with current technology isn`t there anymore

and yes, eventually if the easy progress by schrinking is done for, the designs will be tweaked for effinciency. but that wont have anywhere near as much impact as schrinking does, plus it takes a lot more time and money to develop these new architectures.

a 10 to 20% percent improvement in effecienty per year is where it will be in the not so distant future. i think thats realistic and is proven by the entire tech industry in the last decade.

and until samsung and the likes anounce a new technology that makes asics in general obsolete, we don`t have to be scared there will be anything else to kill asics, because this is the best way to do these kinds of tasks.

[nlsupernova]

Ok so what is so bad about the fact that Bitcoin now uses ASIC's that has everyone using Scrypt or other ASIC protected code?  Is it because they don't want to have to invest the cash for Bitcoin, or is there another reason that seriously disrupts the effectiveness of Bitcoin.  I understand they say it centralizes Bitcoin, but as the price rises the whales will get out earlier because they got in earlier so centralization doesn't seem to be a problem to me.  What is it?  Why use Scrypt?

ASIC's are a very important step towards mining centralization. That's it.

asics are a an important step to making cheap efficient hardware available for everyone. ultimatly fighting centralisation.

And as there would not be centralization if there was no ASIC.
So what your sain is that we have ASIC to fight centralization which was created by ASIC....wouldnt it be easyer not to have ASIC in first place?

Imagine how much more miners would BTC have than it has now. How much harder would it be to centralize it and shut it down if there was no ASIC around...
More miners, more BTC users, more users wider adoption, wider adoption better BTC price....ASICs just ruined initial idea of BTC.

if the network still ran on cpu, a single asic machine develloped by an evil gouverment in secrety would kill btc overnight. asics are not special and defenatly not invented for btc. bitcoin adpted the concept. and if we as a comunity didn`t do it, it would be very easy for the ones who are evil to do just that.

to protect the network we have to make sure it is kept to the highest standard available.

[induktor]

I don't get those people who complain that ASICs won't ROI, and then they go ahead and build giant GPU farms that won't profit unless they can count on reselling their used hardware on eBay for >75% cost after 1 year.

Where do you get this idea?
All of the asics that I have, never made ROI, some of them are close, but not there yet and I has been months! more than 8.
3 of 4 of my GPU farm, give me ROI in less than 5 months. (not counting that I can resell the GPUs that are all working perfectly, and recover 60% more!).
GPUs are FAR more profitable and a more sound business than buy ASICS, my numbers, and the cash I have in my pocket, is telling me that.

[Biomech]

Ok so what is so bad about the fact that Bitcoin now uses ASIC's that has everyone using Scrypt or other ASIC protected code?  Is it because they don't want to have to invest the cash for Bitcoin, or is there another reason that seriously disrupts the effectiveness of Bitcoin.  I understand they say it centralizes Bitcoin, but as the price rises the whales will get out earlier because they got in earlier so centralization doesn't seem to be a problem to me.  What is it?  Why use Scrypt?

ASIC's are a very important step towards mining centralization. That's it.

asics are a an important step to making cheap efficient hardware available for everyone. ultimatly fighting centralisation.

And as there would not be centralization if there was no ASIC.
So what your sain is that we have ASIC to fight centralization which was created by ASIC....wouldnt it be easyer not to have ASIC in first place?

Imagine how much more miners would BTC have than it has now. How much harder would it be to centralize it and shut it down if there was no ASIC around...
More miners, more BTC users, more users wider adoption, wider adoption better BTC price....ASICs just ruined initial idea of BTC.

Yeah, and those damn looms put hand tailors out of business and ruined our lives. Best go throw some wooden shoes in the works. That'll show 'em.

Be real. It was going to happen. You put geeks on a technical issue, they WILL make a better machine. Right now, it's the middle of an arms race. It'll settle down, prices will come down to meet reality, and there will be a place for scrypt as well.

