November 22, 2010 at 19:30Processor Difficulties
Recently, in the Moscow Polytechnic Museum, the computer stand was seriously updated - Intel placed its stand there, which was called " From the sand to the processor ." From now on, this stand will become an integral part of school trips, but even adults I advise you not to postpone visiting the institution for more than five years - by 2016 Intel plans to seriously “upgrade” the museum so that it can enter the top ten science museums in the world!
A three-part lecture cycle of the same name was timed to coincide with this event. Two lectures have already passed - you can find their contents under the cut. Well, if you are interested in all this, then you will still have time to attend the third lecture, information about which is at the end of the post.
I’m not ashamed to admit that most of this text is indeed a summary of the first lecture, which was held by Nikolai Suetin , director of external projects in the field of research and development of Intel in Russia. For the most part, it was about modern semiconductor technologies and the problems that they face.
I propose to start reading interesting, and we will start with the very basics.
Technically, a modern microprocessor is made in the form of one ultra-large integrated circuit, consisting of several billion elements - this is one of the most complex structures created by man. The key elements of any microprocessor are discrete switches - transistors. By blocking and passing electric current (on-off), they enable the logic circuits of the computer to work in two states, that is, in a binary system. The sizes of transistors are measured in nanometers. One nanometer (nm) is one billionth (10–9) part of a meter.
The main part of the work when creating processors is not done at all by people, but by robotic mechanisms - it is they who drag silicon wafers back and forth. The production cycle of each plate can reach up to 2-3 months.
In more detail (and clearly) I will tell about the technology of production of processors, but in the meantime, quite briefly.
The plates are really made of sand - in terms of prevalence in the earth's crust, silicon takes the second place after oxygen. By chemical reactions, silicon oxide (SiO 2) thoroughly cleaned, making it "dirty" clean. For microelectronics, monocrystalline silicon is needed - it is obtained from a melt. It all starts with a small crystal (which is lowered into the melt) - later it turns into a special single-crystal “bull” growing from a person. Then, the main defects are removed and special bulbs (with diamond powder) are cut into disks - each disk is carefully processed to an absolutely smooth and smooth (at atomic level) surface. The thickness of each plate is about 1 mm - solely so that it does not break or bend, that is, so that it can be comfortably worked with.
The diameter of each plate is exactly 300mm - a little later hundreds, or even thousands of processors will “grow” on this area. By the way, Intel, Samsung, Toshiba and TSMC have already announced that they are developing equipment capable of working with 450mm plates (more processors will fit on a larger area, which means the price of each will be lower) - the transition to them is already planned 2012 year.
Here is a cross-sectional image of the processor:
There is a protective metal cover on top, which, in addition to the protective function, also acts as a heat distributor - it is we who abundantly apply thermal grease when we install the cooler. Under the heat distributor is the very piece of silicon that performs all user tasks. Even lower is a special substrate, which is needed for wiring contacts (and increasing the area of the “legs”) so that the processor can be installed in the socket of the motherboard.
The chip itself consists of silicon, on which there are up to 9 layers of metallization (made of copper) - just so many levels are needed so that, according to a certain law, it is possible to connect transistors located on the silicon surface (since it is simply impossible to do all this at the same level). In fact, these layers act as connecting wires, only on a much smaller scale; so that the "wires" do not short-circuit each other, they are separated by a layer of oxide (with a low dielectric constant).
As I wrote above, the unit cell of the processor is a field effect transistor. The first semiconductor products were from Germany and the first transistors were made from it. But as soon as field-effect transistors began to be made (under the gate of which there is a special insulating layer - a thin dielectric film that controls the "on" and "off" of the transistor), germanium immediately "died out", giving way to silicon. Over the past 40 years, silicon dioxide (SiO 2 ) has been used as the main material for the gate dielectric , which was due to its manufacturability and the possibility of systematically improving the characteristics of transistors as their size decreases.
