>. I don't know what Atmel did to deserve their good luck.
There was a long time when it basically was AVR vs PIC when it came to "small developer" aka hobbyist microcontrollers. The PIC may have been there first, but the AVR architecture is much more user friendly and has a following that is at least as large as that of the PIC. The reason why the AVR is being used in the arduino is probably because of it's high-level language compatible microarchitecture.
Seems to me for most uses simply increasing the refresh time interval would save tons of power, and also complexity. If you could get it to a couple of days,
Yes, increasing the refresh time is indeed a way to reduce power consumption of a DRAM. The problem is that you are dealing with billions of memory cells. The median retention time of typical cells is well within the range of seconds. But there is a tiny fraction of cells (1/10000) that lose their charge much quicker, and things may get worse at elevated temperatures etc. Those cells impose limits on the minimum refresh time.
There are ways to work around this by introducing on-the-fly error correction. But this will result in a larger device and added latency, which is obviously not desired in many applications. Nevertheless, there are dedicated low power DRAMs which use this kind of scheme to increase refresh time.
I am a semiconductor scientist, but I completely fail to understand what this news is about. The article does nowhere mention the materials used, the device behavior, the application, the purpose or anything else. A MIM device as is, is a capacitor. And that is exactly what the picture is showing. When this type of capacitor is scaled to the nanometer regime it starts to get leaky due to quantum mechanical tunneling through the dielectric. The abstract mentions 'controlled quantum mechanical tunneling'... Aha, this could be what it is about. But as long as metal electrodes are involved this will only create a nonlinear resistor. Still no idea what the exact purpose is.
Are nanoscale MIM capacitors new? No, not at all. Right now you have billions of them doing their job in your computers main memory. Depending on the vintage of your computer, these capacitors employ nanolaminates of ZrO2 and Al2O3 at a total thickness of 5 to 10 nanometers. Quantum electrical tunneling is of high relevance in these devices, since it leads to loss of stored information. So, is cheap new? A quick calculation suggests that the manufacturing cost of a single MIM device in a DRAM is approximately 10^(-10) US$.
This is not true. You need to be aware of one thing: "Memristors" were not new when they were "discovered". The memory industry knew the concept years before as RRAM. I can assure you that all other nonvolatile memory vendors are developing RRAM or are at least looking into the possibilities. Samsung has been publishing about NiO based RRAM long before it was "discovered" again, IBM has some interesting papers from the Zurich labs. Furthermore, there are several start up companies looking into 3D RRAM which may offer densities far above that of flash. Matrix Semiconductors (bought by Sandisk) and a company by a former Micron guy.
One significant issue with RRAM (and the memristor) so far is that the memory cells have to be "formed". They need an initial high voltage pulse to induce the switching behaviour. This is something that is difficult to do when you have billions of memory cells. To my knowledge no good solution to this problem has been found yet, although there is progress.
This has nothing to do with "leakage" current. As basic field effect transistor theory will teach you, there is a region below the threshold voltage where the current depends exponentially on the gate bias. Yes, exponentially instead of linearly or quadratically as in the "on" region. This means that small changes in the gate bias will allow for a huge change in current. The drawback is here that we are talking about extremely low current. In CMOS logic this equals lower operation frequency.
This idea here is to build circuits that are extremely power efficient but also extremely slow. As mentioned in the article, potential applications may include wireless sensor networks that have to operate on extremely low energy.
Most DRAM companies have operated at a net loss when taking into account the accumulated earnings of the last decade. There is incredibly fierce price competition within the industry.
Do you really feel ripped off when you buy a product that is composed of billions of transistors, has tens of billions of R&D costs behind at at a price of $1 ? (That was the price of a 1Gbit chip not long ago) I don't want to sound like an industry advocate here, but I find this pretty staggering. Is the consumer really ripped off if he has to buy a product that is priced a few months behind Moores law?