As for ASICS enabling centralization, well shit me george, were there not pools taking up most of the network before Avalon? No? sure seemed that way to me. Face it: ASIC's are nothing new. They've been in watches and the like for decades. This is just a new application for an existing process. They do it more efficiently, thus they have taken over one type of cryptocurrency mining. Now there are beginning to be scrypt ASICS as well. Technology is in a constant state of flux. It always will be. Prior to the ASIC revolution, had Bitcoin actually gone mainstream and gotten the attention of the world's rulers, they would have soaked you via taxation to put a supercomputer or two on the job, and they would OWN that network. This has become much more difficult for them. It couldn't be done silently or cheaply at this time. BECAUSE of ASIC's.

Eventually, as others have pointed out, the improvement will become incremental rather than crazy. Already it has slowed somewhat. We're not seeing 40 percent jumps in difficulty anymore, and probably never will again unless somebody develops something even more efficient than the ASIC's. Evolution goes in fits and starts. The revolutionary part is an exclamation point, but it's not the end of the paragraph.

[nlsupernova]

i think by the middle of this year when the neptunes, monacrhs and presperos hit the market we wil see 30% + jumps once more. but that will be the last time until asics as a concept become obsolete in the entire industry.
that will probably be caused by things like grafeen chips or quantum computing or something like that. and thats not consumer ready by a long shot, will take at least 5 years. but more like a decade if you ask me.

a few years from now we`re having a beer somewhere talking about those crazy years when bitcoin was comming to adulthood, and the way people went crazy. we`ll have a few laughs and have warm fuzy feelings when thinking about the way things used to be. and that it actually was a fun thing to be a part of.

[Biomech]

i think by the middle of this year when the neptunes, monacrhs and presperos hit the market we wil see 30% + jumps once more. but that will be the last time until asics as a concept become obsolete in the entire industry.

You may be right, but I actually think it'll still be in the neighborhood of 25 percent max, because the "obsolete" chips are still being disseminated, and 28nm chips from at least one manufacturer are already being put out in multiple configurations as well. I don't see 20nm die shrink as a big deal, honestly. I think it will at best give marginal gains over 28. Optimizing the design, even at higher process nodes, will give a better yield. I think that's where it will go. As already noted, the current bitcoin ASIC's are basically reduced size FPGA's, whereas they could be completely purpose designed. Doing so would make larger silicon real estate more valuable. The race to the bottom on process nodes is kind of a stunt, and definitely a shortcut over what CAN be done. If people start to shy away from the preorder bullshit, I think we'll start to see more "ground up" designs, and more efficiency because of it. There is a lot of redundancy to an FPGA, because it's NOT application specific. Fewer functions equal more real estate, less power consumption. I don't think 110nm was really finished out for potential. There's still room for a lot of improvement, and frankly most of it's design work.

that will probably be caused by things like grafeen chips or quantum computing or something like that. and thats not consumer ready by a long shot, will take at least 5 years. but more like a decade if you ask me.

Quantum's a long shot. I don't think it's going mainstream in a decade. I could be wrong, but I've been hearing how it's "right around the corner" for two decades. Instead we still have expensive and impressive toys in a few universities.

I am utterly unfamiliar with grafeen. Wazzat?

[nlsupernova]

i think by the middle of this year when the neptunes, monacrhs and presperos hit the market we wil see 30% + jumps once more. but that will be the last time until asics as a concept become obsolete in the entire industry.