The scaling rule is simple - reducing the size of the transistor, the thickness of the dielectric should decrease proportionally. So, for example, in chips with a technological process of 65 nm, the thickness of the gate dielectric layer of SiO 2amounted to about 1.2 nm, which is equivalent to five atomic layers. In fact, this is the physical limit for this material, since as a result of a further decrease in the transistor itself (and hence a decrease in the silicon dioxide layer), the leakage current through the gate dielectric increases significantly, which leads to significant current losses and excessive heat generation. In this case, the silicon dioxide layer ceases to be an obstacle to quantum tunneling of electrons, because of which the possibility of guaranteed control of the state of the transistor disappears. Accordingly, even with the perfect manufacture of all transistors (the number of which in the modern processor reaches several billion), the malfunction of at least one of them means the malfunction of the entire processor logic,
The process of miniaturization of transistors did not go against the laws of physics, but computer progress, as we see, did not stop. This means that the problem with the dielectric was somehow solved. And they really decided - when switching to 45nm, Intel began to use a new material, the so-called high-k dielectric, which replaced a hopelessly thin layer of silicon dioxide. A layer based on a rare earth oxide of hafnium with a high (20 vs. 4 for SiO 2) the dielectric constant k (high-k) has become thicker, but this has reduced the leakage current by more than ten times, while maintaining the ability to correctly and stably control the operation of the transistor. The new dielectric turned out to be poorly compatible with the gate made of polysilicon, but this did not become an obstacle - the gate in the new transistors was made of metal to increase the speed of operation.
Thus, Intel became the first company in the world to switch to mass production of microprocessors using hafnium. Moreover, the corporation still belongs to the palm - no one can still reproduce this technology, because a dielectric film is created by atomic deposition, and the material is deposited in successive layers with a thickness of just one atom.
Interestingly, after reading these paragraphs, you got the idea of how billions of transistors design, build and fit in such a small area? And how does it all work in the end and, at the same time, costs quite reasonable money? I was very thoughtful, although before I thought all this was obvious and I even had the conscience to think, “ Hey, why so expensive? For one processor only! ":)
In 1965, one of the founders of Intel Corporation, Gordon Moore, recorded an empirical observation, which later became the famous law of his name. Presenting a graph of the increase in the performance of memory chips, he discovered a curious pattern: new models of chips were developed after equal intervals of time - about 18-24 months - after the appearance of their predecessors, while the capacity of the chips increased approximately twice each time.
Later, Gordon Moore predicted the pattern, suggesting that the number of transistors in microprocessors would double every two years - in fact, constantly creating innovative technologies, Intel has been implementing Moore's law for over 40 years.
The number of transistors continues to grow, although the size of the "output" processor remains relatively unchanged. Again, there is no secret - this becomes clear if you look at the following relationship.
As you can see, once every two years the topological dimensions are reduced by 0.7 times. As a result of reducing the size of transistors, their switching speed is higher, lower price and less power consumption.
Currently, Intel is releasing processors using 32nm technology. Key technical differences from 45nm technology:
- 9 metallization levels
are used - a new generation of high-k dielectric is used (also hafnium oxide, but with special additives - the resulting layer is equivalent to 0.9 nm silicon oxide)
The creation of a new technological process for creating a metal gate led to a 22% increase in the performance of all transistors (compared to 45nm), as well as to the highest density of elements, which required the highest current density.
Intel manufactures processors in three countries - the United States, Israel and Ireland. At the moment, the company has 4 factories for mass production of processors using 32nm technology. These are: D1D and D1C in Oregon, Fab 32 in Arizona and Fab 11X in New Mexico. There are many interesting things in the design of these plants and in their work, but I will talk about this next time.
The cost of such a plant is about $ 5 billion, and if you make several plants at once, then the amount of investment can be safely multiplied. If we take into account that the technology change occurs once every two years, it turns out that the plant has exactly 4 years to “recapture” the $ 5bn invested in it and make a profit. From which the obvious conclusion suggests itself - the economy very much dictates the development of technological progress ... but, despite all these huge figures, the cost of manufacturing a single transistor continues to fall - now it is less than one billionth of a dollar.
It is not necessary to think that with the transition of several factories to 32nm, everything will suddenly begin to be carried out according to this technical process - the same chipsets and other peripheral circuits just do not need it - in most cases they use 45nm. The 22nm boundary is planned to be fully taken as early as next year, and by 2013 it is likely to be 16nm. At least this year, a test plate (22nm) has already been made, on which the performance of all the elements necessary for the processor to work has been demonstrated.
