...what I tell people in "traditional" engineering disciplines is to jump to industry and let your employer pay for grad school
Most good engineering school have a lot of support for grad students, so jumping to industry will just slow you down. With my Research Assistantships and Teaching Assistantships, I haven't paid a dime to go to grad school and I've been here four years. The student salaries are enough for rent, too, so I haven't had to get any loans. I am a traditional engineer, by the way. I'm getting a Ph.D. in Electrical Engineering.
A big problem with industry paying for your master's degree is that they usually don't give you a raise after you finish big enough to compensate what you could have gotten with a Master's in the first place. And you can't just quit, because then you have to pay back some of the fees they paid you!
I own a small business and I can honestly say that I would rather hire someone with 4 years of experience than someone with a PHD.
Probably someone with a Ph.D. would be overqualified for the jobs for which you would hire. On the other hand, if you were hiring people to design adaptively calibrated analog-to-digital converters, for instance, I think you would be better off hiring a Ph.D. than a B.S. with 4 years of circuit design experience. This is because getting a Ph.D. forces you to truly know what you are doing, and when you are working at a very high level, knowing what you are doing is important. In advanced circuit design, for the most part the B.S. and most of the M.S. engineers work for the Ph.D. designers and the best of the M.S. engineers.
Here's what I mean. When I was an undergrad, I could study for tests and get A's without fully grasping what is going on. (How many people who took Calculus REALLY understand Calculus? Maybe one in ten) Likewise, when I was working, I could get by with a cursory knowledge of the material because I had a small part of the project and I was doing rather routine stuff. Now that I'm a Ph.D. student, the success of the project depends entirely on one person, ME. I need to know every last detail because there is no one else who will do it. This means I need to deeply understand the fundamentals, be able to absorb new information quickly, and become truly independent. This is difficult to achieve when you are working on a project and not running it yourself, because when you don't have to deal with the big picture it is easy to become nearsided. Most B.S. engineers I know are fairly competent, but they only know what they can look up. They are fine for routine designs, but they tend to lack the deep knowledge needed to extend the state of the art.
This is not to say every Ph.D. engineer is a better designer than every B.S. engineer, because that is not true. But, I submit that every B.S. engineer would be better if he or she went through the trial by fire of a Ph.D.
I agree with your observations of the current state of the gaming industry, but I don't agree with your conclusions.
If you're old enough to remember 1984, you may remember that demand for the Colecovision (like PS2 today) was VERY strong in 1983/early 1984 and we were all sitting on our hands waiting for the supersystem from Atari, the 7800 (like Xbox today). A huge older system, the Atari 2600 was still selling well (like PS1 today). The market quite suddenly crashed under its own weight and the 7800 was scrapped, only to re-emerge in 1986 as a half-assed American answer to NES. I remember bargin bins of Atari 2600 games for like $9.99 when they were selling for more like $40 and everything was rosey a couple of months before.
1984 is much more similar to today than you think. The big players were raking in mad cash and awesome systems loomed on the horizon, just like today. Your likely scenario may come to pass, but I think a crash may be equally likely.
Someone said s/he thinks gaming won't die, but consoles will. I agree strongly with both that statement and with the substance of the article. There are many parallels between 1984 and 2001 (sounds like a book review!). But remember, computer gaming did NOT die in 1984, just consoles (Atari 2600/5200, Intellivision, Colecovision, etc)
In 1984 my brother and I were both turned off by the crap being sold for consoles and turned to the commodore 64, which had a very healthy gaming market, and never skipped a beat (plus, you could actually PROGRAM it. What a beautiful assembly language it had!). People shouldn't talk about the video game crash of 1984 but rather about the console crash of 1984, because all my friends at the time were playing tons of games on their commodores and apple IIs and we were all lusting after the AMIGA!
Video games didn't crash in 1984; consoles did. I think the same statement may be applicable in 2002, as the author of the article indicates.
First, the converter is clocked at 12.8 GHz, which is very different from running at 12 GS/s. It is definately an oversampled converter, because they mention a digital decimation filter, but they don't give you the oversampling ratio. By oversampling, I mean the input is sampled higher than the Nyquist rate and noise shaped. Then, it is digitally filtered back down to Nyquist with significantly less quantization noise than you would have had without oversampling. A common oversampling ratio is 128 which would put the converter at 90 MHz signal bandwidth, which is quite high for an oversampled converter. I'm totally guessing on the oversampling ratio, though.
