I agree. The continued ability to produce finer and finer lines absolutely astounds me.
I remember when all the experts were saying that 100 nm might be the limit - and the same statement has been made half a dozen times since then.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
The fact that all of this even works and makes chips that actually work is absolutely staggering. Implying that Intel of all companies is just dragging to extract more profits from the current technology node is absolutely ridiculous. If it weren't for Intel, we would probably still be stuck at 65 nm or something, the amount of money they throw at advancing chip production technology basically funds half of the semiconductor production supply industry.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
I also like the idea of taking steps to get there for the sole purpose of giving us more time.
Us meaning them, but work with me here. Why do we (they) need time? Well, we're working on the Next Big Thing.
Quantum computing.
Intel has a classical chipset roadmap out to Skymont in 2017. After that, it might not be possible to make chips smaller. If not, there'll have to be work done in other areas. Sure, we can mess around with different types of transistors and 3D circuits and what not, but it looks like quantum stuff is where we'll need to go in the end. More work in high-temperature superconductors certainly won't hurt that.
We might see the first personal quantum computer in the mid to late '20s.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
The fact that all of this even works and makes chips that actually work is absolutely staggering. Implying that Intel of all companies is just dragging to extract more profits from the current technology node is absolutely ridiculous. If it weren't for Intel, we would probably still be stuck at 65 nm or something, the amount of money they throw at advancing chip production technology basically funds half of the semiconductor production supply industry.
And that reinforces my point. Years ago, the consensus was that the limit was somewhere around 100 nm. Later, the consensus was that the limit was around 60 nm. Then the consensus was that the limit was around 30 nm. And so on.
Eventually, they'll be right, but it's really interesting how many times the consensus has been wrong on this particular issue. Either the scientists are incredibly brilliant or the 'experts' are incredibly uninformed. Or both.
This has nothing to do with manufacturing process issues. Intel could go straight to 11 nm and beyond (nanotechnology) if they wanted right now. It is just milking the process as much as possible.
Sure they can. They can built the circuit by manipulating atoms one at a time to create wire as thin as 1 atom. Of course each chip will probably cost you more than 100k to buy since no high volume production machines are available for that kind of process.
And that reinforces my point. Years ago, the consensus was that the limit was somewhere around 100 nm. Later, the consensus was that the limit was around 60 nm. Then the consensus was that the limit was around 30 nm. And so on.
Eventually, they'll be right, but it's really interesting how many times the consensus has been wrong on this particular issue. Either the scientists are incredibly brilliant or the 'experts' are incredibly uninformed. Or both.
Yeah, I think of this often. I remember researching a report on chip fabrication back in the day, and many books (LOL--books as sources of information on cutting edge technology--remember that?) held up the 100 nm limit as a theoretical limit. Several sources doubted that even getting close to 100 was technologically feasible even if it was theoretically possible.
It does boggle the mind how fast the impossible becomes possible these days!
"Maloney attributes the "adjustment" to "the new manufacturing process needed to make the smaller chips."
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
It really is amazing. Of course I'm also amazed at the ceramic coating on my Toto toilet that makes cleaning the bowl unnecessary.
$100 billion, are you suggesting Apple is saving up to buy Intel?
Imagine the fear and furore from the rest of the PC industry!
I don't think such a move will make it past the regulators.
Well, you seem to be a bit new to how the WORLD NOW WORKS. If Apple wanted to buy INTEL, they'd have to rejoin the US Chamber of Commerce, and support a lot of candidates in November (money = speech, so Apple could be loudest). Then they'd hire the usual expert pundits to go around the radio and TV and explain why we all NEED Apple + INTEL to make us competitive with the Chinese.
Eventually, everyone will just understand that this is the best possible thing for US competition and jobs, jobs, jobs. (in Brazil and China -- but who's paying attention, right?).
Apple spreads a bit more money around and former FCC and Antitrust bureaucrats are now "history" consultants for Apple's board of directors and make $750,000 a year for two week seminars at some tropical island getaway (they'll show some power-points {OK, Keynotes} so it's real legit). That's chump change.
>> if Apple wanted to, they could certainly buy INTEL -- but they shouldn't. They should buy ARM.
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
You are right, 22nm was a MAJOR hurdle, I do remember that it was deemed impossible at one time. The jump down to 14 nm may not happen at the same pace -- I'm pretty sure. At this point, they are doing more error correction for stray electrons than computations. Some new material besides Silicon will need to be used.
Maybe it will be on diamond -- who knows? There are a few meta-materials that show promise. I'm a fan of light myself -- even though it is in some ways requires larger gates -- you can process with more than one frequency at a time if you can create gates that can stop light at specific frequencies at the same time.
>> Anywho, this is probably someone who thinks modern chip technology is all based on captured Alien tech, and who haven't tracked the blood, sweat and tears of hard working humans who have incrementally improved chips over the years.
Comments
The Mac accounts for an increasingly smaller percentage of Apple's revenue each year....
But that is still MORE Macs being produced every year.
An ever INCREASING need for chips to build them.
I agree. The continued ability to produce finer and finer lines absolutely astounds me.
I remember when all the experts were saying that 100 nm might be the limit - and the same statement has been made half a dozen times since then.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
The fact that all of this even works and makes chips that actually work is absolutely staggering. Implying that Intel of all companies is just dragging to extract more profits from the current technology node is absolutely ridiculous. If it weren't for Intel, we would probably still be stuck at 65 nm or something, the amount of money they throw at advancing chip production technology basically funds half of the semiconductor production supply industry.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
I also like the idea of taking steps to get there for the sole purpose of giving us more time.