* UPD by nE0 * The need to reduce the thickness of the gate dielectric is dictated by the simple formula of a flat capacitor:
The gate area of the transistor is reduced, and for the transistor to work, the capacity of the gate dielectric must be preserved.
Therefore, it was necessary to reduce its thickness, and when it became impossible to find a material with a higher dielectric constant.
When will the silicon era end? The exact date is still unknown, but it is definitely not far off. In the 22nm technology, it will definitely “fight”, most likely it will remain in the 16nm ... but then the fun part will begin. The periodic table, in principle, is large enough and there is plenty to choose from) But most likely, everything will rest not only in chemistry. It will be possible to increase the efficiency of the processor either by reducing the topological dimensions (they are doing it now), or by using other compounds with higher carrier mobility - perhaps gallium arsenide, possibly “sensational” and promising graphene (by the way, it has hundreds of times the mobility higher than silicon). But there are problems. Now the technology is designed to process plates with a diameter of 300 mm - the amount of gallium arsenide needed for such a plate is simply not in nature,
Most likely, the next step will be the deposition of monocrystalline gallium arsenide on silicon, and then graphene. And, perhaps, the development of microelectronics will go not only along the path of improving technologies, but also along the path of developing a fundamentally new logic - this, too, cannot be ruled out. Let's bet, gentlemen? ;)
In general, now there is a struggle for technology and high mobility. But one thing is clear - there is no reason to stop progress.
The process of manufacturing processors consists of two large "parts". For the first, you need to have the manufacturing technology itself, and for the second you need an understanding of WHAT to make and how - the architecture (how the transistors are connected). If both new architecture and new technology are made, then in case of failure it will be difficult to find the “culprits” - some will say that the “architects” are to blame, others that the technologists. In general, following such a strategy is very short-sighted.
At Intel, the introduction of new technology and architecture is spread out over time - technology is introduced in one year (and already developed architecture is produced using the new technology - if something goes wrong, then the technologists will be to blame); and when the new technology is worked out, the architects will make a new architecture for it, and if something does not work on the proven technology, then the architects will be to blame. This strategy was called "Tick-Tack."
More clearly:
With the current pace of technology development, a fantastic amount of investment in research and development is required - annually Intel invests $ 4-5bn in this business. Part of the work is going on inside the company, but a lot is going on outside it. It’s almost impossible to keep an entire laboratory in the company like Bell Labs (the forge of Nobel laureates).
As a rule, the first ideas are laid in universities - so that universities know what exactly makes sense to work on (which technologies are in demand and what will be relevant), all “semiconductor companies” were united in a consortium. After that, they provide a kind of roadmap - it talks about all the problems that the semiconductor industry will face in the next 3-5-7 years. In theory, any company has the right to literally go to the university and “take advantage” of one or another innovative development, but the developer university, as a rule, retains the rights to it - this approach is called “open innovation”. Intel was no exception and periodically listens to students' ideas - after protection, selection at the engineering level and testing in real conditions,
Here is a list of research centers around the world that Intel works with (except universities):
Increased productivity leads to more expensive factories, and this in turn leads to natural selection. So, for example, in order to recoup itself in 4 years, each Intel factory must produce at least 100 working plates per hour. There are thousands of chips on each plate ... and if you make certain calculations, it will become clear - if Intel had not 80% of the global processor market, the company simply could not have recouped the costs. The conclusion - to have both our own “design” and our own production in our time is quite unprofitable - at least you need to have a huge market. The result of natural selection can be seen below - as you can see, with its “design” and production, fewer and fewer companies keep pace with technological progress. Everyone else had to switch to fabless mode - for example, neither Apple nor NVIDIA,
In addition to Intel, only two companies are potentially ready for 22nm technology worldwide - Samsung and TSMC, which invested more than $ 1 billion last year in their factories. Moreover, TSMC does not have its own design unit (only foundry) - in fact, it is just a high-tech forge that accepts orders from other companies and often does not even know what it forges.