As for the number of bits it can do per sampling, that is largely irrelevent in a communications context. What matters is the SNR. You can figure out the number of bits of linearity with the following equation: SNR = (6*B + 2) dB where B=the effective number of bits. In other words, if you have an SNR of 60 dB, you have just about 10 bits of accuracy. The reason accuracy is specified as an SNR rather than a number of bits is if the converter is nonlinear (as it always is in practice) then even if you have more bits of resolution, the additional bits are inaccurate and should be ignored. For example, if you have a so-called "12 bit" converter with an SNR of 63 dB, then although you have 4096 possible output codes, the uncertainty between them is enough that the lower 2 LSBs are garbage, and, although you have 12 bits of resolution, you only have 10 bits of accuracy. This is a common way for manufacturers to lie on data sheets. Keep in mind, though, there are situations where resolution is more important than accuracy (such as digital imaging) and you'd rather have a 12-bit / 10 bit accurate converter than a 10-bit / 10 bit accurate converter.
Hope that cleared things up. If you have more questions, email me or reply to this post. I just love talking about data-converters!
Applying concepts like that to the brain is useless. The brain is unbelieveably parallel, and runs asynchronously, so "how fast can we think?" really doesn't mean anything. I know you were just thinking of the sci-fi aspects of this and all, but this is one nut that just ain't gonna be cracked.
Disclaimer: I design analog-to-digital converters for a living.
The market the article talks about is the analog to digital converter market, not the desktop market.
True enough. I stand by my statements, however. For one thing, the article discusses an A/D fabricated in the technology. The die size was 1 cm^2. That is truly enormous and very would be extremely expensive as a product. Even if they can bring down the lithography, it is still very expensive. Second, while they say it runs at 12.8 GHz, because of the decimation filter it is obviously an oversampled converter but they don't give the oversampling ratio, so we have no idea of the actual conversion rate. I drew parallels between III-V materials (which have been around since the 1960s and must have achieved some kind of maturity) with superconducting electronics (which have been around since the late 1970s) and I think they still stand.
My belief is that this is a laboratory curiosity with little commerical potential. I'm sure the military is very interested in using it with radar, however.
By the way, the IEEE is well known for pumping up "cutting edge" technologies that never reach their potential. Remember "fuzzy logic"?
Doesn't this wildly violate it? Right now we're at 1.5 GHz. Moore's law states that this will double every 18
months. That's about three times it should double in the next five years. La la la, David does some simple
math:
How many times has this been said on/.? Moore's law is about the density of DRAM, and by implication, the density of other CMOS circuits. It says nothing about clock rate. It is true that smaller transistors are faster, but there are other problems with clocks that smaller transistors make more difficult, most notably clock skew. There is already logic out there that can go faster than 20 GHz, but it is LSI GaAs logic (flip-flops / gates / adders).
Moore's law has nothing to do with superconductors, at all. We may never see mass-market superconducting logic. It will be just too expensive, and it could be impossible to integrate well enough for computers.
Superconducting logic has been out for a VERY VERY long time. In fact, IBM burned tens of millions of dollars on the subject in the 1970s. The problems with superconductors are even WORSE than the problems with superfast III-V logic. UCSB has 70 GHz flip-flops made out of transferred-substrate heterojunction (III-V, Indium Phosphide and something else) transistors, but nobody thinks they will revolutionize computing, because they won't. So it is with superconducting logic.
There are two huge problems with superconducting logic that don't seem solvable in the near future. They are:
1. Cost : These things are enormously expensive to manufacture and operate, and it is the economy of scale of CMOS techology which has enabled, more than anything, the current computing revolution. Do you have any idea how expensive coolant and the dewar to use it in are to get something to 5K? Even the so-called "high-temperature" superconductors have to be pretty damn cold to function; they just don't need to go so close to absolute-zero.
2. Integration This is probably the killer. It will be extremely difficult to integrate many devices together. Even if myriad technical difficulties are overcome, the solution is not likely to be inexpensive, as CMOS technology is. For III-V semiconductors (which use much less exotic materials than superconductors), high defect rate, problems with lattic matching of the materials, and the lack of a high quality native oxide (like SiO2 in silicon) have made it impossible to achieve integration levels anywhere close to that achieved in silicon. Even GaAs, the most well-understood III-V semiconductor, can't be integrated to more than a few thousand devices. That's why we don't have 20 GHz GaAs microprosessors. And superconductors are even HARDER to deal with.