Growing from $50m to $1B in five years would mean that the market for open sources increases twentyfold in just five years. OSS has been around since the 80ies, OSS companies have been around since the nineties. So it took 15 years for the market to grow to $50m, why on earth should it increase twentyfold in just five measly years? Like many analysts, he is just making up numbers.
Aren't start-ups almost always spinoffs of a university or research institute? Liquavista is a spin-off of what used to be Philips Research (Natlab), one of last strongholds of real industrial research in Europe. It used to be a pretty amazing place, maybe the closest of what Europe had to the Bell Labs. Unfortunately it was messed up pretty badly during the last decade due to various splits, carve outs and a general move away from industry backed research.
Actually you don't need superconductors for this. High-voltage direct current transmission lines are very well capable of delivering electricity with high efficiency across long distance without superconductors. Existing projects, like the Quebec-New Englad transmission line are capable of carrying >2GW of electrical energy over distances of >1100km. This is far more than even the largest photovoltaic power plant can generate today.
Generally agreed. But I'd like to point out the semiconductor manufacturing uses several "nanotechnology" methods besides lithography. For example the high-k deposition employs an atomic layer deposition process (ALD), that allows precise control of film thicknesses down to tenths of a monolayer. This is achieved by surface limited reactions, very similar to many techniques within the realm of "self-assembly" or bottom-up.
Having an "assembler" on the atomic level would of course be a long time goal. However it is very likely that this is not possible. Atomic interactions simply do not allow for arbitrary combinations of individual atoms.
This is a cross section of the pmos transistors in one of Intels 45nm high-k metal gate CPUs. As you can see there are many layers with a horizontal and lateral extend far below 10 nm. In fact the thinnest layers are in the order of 1-2nm - The gate stack itself consists of a multilayer stack of SiO2/HfO2/TiN, where each of the layers is only 1-3 nm thick.
How is this not nanotechnology?
Most of the known bottom up approaches that are hyped and studied at universities, such as nanoparticles and nanowires, lead to significantly larger structures.
Top down beat bottom up years ago. Sorry guys, it's a nice phd topic but the industry is already there.
not been physically realized and connected with that theory until recently.
Actually that is not true. Resistive switching had been demonstrated even before 1971 (there are some examples on Wikipedia).
Let's see, what happened here?
1) Someone found resistive switching 2) Someone developed a trivial algebraic model of a resistor with a memory effect 3) Someone with good marketing skills and connections combines 1) and 2) and manages to ram a paper into Nature. 4) Lots of press hype ensues 5) Profit? nope. Novel? nope. Publicity? Way too much, given for the nonevent. Application? Was already there at 1) in form of RRAM.
Did anybody notice that a new class of high temperature superconductors was found recently? Or multiferroics? Or commercial availability of phase change memory (PCRAM), which may render RRAM obselete before it went anywhere?
Sorry to be so harsh, but the specific experiment reported here is of little to none value outside of science. Why?
Hysteretic resistive switching in metal oxide systems is a well known phenomenon (RRAM) and occurs in all transition metal oxides with noble eletrodes. This is what has been recristened as "Memrestor" by HP. It is widely agreed upon that this switching mechanism is due to a redox reaction where oxygen is added or removed from the insulator. The specifics (filament, interfacial barrier lowering etc.) are still subject of current research though.
The experiment in the paper takes a slightly different approach: vanadium oxide has a very interesting property where its resistance switches apruptly by orders of magnitude at a certain temperature due to a reorganisation of its electronic structure. This phenomenon is known as metal to insulator (MTI) transition and has been research for at least 50 years.
The MTI has a hysteretic behavior which means that it retains its state if you vary the temperature only a little above or below the critical MTI temperature Tc. The researchers have now shown that if you keep the temperature of the system close to Tc, you can use an additional electric current to switch the resistivity of the system. A possible explanation could be self heating.
Why is it useless for practical application?
1) The phenomenon instrinsically only works at a certain temperature. Deviations by fractions of degrees K will destroy all information.