As you can see, natural selection passed quite quickly - in just 3 years. Two conclusions can be drawn from this. The first is that it is unlikely to become an industry leader without an factory; the second - in fact, you can succeed without your plant. By and large, a good computer, brains and the ability to “draw” will be enough - the threshold for entering the market has greatly decreased, and for this reason a lot of “startups” have appeared. Someone comes up with a certain scheme for which there is or a certain market is artificially created - novice manufacturers are rising ... PROFIT! But the threshold to the foundry market has risen dramatically and will only continue to grow ...
What else has changed in recent years? If you recall, then until 2004, the statement “the higher the processor frequency, the better” was quite fair. Starting from 2004-2005, the frequency of processors almost stopped growing, which is associated with access to a kind of physical restrictions. Now you can increase productivity due to multi-core - performing tasks in parallel. But to make many cores on one chip is not a big problem - it is much more difficult to make them work correctly in the load. As a result, from this moment the role of software has dramatically increased, and the importance of the profession “programmer” will only gain momentum in the near future.
In general, summing up the above :
- Moore's Law continues to apply
- The increase in the cost of developing new technologies and materials, as well as the costs of maintaining factories are
growing. - Productivity is also growing. A jump is expected when switching to 450mm plates.
As a result :
- Separation of companies into "fabless" and "foundry"
- Outsource of basic R&D
- Differentiation due to software development
Was it interesting for you to read? I hope so. At the very least, it was interesting for me to write all this and it was even more interesting to listen to it ... although I also thought at first, "what will they tell in this lecture."
Last week at the Moscow Polytechnic Museum, a second lecture was held by Intel Fellow Academician Boris Babayan . The lecture was dedicated to the history of Russian microprocessor developments - unfortunately, I just won’t have time to write a similar article on the second lecture. Therefore, for those who could not come, I can offer only this audio recording of the lecture.
This Wednesday (November 24, 2010, in the same place) will be the final lecture from the cycle "From sand to processor" - it will be conducted by Oleg Semenov, head of the Russian Intel Labs. Come, admission is free.
See you soon!
A three-part lecture cycle of the same name was timed to coincide with this event. Two lectures have already passed - you can find their contents under the cut. Well, if you are interested in all this, then you will still have time to attend the third lecture, information about which is at the end of the post.
I’m not ashamed to admit that most of this text is indeed a summary of the first lecture, which was held by Nikolai Suetin , director of external projects in the field of research and development of Intel in Russia. For the most part, it was about modern semiconductor technologies and the problems that they face.
I propose to start reading interesting, and we will start with the very basics.
CPU
Technically, a modern microprocessor is made in the form of one ultra-large integrated circuit, consisting of several billion elements - this is one of the most complex structures created by man. The key elements of any microprocessor are discrete switches - transistors. By blocking and passing electric current (on-off), they enable the logic circuits of the computer to work in two states, that is, in a binary system. The sizes of transistors are measured in nanometers. One nanometer (nm) is one billionth (10–9) part of a meter.| More than 2000 transistor gates made using 45-nm manufacturing technology can be placed on a section of a single human hair. If we talk about nanotechnology in more detail, then in 2008, more than 227 billion dollars of issued semiconductor chips accounted for almost the entire "nano" market, while magnetic disks, optoelectronics, etc. still makes up no more than 15%. |
In more detail (and clearly) I will tell about the technology of production of processors, but in the meantime, quite briefly.
The plates are really made of sand - in terms of prevalence in the earth's crust, silicon takes the second place after oxygen. By chemical reactions, silicon oxide (SiO 2) thoroughly cleaned, making it "dirty" clean. For microelectronics, monocrystalline silicon is needed - it is obtained from a melt. It all starts with a small crystal (which is lowered into the melt) - later it turns into a special single-crystal “bull” growing from a person. Then, the main defects are removed and special bulbs (with diamond powder) are cut into disks - each disk is carefully processed to an absolutely smooth and smooth (at atomic level) surface. The thickness of each plate is about 1 mm - solely so that it does not break or bend, that is, so that it can be comfortably worked with.
The diameter of each plate is exactly 300mm - a little later hundreds, or even thousands of processors will “grow” on this area. By the way, Intel, Samsung, Toshiba and TSMC have already announced that they are developing equipment capable of working with 450mm plates (more processors will fit on a larger area, which means the price of each will be lower) - the transition to them is already planned 2012 year.