In summary, even if researchers are able to overcome almost insurrmountable odds to find away to reliably integrate meaningful numbers of these devices on a single die, I think it is very unlikely they will be able to do it cheaply, which is just as important as being able to do it at all. Otherwise, this technology will be of interest only to the military.
By the way, I know III-V semiconductors have a lot of very important uses, especially in optics and RF. It is a fact, however, that III-V logic is mainly of interest to the military and the space industry.
The seasons - the 23-degree axial tilt that lets the sun drive energy back and forth across the surface of the planet I'll accept that big whacks are pretty common (Uranus' axial tilt, etc.)
The seasons don't do anything to promote life. By far the greatest concentrations of life in the tropics; both on land and in the sea, and they don't have discernable seasons. In fact, seasons are a hurdle to life: places with the largest seasonal differences (like the arctic) have the least biodiversity.
So what the fuck do these 100 people (including 30 PhDs) in the research department work on?
In large engineering projects, such as search engines, the amount of work to be done is something more than linearly related to the size of the project. Back when google was two guys, they had defined the key algorithm and it was really cool. But.... really cool is a long, long way from a business. For example, how do you index over a billion pages and still keep the search fast? How do you distribute the database and the searches over linux machines to reduce cost at little (or no) expense in reliability? How do you keep crafty webmasters from tricking your algorithms?
Do you really think google is the same as it was, just bigger? Give me a break.
Just why do the companies, even the great ones, think that their headcount MUST grow?
They don't. Exar is the world leader in analog interfaces for digital imagers and they've been around since the 1960s. They have around 200 employees.
It has been tried before, but an insulating substrate is costly. Both RCA and HP tried "SOS" chips, silicon-on-saphire, but the advantages didn't compensate for the much higher cost.
SOI is significantly different from SOS, for one thing the insulator is not sapphire, and the advantages do look like they will be able to compensate for the higher cost. Chief among them are freedom from body effect and latch-up and ultra low-voltage operation. People have been interested for years in SOI technologies, including SOS, but have in the past only used them for radiation-hard military/space type things because of the cost. Today, with much lower defect densities in SOI wafers, the cost is decreasing dramatically.
VCRs, TVs, dryers and refrigerators are tested as entire, complete units. There is little variation from one
unit to another as it comes off the assembly line. Computers are built from a Tower of Babel of separate components
Actually, VCRs, TVs, et al. use a LOT of different components. A TV will probably use amplifiers, tuners, a tube, microprocessor, power supply regulator, etc. each from a different company, much like a computer. And, the vast majority of computer users keep their computers as delivered from the factory and don't replace components.
From your reasoning, all these products you described are Tower of Babels. I think computers could be much better. Have you met some of the guys that call themselves test "engineers"? Some of them are professional but way too many only know how to run Excel scripts someone else wrote for them.
which is distinct from the idea that government should freely interfere with private finances and personal issues
of morality on a regular basis (Democratic view)
Maybe it isn't far to say the Democratic view is to freely interfere with private finances but the interference on personal issues of morality is a very Republican viewpoint. Republicans want to take away a woman's right to control her own body (it's in the party platform) and Republicans don't want to let homosexuals enjoy the same privleges as heterosexuals (marriage, inheritance, etc.). Democrats, as a party, have the opposite views.
Nothing you said makes it difficults to view politics from left to right. It goes: (from left to right)
GREEN -> DEMOCRAT -> REPUBLICAN/REFORM
Libertarians sometimes act like democrats and sometimes like republicans. They can be independent thinkers. Independent thinkers I almost NEVER agree with... but that's another story.
Crusoe chip which uses software to perform many functions previously done by hardware,
enabling lighter PC notebooks with much longer battery life.
Anyone remember microcode? You could put your CPU control unit into a set of microinstructions in ROM that would tell your ALU what operations to take and you wouldn't have to design a complex controller. The above sounds similar. Is that, essentially, what Crusoe does? I know it is a lot more complex than the Mircocode of the 1970s and 1980s but one of the coolest aspects of Microcode is that you could emulate other instructions and so it made it easier to make a CPU compatible with earlier units.
It seems to me like Crusoe is a very advanced implementation of microcode, but purely in software. Correct me if I'm wrong but isn't one of the primary features of Crusoe that it emulates the Instruction Set of different processors, such as x86, in Software?