2) As far as I can see they only demonstrated electrical switching into one direction. To erase the memory both would be required.
All in all a nice experiment, but again with typical university style hype, piggybacking on the Memristor craze.
I am also relatively certain that current driven MTI switching has been reported before. I am aware of a couple of experiments where a field switched MTI transition was proposed for transistors. Those devices should exhibit exactly the same hysteresis and "memory" properties.
I think it is quite useful of you want to work while travelling in an airplane on a train. The 9" netbooks are not really good for anything that involves a lot of typing.
A bought a DELL latitude x200 off ebay a couple of years ago for exactly that reason and I have never regretted it. Back then this was still a business notebook and costed $3000+ (I paid $250, years later). The $999 price point is not too bad.
The main drawback seems to be the battery. But did you know they had outlets in many european trains?
Ok, so they are able to extract some information from a ciphertext that was created from a low entropy plaintext. It is quite obvious that redundancy in data creates vulnerabilities. This is exactly the reason why any serious encryption software applied data compression (entropy coder) before encrypting.
Actually simply storing images in any format with data compression would completely invalidate this method. (eg. JPG, TIFF, PNG, GIF)
There is an obvious flaw in your assertion: The people who don't cope with linux today are the 95% of the population that did not own a computer at all in the 80ies and early 90ies.
Computers are not targetted at professionals and enthusiasts anymore.
Sorry, but Origins spread sheet functions are horrible and rudimentary. There is a reason why it offers embedded Excel spreadsheets.
Working with Matlab is only helpful if you work with a large amount of similar data. In many experimental fields you have to deal with small datasets of varying formats. Especially if you are fiddling with the set up of new characterization methods. Excel is really quite useful for that.
Indeed, they have not even demonstrated working devices yet. The press release is nothing but the announcement of the utilization of one specific technique.
09/19/2008, The internet. Slashdot user Bender_ announces to leapfrog IBM and Intel by intending to build 10 nm structures in his garage.
Slashdot Mirror
>. I don't know what Atmel did to deserve their good luck.
There was a long time when it basically was AVR vs PIC when it came to "small developer" aka hobbyist microcontrollers. The PIC may have been there first, but the AVR architecture is much more user friendly and has a following that is at least as large as that of the PIC. The reason why the AVR is being used in the arduino is probably because of it's high-level language compatible microarchitecture.
Seems to me for most uses simply increasing the refresh time interval would save tons of power, and also complexity. If you could get it to a couple of days,
Yes, increasing the refresh time is indeed a way to reduce power consumption of a DRAM. The problem is that you are dealing with billions of memory cells. The median retention time of typical cells is well within the range of seconds. But there is a tiny fraction of cells (1/10000) that lose their charge much quicker, and things may get worse at elevated temperatures etc. Those cells impose limits on the minimum refresh time.
There are ways to work around this by introducing on-the-fly error correction. But this will result in a larger device and added latency, which is obviously not desired in many applications. Nevertheless, there are dedicated low power DRAMs which use this kind of scheme to increase refresh time.
I am a semiconductor scientist, but I completely fail to understand what this news is about. The article does nowhere mention the materials used, the device behavior, the application, the purpose or anything else.
A MIM device as is, is a capacitor. And that is exactly what the picture is showing. When this type of capacitor is scaled to the nanometer regime it starts to get leaky due to quantum mechanical tunneling through the dielectric. The abstract mentions 'controlled quantum mechanical tunneling'... Aha, this could be what it is about. But as long as metal electrodes are involved this will only create a nonlinear resistor. Still no idea what the exact purpose is.
Are nanoscale MIM capacitors new? No, not at all. Right now you have billions of them doing their job in your computers main memory. Depending on the vintage of your computer, these capacitors employ nanolaminates of ZrO2 and Al2O3 at a total thickness of 5 to 10 nanometers. Quantum electrical tunneling is of high relevance in these devices, since it leads to loss of stored information. So, is cheap new? A quick calculation suggests that the manufacturing cost of a single MIM device in a DRAM is approximately 10^(-10) US$.