Here is a cross-sectional image of the processor:
There is a protective metal cover on top, which, in addition to the protective function, also acts as a heat distributor - it is we who abundantly apply thermal grease when we install the cooler. Under the heat distributor is the very piece of silicon that performs all user tasks. Even lower is a special substrate, which is needed for wiring contacts (and increasing the area of the “legs”) so that the processor can be installed in the socket of the motherboard.
The chip itself consists of silicon, on which there are up to 9 layers of metallization (made of copper) - just so many levels are needed so that, according to a certain law, it is possible to connect transistors located on the silicon surface (since it is simply impossible to do all this at the same level). In fact, these layers act as connecting wires, only on a much smaller scale; so that the "wires" do not short-circuit each other, they are separated by a layer of oxide (with a low dielectric constant).
As I wrote above, the unit cell of the processor is a field effect transistor. The first semiconductor products were from Germany and the first transistors were made from it. But as soon as field-effect transistors began to be made (under the gate of which there is a special insulating layer - a thin dielectric film that controls the "on" and "off" of the transistor), germanium immediately "died out", giving way to silicon. Over the past 40 years, silicon dioxide (SiO 2 ) has been used as the main material for the gate dielectric , which was due to its manufacturability and the possibility of systematically improving the characteristics of transistors as their size decreases.
The scaling rule is simple - reducing the size of the transistor, the thickness of the dielectric should decrease proportionally. So, for example, in chips with a technological process of 65 nm, the thickness of the gate dielectric layer of SiO 2amounted to about 1.2 nm, which is equivalent to five atomic layers. In fact, this is the physical limit for this material, since as a result of a further decrease in the transistor itself (and hence a decrease in the silicon dioxide layer), the leakage current through the gate dielectric increases significantly, which leads to significant current losses and excessive heat generation. In this case, the silicon dioxide layer ceases to be an obstacle to quantum tunneling of electrons, because of which the possibility of guaranteed control of the state of the transistor disappears. Accordingly, even with the perfect manufacture of all transistors (the number of which in the modern processor reaches several billion), the malfunction of at least one of them means the malfunction of the entire processor logic,
The process of miniaturization of transistors did not go against the laws of physics, but computer progress, as we see, did not stop. This means that the problem with the dielectric was somehow solved. And they really decided - when switching to 45nm, Intel began to use a new material, the so-called high-k dielectric, which replaced a hopelessly thin layer of silicon dioxide. A layer based on a rare earth oxide of hafnium with a high (20 vs. 4 for SiO 2) the dielectric constant k (high-k) has become thicker, but this has reduced the leakage current by more than ten times, while maintaining the ability to correctly and stably control the operation of the transistor. The new dielectric turned out to be poorly compatible with the gate made of polysilicon, but this did not become an obstacle - the gate in the new transistors was made of metal to increase the speed of operation.
Thus, Intel became the first company in the world to switch to mass production of microprocessors using hafnium. Moreover, the corporation still belongs to the palm - no one can still reproduce this technology, because a dielectric film is created by atomic deposition, and the material is deposited in successive layers with a thickness of just one atom.
| Hafnium (lat. Hafnium, Hf) - a heavy refractory silver-white metal, 72 element of the periodic system, opened in 1923. In the world, an average of about 70 tons of hafnium is mined annually. Despite the fact that the metal is rare earth and relatively few are mined, there is no cause for concern. Firstly, oxide is used, and secondly, the thickness of the oxide film will only decrease over time. And thirdly - if you take one cubic centimeter of hafnium and distribute it over the surface with a layer of such a thickness that is used in chips, then an area equal to 10 football fields will be covered with a film of hafnium. Something like this :) |
In 1965, one of the founders of Intel Corporation, Gordon Moore, recorded an empirical observation, which later became the famous law of his name. Presenting a graph of the increase in the performance of memory chips, he discovered a curious pattern: new models of chips were developed after equal intervals of time - about 18-24 months - after the appearance of their predecessors, while the capacity of the chips increased approximately twice each time.
Later, Gordon Moore predicted the pattern, suggesting that the number of transistors in microprocessors would double every two years - in fact, constantly creating innovative technologies, Intel has been implementing Moore's law for over 40 years.