I don't think anyone on the face of the Earth thinks Moore's Law is a law in the same sense Snell's Law or Boyle's Law are laws. It was just a rather offhand comment that Gordon Moore made in the late 1970s at a VLSI conference in Caltech. It is very interesting that engineers have continuously innovated to keep Moore's Law going, but of course it will eventually stop. In fact, people were predicting the end of Moore's Law at 1 Micron, but the miracle of optical lithography has us down to 0.15 micron!
"Moore's Law" has always been considered more of a goal that a "Law".
For computer chips this means that as the physical features on the chip get smaller and closer together, the electrons will be able to tunnel from one wire to another. This is called a tunneling current. As features approach 100nm it becomes fairly noticeable, and you have to start taking it into account.
I've never heard of electrons tunneling between wires. This would be a severe, perhaps fatal, form of crosstalk, and even in a 0.1um technology, the wires aren't necessarily anywhere near that close together. What you do see, however, is something called induced gate current where a MOSFET with supposedly infinite input impedence exhibts a bias current into its gate. This is because the silicon-dioxide layer between the gate and the channel is so small electrons in the channel can tunnel through the gate oxide and escape out the gate lead. This tends to make the MOSFET look a little like a Bipolar Junction Transistors, which people have been dealing with forever. The main effect of this induced gate current is increased power dissipation.
What is interesting is that a similar induced gate current can occur when operating a MOSFET at very high frequencies. The problem here is that when the frequency gets too high the capacitance between the gate and the channel tends to short out and provide a conducting path through the gate terminal. This is observed (and taken into account) in CMOS wireless/RF circuits.
Nah, at some point we'll actually run out of things which require such processing power.
I have to dissagree strongly here. Applications expand, like a gas, to fit the available capability of any given technology. Could people using "powerful" $100,000 minicomputers in the 1970's ever dream how much computer power we have today or what we would use it for, or how cheap it would be? Streaming MP3s and video would sound like Star Trek to someone in the early 1980's!
Besides, even if the application doesn't change, new processing capability can be used for many things, such as automatic calibration of analog circuits (hard problem) and massively reconfigurable systems.
They'll always be things we can do to make stuff better, faster, or cheaper.
While photolithography certainly is one of the potential limits to Moore's law, it is not the only one, nor is it the most difficult. For years we have had electron-beam lithography but it is expensive, and that is why we have pushed optical lithography to such dizzying heights. There is no technical reason not to use E-beam lithography, but there are economic reasons.
But consider:
1. interconnect: as feature sizes diminish, the physical height of metal lines becomes greater than their width, making them look like skyscrapers, and the IC isn't so planar anymore. The problem then becomes the physical strength of the conductor, as it easily breaks as it is forced to bend over the surface of the chip. Copper interconnect is one partial solution to this problem, but it is not a magic bullet and things are getting worse all the time.
2. leakage: as transistors shrink, their gate oxide also scales. Therefore, for a given supply voltage, the electric field in the transistor increases until the gate blows out. So, then power supply voltages are scaled. Unfortunately, this tends to slow down the transistor unless the threshold voltage is also reduced, but then we have increased leakage current. This is quite a trade off, as increased leakage current not only increases the power dissipation (more on this next) but it also makes it more difficult to design RAM and mixed-signal/analog blocks.
3. Power Dissipation: Even though the supply voltage is decreased, and power dissipation of a single transistor decreases as the square of the supply voltage, overall power will increase for two reasons. First, there are many more transistors on the chip switching ever faster, and second, the reduced threshold voltages mean there will be significant static power drain even in CMOS logic. 1 nA of leakage/transistor in a 1 Volt, 1 Billion Transistor microprocessor of the future would burn a full Watt even without switching! This is a very serious problem not only for portable applications because it is difficult to package such a power hungry chip cheaply and efficiently.
While this is an interesting development to optical lithography, I don't think it will have much impact on Moore's law. In fact, I'm much more worried about the power issue and The Interconnect Problem.
My, my is this a strange "news" story. Someone looking at Slashdot for the first time would think that to Nerds, the stuff that matters is constantly whining about Microsoft and how evil they are. We've discussed things just like this over and over again (remember hotmail and BSD)? and got nowhere and said next to nothing.