Us meaning them, but work with me here. Why do we (they) need time? Well, we're working on the Next Big Thing.
Quantum computing.
Intel has a classical chipset roadmap out to Skymont in 2017. After that, it might not be possible to make chips smaller. If not, there'll have to be work done in other areas. Sure, we can mess around with different types of transistors and 3D circuits and what not, but it looks like quantum stuff is where we'll need to go in the end. More work in high-temperature superconductors certainly won't hurt that.
We might see the first personal quantum computer in the mid to late '20s.
I think right now, the general consensus is that the limit is somewhere between 8 and 10 nm, but before we get there, lots and lots and lots of things need to be sorted out, as we've more or less reached the point where the box of tricks that allows us to use deep UV light (which has a wavelength of about 190 nm) to print features at the current nodes (which are 22 and 28 nm) is almost exhausted.
The fact that all of this even works and makes chips that actually work is absolutely staggering. Implying that Intel of all companies is just dragging to extract more profits from the current technology node is absolutely ridiculous. If it weren't for Intel, we would probably still be stuck at 65 nm or something, the amount of money they throw at advancing chip production technology basically funds half of the semiconductor production supply industry.
And that reinforces my point. Years ago, the consensus was that the limit was somewhere around 100 nm. Later, the consensus was that the limit was around 60 nm. Then the consensus was that the limit was around 30 nm. And so on.
Eventually, they'll be right, but it's really interesting how many times the consensus has been wrong on this particular issue. Either the scientists are incredibly brilliant or the 'experts' are incredibly uninformed. Or both.
Twice the resolution = four times the pixels.
Super Hi-Vision, for example, has four times the resolution and 16 times the pixels as 1080p. Get it?
From the silence I think he's discovering his own math skills relative to the context of the subject matter is sorely lacking.
This has nothing to do with manufacturing process issues. Intel could go straight to 11 nm and beyond (nanotechnology) if they wanted right now. It is just milking the process as much as possible.
Sure they can. They can built the circuit by manipulating atoms one at a time to create wire as thin as 1 atom. Of course each chip will probably cost you more than 100k to buy since no high volume production machines are available for that kind of process.
And that reinforces my point. Years ago, the consensus was that the limit was somewhere around 100 nm. Later, the consensus was that the limit was around 60 nm. Then the consensus was that the limit was around 30 nm. And so on.
Eventually, they'll be right, but it's really interesting how many times the consensus has been wrong on this particular issue. Either the scientists are incredibly brilliant or the 'experts' are incredibly uninformed. Or both.
Yeah, I think of this often. I remember researching a report on chip fabrication back in the day, and many books (LOL--books as sources of information on cutting edge technology--remember that?) held up the 100 nm limit as a theoretical limit. Several sources doubted that even getting close to 100 was technologically feasible even if it was theoretically possible.
It does boggle the mind how fast the impossible becomes possible these days!
(So where is my FTL drive??)
Problems with the 22nm process:
"Maloney attributes the "adjustment" to "the new manufacturing process needed to make the smaller chips."
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
It really is amazing. Of course I'm also amazed at the ceramic coating on my Toto toilet that makes cleaning the bowl unnecessary.
$100 billion, are you suggesting Apple is saving up to buy Intel?
Imagine the fear and furore from the rest of the PC industry!
I don't think such a move will make it past the regulators.
Well, you seem to be a bit new to how the WORLD NOW WORKS. If Apple wanted to buy INTEL, they'd have to rejoin the US Chamber of Commerce, and support a lot of candidates in November (money = speech, so Apple could be loudest). Then they'd hire the usual expert pundits to go around the radio and TV and explain why we all NEED Apple + INTEL to make us competitive with the Chinese.
Eventually, everyone will just understand that this is the best possible thing for US competition and jobs, jobs, jobs. (in Brazil and China -- but who's paying attention, right?).
Apple spreads a bit more money around and former FCC and Antitrust bureaucrats are now "history" consultants for Apple's board of directors and make $750,000 a year for two week seminars at some tropical island getaway (they'll show some power-points {OK, Keynotes} so it's real legit). That's chump change.
>> if Apple wanted to, they could certainly buy INTEL -- but they shouldn't. They should buy ARM.
Too many of you have no idea how amazing it is to produce 22nm chips. It doesn't seem like all that long ago (maybe 20 years) when there were discussions about whether chip manufacturers could break the 100nm barrier.
At 22nm I think someone had to have sold their soul to dark forces.
You are right, 22nm was a MAJOR hurdle, I do remember that it was deemed impossible at one time. The jump down to 14 nm may not happen at the same pace -- I'm pretty sure. At this point, they are doing more error correction for stray electrons than computations. Some new material besides Silicon will need to be used.
Maybe it will be on diamond -- who knows? There are a few meta-materials that show promise. I'm a fan of light myself -- even though it is in some ways requires larger gates -- you can process with more than one frequency at a time if you can create gates that can stop light at specific frequencies at the same time.
>> Anywho, this is probably someone who thinks modern chip technology is all based on captured Alien tech, and who haven't tracked the blood, sweat and tears of hard working humans who have incrementally improved chips over the years.