The number of transistors continues to grow, although the size of the "output" processor remains relatively unchanged. Again, there is no secret - this becomes clear if you look at the following relationship.
As you can see, once every two years the topological dimensions are reduced by 0.7 times. As a result of reducing the size of transistors, their switching speed is higher, lower price and less power consumption.
Currently, Intel is releasing processors using 32nm technology. Key technical differences from 45nm technology:
- 9 metallization levels
are used - a new generation of high-k dielectric is used (also hafnium oxide, but with special additives - the resulting layer is equivalent to 0.9 nm silicon oxide)
The creation of a new technological process for creating a metal gate led to a 22% increase in the performance of all transistors (compared to 45nm), as well as to the highest density of elements, which required the highest current density.
Production
Intel manufactures processors in three countries - the United States, Israel and Ireland. At the moment, the company has 4 factories for mass production of processors using 32nm technology. These are: D1D and D1C in Oregon, Fab 32 in Arizona and Fab 11X in New Mexico. There are many interesting things in the design of these plants and in their work, but I will talk about this next time.
The cost of such a plant is about $ 5 billion, and if you make several plants at once, then the amount of investment can be safely multiplied. If we take into account that the technology change occurs once every two years, it turns out that the plant has exactly 4 years to “recapture” the $ 5bn invested in it and make a profit. From which the obvious conclusion suggests itself - the economy very much dictates the development of technological progress ... but, despite all these huge figures, the cost of manufacturing a single transistor continues to fall - now it is less than one billionth of a dollar.
It is not necessary to think that with the transition of several factories to 32nm, everything will suddenly begin to be carried out according to this technical process - the same chipsets and other peripheral circuits just do not need it - in most cases they use 45nm. The 22nm boundary is planned to be fully taken as early as next year, and by 2013 it is likely to be 16nm. At least this year, a test plate (22nm) has already been made, on which the performance of all the elements necessary for the processor to work has been demonstrated.
* UPD by nE0 * The need to reduce the thickness of the gate dielectric is dictated by the simple formula of a flat capacitor:
The gate area of the transistor is reduced, and for the transistor to work, the capacity of the gate dielectric must be preserved.Therefore, it was necessary to reduce its thickness, and when it became impossible to find a material with a higher dielectric constant.
When will the silicon era end? The exact date is still unknown, but it is definitely not far off. In the 22nm technology, it will definitely “fight”, most likely it will remain in the 16nm ... but then the fun part will begin. The periodic table, in principle, is large enough and there is plenty to choose from) But most likely, everything will rest not only in chemistry. It will be possible to increase the efficiency of the processor either by reducing the topological dimensions (they are doing it now), or by using other compounds with higher carrier mobility - perhaps gallium arsenide, possibly “sensational” and promising graphene (by the way, it has hundreds of times the mobility higher than silicon). But there are problems. Now the technology is designed to process plates with a diameter of 300 mm - the amount of gallium arsenide needed for such a plate is simply not in nature,
Most likely, the next step will be the deposition of monocrystalline gallium arsenide on silicon, and then graphene. And, perhaps, the development of microelectronics will go not only along the path of improving technologies, but also along the path of developing a fundamentally new logic - this, too, cannot be ruled out. Let's bet, gentlemen? ;)
In general, now there is a struggle for technology and high mobility. But one thing is clear - there is no reason to stop progress.
Tick tac
The process of manufacturing processors consists of two large "parts". For the first, you need to have the manufacturing technology itself, and for the second you need an understanding of WHAT to make and how - the architecture (how the transistors are connected). If both new architecture and new technology are made, then in case of failure it will be difficult to find the “culprits” - some will say that the “architects” are to blame, others that the technologists. In general, following such a strategy is very short-sighted.