I'm tired of little digs at Microsoft masquarading as news around here. I use Unix and I don't like Microsoft very much. But I don't want to whine like a spoiled puppy about it either. Can we stick to the real stuff the matters, like new products, on-line rights and privacy, and science?
fix nerve damage (I'd love to see Christopher Reeve and many others walk again someday)
I don't see this discovery putting us any closer to fixing nerve damage. The problem in fixing nerve damage isn't the lack of a suitable conductor, the problem is the unbelievable interconnect complexity and the vast number of wildly different signals. We can already use 2 digital signal processors and copper wire to jump a break in the spine. The problem is what signals need to be sent where and how quickly. So, it is an issue of DSPs that are too slow as well as a lack of understanding of the central nervous system. DNA trickery won't help us there.
Imagine, a circuit that can actually change it's physical
wiring to handle new conditions and/or optimize itself...
Actually, this isn't very far off from an adaptive Field-Programmable Gate Array (FPGA). While it doesn't physically change its wiring, it does the equivilent and reconfigures it's connections and balance of processing power to improve it's own performance. There is a whole class of adaptive filters that select tap weights to improve their performance when in contact with their environment. They are quite useful devices, and the current Internet would be impossible without them.
I think reconfigurable systems are more useful that systems that could "change it's physical wiring" because such systems would 1) Give nasty glitches when they are in the act of rewiring, and 2) Be VERY hard to analyze and test.
I wonder how we can emulate something we don't understand? Maybe we can simulate how we think a black hole works using OpenGL but we certainly can't emulate one.
Why are people interested on running Linux on SPARC rather than x86? From a price/performance point of view I think x86 machines are a much better value; unless you are running Solaris. Linux still isn't quite where Solaris is when it comes to stability and scalability so it seems the main reason to buy a SPARC computer from Sun is to run Solaris on it.
Besides, a lot of the software people buy Sun Workstations to run aren't ported to Linux yet (such as HSPICE and Cadence).
Is running Linux on a Sun really that much better than running it on a PIII?
Slashdot Mirror
Most good engineering school have a lot of support for grad students, so jumping to industry will just slow you down. With my Research Assistantships and Teaching Assistantships, I haven't paid a dime to go to grad school and I've been here four years. The student salaries are enough for rent, too, so I haven't had to get any loans. I am a traditional engineer, by the way. I'm getting a Ph.D. in Electrical Engineering.
A big problem with industry paying for your master's degree is that they usually don't give you a raise after you finish big enough to compensate what you could have gotten with a Master's in the first place. And you can't just quit, because then you have to pay back some of the fees they paid you!
Probably someone with a Ph.D. would be overqualified for the jobs for which you would hire. On the other hand, if you were hiring people to design adaptively calibrated analog-to-digital converters, for instance, I think you would be better off hiring a Ph.D. than a B.S. with 4 years of circuit design experience. This is because getting a Ph.D. forces you to truly know what you are doing, and when you are working at a very high level, knowing what you are doing is important. In advanced circuit design, for the most part the B.S. and most of the M.S. engineers work for the Ph.D. designers and the best of the M.S. engineers.
Here's what I mean. When I was an undergrad, I could study for tests and get A's without fully grasping what is going on. (How many people who took Calculus REALLY understand Calculus? Maybe one in ten) Likewise, when I was working, I could get by with a cursory knowledge of the material because I had a small part of the project and I was doing rather routine stuff. Now that I'm a Ph.D. student, the success of the project depends entirely on one person, ME. I need to know every last detail because there is no one else who will do it. This means I need to deeply understand the fundamentals, be able to absorb new information quickly, and become truly independent. This is difficult to achieve when you are working on a project and not running it yourself, because when you don't have to deal with the big picture it is easy to become nearsided. Most B.S. engineers I know are fairly competent, but they only know what they can look up. They are fine for routine designs, but they tend to lack the deep knowledge needed to extend the state of the art.
This is not to say every Ph.D. engineer is a better designer than every B.S. engineer, because that is not true. But, I submit that every B.S. engineer would be better if he or she went through the trial by fire of a Ph.D.
If you're old enough to remember 1984, you may remember that demand for the Colecovision (like PS2 today) was VERY strong in 1983/early 1984 and we were all sitting on our hands waiting for the supersystem from Atari, the 7800 (like Xbox today). A huge older system, the Atari 2600 was still selling well (like PS1 today). The market quite suddenly crashed under its own weight and the 7800 was scrapped, only to re-emerge in 1986 as a half-assed American answer to NES. I remember bargin bins of Atari 2600 games for like $9.99 when they were selling for more like $40 and everything was rosey a couple of months before.