You may be right, but I actually think it'll still be in the neighborhood of 25 percent max, because the "obsolete" chips are still being disseminated, and 28nm chips from at least one manufacturer are already being put out in multiple configurations as well. I don't see 20nm die shrink as a big deal, honestly. I think it will at best give marginal gains over 28. Optimizing the design, even at higher process nodes, will give a better yield. I think that's where it will go. As already noted, the current bitcoin ASIC's are basically reduced size FPGA's, whereas they could be completely purpose designed. Doing so would make larger silicon real estate more valuable. The race to the bottom on process nodes is kind of a stunt, and definitely a shortcut over what CAN be done. If people start to shy away from the preorder bullshit, I think we'll start to see more "ground up" designs, and more efficiency because of it. There is a lot of redundancy to an FPGA, because it's NOT application specific. Fewer functions equal more real estate, less power consumption. I don't think 110nm was really finished out for potential. There's still room for a lot of improvement, and frankly most of it's design work.

that will probably be caused by things like grafeen chips or quantum computing or something like that. and thats not consumer ready by a long shot, will take at least 5 years. but more like a decade if you ask me.

Quantum's a long shot. I don't think it's going mainstream in a decade. I could be wrong, but I've been hearing how it's "right around the corner" for two decades. Instead we still have expensive and impressive toys in a few universities.

I am utterly unfamiliar with grafeen. Wazzat?

it mostly depends on how much funding goes to these types of things. if a country like the USA or something finds a way to have a a big competitive advantage in some field by having that technology, things can move very quickly. but on the other side, it might never come at all. i don`t know enough about it to make any usefull predictions.
the only reason the atomic bomb was develloped in 5 years was because of war and endless funding. if it wasn`t for ww2 nuclair technology would probably have been 25 years behind and nuclair reactors where just now becoming the new cool thing.

grafeen is basicly a molecule thick cunductor when used in chips, but it has a wide range of uses. very versitile.. its actually just carbon i believe. its nano technology on steroids. i don`t know that much about it either. but if your into that kind of technology you schould check it out. crazy stuff but very very cool!

[Biomech]

i think by the middle of this year when the neptunes, monacrhs and presperos hit the market we wil see 30% + jumps once more. but that will be the last time until asics as a concept become obsolete in the entire industry.

You may be right, but I actually think it'll still be in the neighborhood of 25 percent max, because the "obsolete" chips are still being disseminated, and 28nm chips from at least one manufacturer are already being put out in multiple configurations as well. I don't see 20nm die shrink as a big deal, honestly. I think it will at best give marginal gains over 28. Optimizing the design, even at higher process nodes, will give a better yield. I think that's where it will go. As already noted, the current bitcoin ASIC's are basically reduced size FPGA's, whereas they could be completely purpose designed. Doing so would make larger silicon real estate more valuable. The race to the bottom on process nodes is kind of a stunt, and definitely a shortcut over what CAN be done. If people start to shy away from the preorder bullshit, I think we'll start to see more "ground up" designs, and more efficiency because of it. There is a lot of redundancy to an FPGA, because it's NOT application specific. Fewer functions equal more real estate, less power consumption. I don't think 110nm was really finished out for potential. There's still room for a lot of improvement, and frankly most of it's design work.

that will probably be caused by things like grafeen chips or quantum computing or something like that. and thats not consumer ready by a long shot, will take at least 5 years. but more like a decade if you ask me.

Quantum's a long shot. I don't think it's going mainstream in a decade. I could be wrong, but I've been hearing how it's "right around the corner" for two decades. Instead we still have expensive and impressive toys in a few universities.

I am utterly unfamiliar with grafeen. Wazzat?

it mostly depends on how much funding goes to these types of things. if a country like the USA or something finds a way to have a a big competitive advantage in some field by having that technology, things can move very quickly. but on the other side, it might never come at all. i don`t know enough about it to make any usefull predictions.

grafeen is basicly a molecule thick cunductor. nano technology on steroids. i don`t know that much about it either. but if your into that kind of technology you schould check it out. crazy stuff but very very cool!

Been into nanotech since Engines of Creation hit the market. Now I'm just waiting for the final breakthrough for assemblers Yeah, I'll definitely check it out. Thanks!

[nlsupernova]

yeah i like that stuf to, but i don`t have enough background in those fields to really get into it. but its fascinating nonetheless