At Intel, the introduction of new technology and architecture is spread out over time - technology is introduced in one year (and already developed architecture is produced using the new technology - if something goes wrong, then the technologists will be to blame); and when the new technology is worked out, the architects will make a new architecture for it, and if something does not work on the proven technology, then the architects will be to blame. This strategy was called "Tick-Tack."
| Tick- tock is an extensive microprocessor development strategy announced by Intel at the Intel Developer Forum in September 2006. The development cycle is divided into two stages - tick and so. “Tick” means a miniaturization of the process and relatively small improvements to the microarchitecture. “So” means the release of processors with a new microarchitecture, but using the existing process. According to Intel's plans, each part of the cycle should take about a year. |
With the current pace of technology development, a fantastic amount of investment in research and development is required - annually Intel invests $ 4-5bn in this business. Part of the work is going on inside the company, but a lot is going on outside it. It’s almost impossible to keep an entire laboratory in the company like Bell Labs (the forge of Nobel laureates).
As a rule, the first ideas are laid in universities - so that universities know what exactly makes sense to work on (which technologies are in demand and what will be relevant), all “semiconductor companies” were united in a consortium. After that, they provide a kind of roadmap - it talks about all the problems that the semiconductor industry will face in the next 3-5-7 years. In theory, any company has the right to literally go to the university and “take advantage” of one or another innovative development, but the developer university, as a rule, retains the rights to it - this approach is called “open innovation”. Intel was no exception and periodically listens to students' ideas - after protection, selection at the engineering level and testing in real conditions,
Here is a list of research centers around the world that Intel works with (except universities):
Increased productivity leads to more expensive factories, and this in turn leads to natural selection. So, for example, in order to recoup itself in 4 years, each Intel factory must produce at least 100 working plates per hour. There are thousands of chips on each plate ... and if you make certain calculations, it will become clear - if Intel had not 80% of the global processor market, the company simply could not have recouped the costs. The conclusion - to have both our own “design” and our own production in our time is quite unprofitable - at least you need to have a huge market. The result of natural selection can be seen below - as you can see, with its “design” and production, fewer and fewer companies keep pace with technological progress. Everyone else had to switch to fabless mode - for example, neither Apple nor NVIDIA,
In addition to Intel, only two companies are potentially ready for 22nm technology worldwide - Samsung and TSMC, which invested more than $ 1 billion last year in their factories. Moreover, TSMC does not have its own design unit (only foundry) - in fact, it is just a high-tech forge that accepts orders from other companies and often does not even know what it forges.
As you can see, natural selection passed quite quickly - in just 3 years. Two conclusions can be drawn from this. The first is that it is unlikely to become an industry leader without an factory; the second - in fact, you can succeed without your plant. By and large, a good computer, brains and the ability to “draw” will be enough - the threshold for entering the market has greatly decreased, and for this reason a lot of “startups” have appeared. Someone comes up with a certain scheme for which there is or a certain market is artificially created - novice manufacturers are rising ... PROFIT! But the threshold to the foundry market has risen dramatically and will only continue to grow ...
What else has changed in recent years? If you recall, then until 2004, the statement “the higher the processor frequency, the better” was quite fair. Starting from 2004-2005, the frequency of processors almost stopped growing, which is associated with access to a kind of physical restrictions. Now you can increase productivity due to multi-core - performing tasks in parallel. But to make many cores on one chip is not a big problem - it is much more difficult to make them work correctly in the load. As a result, from this moment the role of software has dramatically increased, and the importance of the profession “programmer” will only gain momentum in the near future.
In general, summing up the above :
- Moore's Law continues to apply
- The increase in the cost of developing new technologies and materials, as well as the costs of maintaining factories are
growing. - Productivity is also growing. A jump is expected when switching to 450mm plates.
As a result :
- Separation of companies into "fabless" and "foundry"
- Outsource of basic R&D
- Differentiation due to software development
The end
Was it interesting for you to read? I hope so. At the very least, it was interesting for me to write all this and it was even more interesting to listen to it ... although I also thought at first, "what will they tell in this lecture."
Last week at the Moscow Polytechnic Museum, a second lecture was held by Intel Fellow Academician Boris Babayan . The lecture was dedicated to the history of Russian microprocessor developments - unfortunately, I just won’t have time to write a similar article on the second lecture. Therefore, for those who could not come, I can offer only this audio recording of the lecture.
[ Download / 180Mb]
This Wednesday (November 24, 2010, in the same place) will be the final lecture from the cycle "From sand to processor" - it will be conducted by Oleg Semenov, head of the Russian Intel Labs. Come, admission is free.
See you soon!