1984 is much more similar to today than you think. The big players were raking in mad cash and awesome systems loomed on the horizon, just like today. Your likely scenario may come to pass, but I think a crash may be equally likely.
In 1984 my brother and I were both turned off by the crap being sold for consoles and turned to the commodore 64, which had a very healthy gaming market, and never skipped a beat (plus, you could actually PROGRAM it. What a beautiful assembly language it had!). People shouldn't talk about the video game crash of 1984 but rather about the console crash of 1984, because all my friends at the time were playing tons of games on their commodores and apple IIs and we were all lusting after the AMIGA!
Video games didn't crash in 1984; consoles did. I think the same statement may be applicable in 2002, as the author of the article indicates.
First, the converter is clocked at 12.8 GHz, which is very different from running at 12 GS/s. It is definately an oversampled converter, because they mention a digital decimation filter, but they don't give you the oversampling ratio. By oversampling, I mean the input is sampled higher than the Nyquist rate and noise shaped. Then, it is digitally filtered back down to Nyquist with significantly less quantization noise than you would have had without oversampling. A common oversampling ratio is 128 which would put the converter at 90 MHz signal bandwidth, which is quite high for an oversampled converter. I'm totally guessing on the oversampling ratio, though.
As for the number of bits it can do per sampling, that is largely irrelevent in a communications context. What matters is the SNR. You can figure out the number of bits of linearity with the following equation: SNR = (6*B + 2) dB where B=the effective number of bits. In other words, if you have an SNR of 60 dB, you have just about 10 bits of accuracy. The reason accuracy is specified as an SNR rather than a number of bits is if the converter is nonlinear (as it always is in practice) then even if you have more bits of resolution, the additional bits are inaccurate and should be ignored. For example, if you have a so-called "12 bit" converter with an SNR of 63 dB, then although you have 4096 possible output codes, the uncertainty between them is enough that the lower 2 LSBs are garbage, and, although you have 12 bits of resolution, you only have 10 bits of accuracy. This is a common way for manufacturers to lie on data sheets. Keep in mind, though, there are situations where resolution is more important than accuracy (such as digital imaging) and you'd rather have a 12-bit / 10 bit accurate converter than a 10-bit / 10 bit accurate converter.
Hope that cleared things up. If you have more questions, email me or reply to this post. I just love talking about data-converters!
Applying concepts like that to the brain is useless. The brain is unbelieveably parallel, and runs asynchronously, so "how fast can we think?" really doesn't mean anything. I know you were just thinking of the sci-fi aspects of this and all, but this is one nut that just ain't gonna be cracked.
The market the article talks about is the analog to digital converter market, not the desktop market.
True enough. I stand by my statements, however. For one thing, the article discusses an A/D fabricated in the technology. The die size was 1 cm^2. That is truly enormous and very would be extremely expensive as a product. Even if they can bring down the lithography, it is still very expensive. Second, while they say it runs at 12.8 GHz, because of the decimation filter it is obviously an oversampled converter but they don't give the oversampling ratio, so we have no idea of the actual conversion rate. I drew parallels between III-V materials (which have been around since the 1960s and must have achieved some kind of maturity) with superconducting electronics (which have been around since the late 1970s) and I think they still stand.
My belief is that this is a laboratory curiosity with little commerical potential. I'm sure the military is very interested in using it with radar, however.
By the way, the IEEE is well known for pumping up "cutting edge" technologies that never reach their potential. Remember "fuzzy logic"?
How many times has this been said on /.? Moore's law is about the density of DRAM, and by implication, the density of other CMOS circuits. It says nothing about clock rate. It is true that smaller transistors are faster, but there are other problems with clocks that smaller transistors make more difficult, most notably clock skew. There is already logic out there that can go faster than 20 GHz, but it is LSI GaAs logic (flip-flops / gates / adders).
Moore's law has nothing to do with superconductors, at all. We may never see mass-market superconducting logic. It will be just too expensive, and it could be impossible to integrate well enough for computers.
There are two huge problems with superconducting logic that don't seem solvable in the near future. They are:
1. Cost : These things are enormously expensive to manufacture and operate, and it is the economy of scale of CMOS techology which has enabled, more than anything, the current computing revolution. Do you have any idea how expensive coolant and the dewar to use it in are to get something to 5K? Even the so-called "high-temperature" superconductors have to be pretty damn cold to function; they just don't need to go so close to absolute-zero.
2. Integration This is probably the killer. It will be extremely difficult to integrate many devices together. Even if myriad technical difficulties are overcome, the solution is not likely to be inexpensive, as CMOS technology is. For III-V semiconductors (which use much less exotic materials than superconductors), high defect rate, problems with lattic matching of the materials, and the lack of a high quality native oxide (like SiO2 in silicon) have made it impossible to achieve integration levels anywhere close to that achieved in silicon. Even GaAs, the most well-understood III-V semiconductor, can't be integrated to more than a few thousand devices. That's why we don't have 20 GHz GaAs microprosessors. And superconductors are even HARDER to deal with.
In summary, even if researchers are able to overcome almost insurrmountable odds to find away to reliably integrate meaningful numbers of these devices on a single die, I think it is very unlikely they will be able to do it cheaply, which is just as important as being able to do it at all. Otherwise, this technology will be of interest only to the military.
By the way, I know III-V semiconductors have a lot of very important uses, especially in optics and RF. It is a fact, however, that III-V logic is mainly of interest to the military and the space industry.
The seasons don't do anything to promote life. By far the greatest concentrations of life in the tropics; both on land and in the sea, and they don't have discernable seasons. In fact, seasons are a hurdle to life: places with the largest seasonal differences (like the arctic) have the least biodiversity.
In large engineering projects, such as search engines, the amount of work to be done is something more than linearly related to the size of the project. Back when google was two guys, they had defined the key algorithm and it was really cool. But.... really cool is a long, long way from a business. For example, how do you index over a billion pages and still keep the search fast? How do you distribute the database and the searches over linux machines to reduce cost at little (or no) expense in reliability? How do you keep crafty webmasters from tricking your algorithms?
Do you really think google is the same as it was, just bigger? Give me a break.
Just why do the companies, even the great ones, think that their headcount MUST grow?
They don't. Exar is the world leader in analog interfaces for digital imagers and they've been around since the 1960s. They have around 200 employees.
SOI is significantly different from SOS, for one thing the insulator is not sapphire, and the advantages do look like they will be able to compensate for the higher cost. Chief among them are freedom from body effect and latch-up and ultra low-voltage operation. People have been interested for years in SOI technologies, including SOS, but have in the past only used them for radiation-hard military/space type things because of the cost. Today, with much lower defect densities in SOI wafers, the cost is decreasing dramatically.
Actually, VCRs, TVs, et al. use a LOT of different components. A TV will probably use amplifiers, tuners, a tube, microprocessor, power supply regulator, etc. each from a different company, much like a computer. And, the vast majority of computer users keep their computers as delivered from the factory and don't replace components.
From your reasoning, all these products you described are Tower of Babels. I think computers could be much better. Have you met some of the guys that call themselves test "engineers"? Some of them are professional but way too many only know how to run Excel scripts someone else wrote for them.
Maybe it isn't far to say the Democratic view is to freely interfere with private finances but the interference on personal issues of morality is a very Republican viewpoint. Republicans want to take away a woman's right to control her own body (it's in the party platform) and Republicans don't want to let homosexuals enjoy the same privleges as heterosexuals (marriage, inheritance, etc.). Democrats, as a party, have the opposite views.
Nothing you said makes it difficults to view politics from left to right. It goes: (from left to right)
GREEN -> DEMOCRAT -> REPUBLICAN/REFORM
Libertarians sometimes act like democrats and sometimes like republicans. They can be independent thinkers. Independent thinkers I almost NEVER agree with... but that's another story.
Anyone remember microcode? You could put your CPU control unit into a set of microinstructions in ROM that would tell your ALU what operations to take and you wouldn't have to design a complex controller. The above sounds similar. Is that, essentially, what Crusoe does? I know it is a lot more complex than the Mircocode of the 1970s and 1980s but one of the coolest aspects of Microcode is that you could emulate other instructions and so it made it easier to make a CPU compatible with earlier units.
It seems to me like Crusoe is a very advanced implementation of microcode, but purely in software. Correct me if I'm wrong but isn't one of the primary features of Crusoe that it emulates the Instruction Set of different processors, such as x86, in Software?
How is that 5 years ahead?
"Moore's Law" has always been considered more of a goal that a "Law".
Well, these guys claim they can switch a single hydrogen atom between two silicon atoms.
Check out the press release
And the slashdot discussion about it
I've never heard of electrons tunneling between wires. This would be a severe, perhaps fatal, form of crosstalk, and even in a 0.1um technology, the wires aren't necessarily anywhere near that close together. What you do see, however, is something called induced gate current where a MOSFET with supposedly infinite input impedence exhibts a bias current into its gate. This is because the silicon-dioxide layer between the gate and the channel is so small electrons in the channel can tunnel through the gate oxide and escape out the gate lead. This tends to make the MOSFET look a little like a Bipolar Junction Transistors, which people have been dealing with forever. The main effect of this induced gate current is increased power dissipation.
What is interesting is that a similar induced gate current can occur when operating a MOSFET at very high frequencies. The problem here is that when the frequency gets too high the capacitance between the gate and the channel tends to short out and provide a conducting path through the gate terminal. This is observed (and taken into account) in CMOS wireless/RF circuits.
I have to dissagree strongly here. Applications expand, like a gas, to fit the available capability of any given technology. Could people using "powerful" $100,000 minicomputers in the 1970's ever dream how much computer power we have today or what we would use it for, or how cheap it would be? Streaming MP3s and video would sound like Star Trek to someone in the early 1980's!
Besides, even if the application doesn't change, new processing capability can be used for many things, such as automatic calibration of analog circuits (hard problem) and massively reconfigurable systems.
They'll always be things we can do to make stuff better, faster, or cheaper.
But consider:
1. interconnect: as feature sizes diminish, the physical height of metal lines becomes greater than their width, making them look like skyscrapers, and the IC isn't so planar anymore. The problem then becomes the physical strength of the conductor, as it easily breaks as it is forced to bend over the surface of the chip. Copper interconnect is one partial solution to this problem, but it is not a magic bullet and things are getting worse all the time.
2. leakage: as transistors shrink, their gate oxide also scales. Therefore, for a given supply voltage, the electric field in the transistor increases until the gate blows out. So, then power supply voltages are scaled. Unfortunately, this tends to slow down the transistor unless the threshold voltage is also reduced, but then we have increased leakage current. This is quite a trade off, as increased leakage current not only increases the power dissipation (more on this next) but it also makes it more difficult to design RAM and mixed-signal/analog blocks.
3. Power Dissipation: Even though the supply voltage is decreased, and power dissipation of a single transistor decreases as the square of the supply voltage, overall power will increase for two reasons. First, there are many more transistors on the chip switching ever faster, and second, the reduced threshold voltages mean there will be significant static power drain even in CMOS logic. 1 nA of leakage/transistor in a 1 Volt, 1 Billion Transistor microprocessor of the future would burn a full Watt even without switching! This is a very serious problem not only for portable applications because it is difficult to package such a power hungry chip cheaply and efficiently.
While this is an interesting development to optical lithography, I don't think it will have much impact on Moore's law. In fact, I'm much more worried about the power issue and The Interconnect Problem.
I'm tired of little digs at Microsoft masquarading as news around here. I use Unix and I don't like Microsoft very much. But I don't want to whine like a spoiled puppy about it either. Can we stick to the real stuff the matters, like new products, on-line rights and privacy, and science?
fix nerve damage (I'd love to see Christopher Reeve and many others walk again someday)
I don't see this discovery putting us any closer to fixing nerve damage. The problem in fixing nerve damage isn't the lack of a suitable conductor, the problem is the unbelievable interconnect complexity and the vast number of wildly different signals. We can already use 2 digital signal processors and copper wire to jump a break in the spine. The problem is what signals need to be sent where and how quickly. So, it is an issue of DSPs that are too slow as well as a lack of understanding of the central nervous system. DNA trickery won't help us there.
Actually, this isn't very far off from an adaptive Field-Programmable Gate Array (FPGA). While it doesn't physically change its wiring, it does the equivilent and reconfigures it's connections and balance of processing power to improve it's own performance. There is a whole class of adaptive filters that select tap weights to improve their performance when in contact with their environment. They are quite useful devices, and the current Internet would be impossible without them.
I think reconfigurable systems are more useful that systems that could "change it's physical wiring" because such systems would 1) Give nasty glitches when they are in the act of rewiring, and 2) Be VERY hard to analyze and test.
I wonder how we can emulate something we don't understand? Maybe we can simulate how we think a black hole works using OpenGL but we certainly can't emulate one.
Besides, a lot of the software people buy Sun Workstations to run aren't ported to Linux yet (such as HSPICE and Cadence).
Is running Linux on a Sun really that much better than running it on a PIII?