This is not true. You need to be aware of one thing: "Memristors" were not new when they were "discovered". The memory industry knew the concept years before as RRAM. I can assure you that all other nonvolatile memory vendors are developing RRAM or are at least looking into the possibilities. Samsung has been publishing about NiO based RRAM long before it was "discovered" again, IBM has some interesting papers from the Zurich labs. Furthermore, there are several start up companies looking into 3D RRAM which may offer densities far above that of flash. Matrix Semiconductors (bought by Sandisk) and a company by a former Micron guy.
One significant issue with RRAM (and the memristor) so far is that the memory cells have to be "formed". They need an initial high voltage pulse to induce the switching behaviour. This is something that is difficult to do when you have billions of memory cells. To my knowledge no good solution to this problem has been found yet, although there is progress.
This has nothing to do with "leakage" current. As basic field effect transistor theory will teach you, there is a region below the threshold voltage where the current depends exponentially on the gate bias. Yes, exponentially instead of linearly or quadratically as in the "on" region. This means that small changes in the gate bias will allow for a huge change in current. The drawback is here that we are talking about extremely low current. In CMOS logic this equals lower operation frequency.
This idea here is to build circuits that are extremely power efficient but also extremely slow. As mentioned in the article, potential applications may include wireless sensor networks that have to operate on extremely low energy.
You know what the most disturbing thing is?
Most DRAM companies have operated at a net loss when taking into account the accumulated earnings of the last decade. There is incredibly fierce price competition within the industry.
Do you really feel ripped off when you buy a product that is composed of billions of transistors, has tens of billions of R&D costs behind at at a price of $1 ? (That was the price of a 1Gbit chip not long ago) I don't want to sound like an industry advocate here, but I find this pretty staggering.
Is the consumer really ripped off if he has to buy a product that is priced a few months behind Moores law?
Growing from $50m to $1B in five years would mean that the market for open sources increases twentyfold in just five years. OSS has been around since the 80ies, OSS companies have been around since the nineties. So it took 15 years for the market to grow to $50m, why on earth should it increase twentyfold in just five measly years? Like many analysts, he is just making up numbers.
Well, maybe it cannot be done because it is sealed into a rigid structure?
This sounds very much like a Dye-sensitized solar cell, also known as Graetzel cell.
Unfortunately that means that the new invention does probably share the same (unsolved) long term stability problems.
Aren't start-ups almost always spinoffs of a university or research institute? Liquavista is a spin-off of what used to be Philips Research (Natlab), one of last strongholds of real industrial research in Europe. It used to be a pretty amazing place, maybe the closest of what Europe had to the Bell Labs. Unfortunately it was messed up pretty badly during the last decade due to various splits, carve outs and a general move away from industry backed research.
Both Berlin and Paris had a networks with a total length of more than 400km.
obvious link
Fewer instructions does not always mean that the CPU architecture gets more optimized.
To my knowledge, in terms of gatecount this is the most efficient CPU around:
http://www.opencores.org/project,mcpu
32 Macrocells in a CPLD.
Actually you don't need superconductors for this. High-voltage direct current transmission lines are very well capable of delivering electricity with high efficiency across long distance without superconductors. Existing projects, like the Quebec-New Englad transmission line are capable of carrying >2GW of electrical energy over distances of >1100km. This is far more than even the largest photovoltaic power plant can generate today.
Generally agreed. But I'd like to point out the semiconductor manufacturing uses several "nanotechnology" methods besides lithography. For example the high-k deposition employs an atomic layer deposition process (ALD), that allows precise control of film thicknesses down to tenths of a monolayer. This is achieved by surface limited reactions, very similar to many techniques within the realm of "self-assembly" or bottom-up.
Having an "assembler" on the atomic level would of course be a long time goal. However it is very likely that this is not possible. Atomic interactions simply do not allow for arbitrary combinations of individual atoms.
well, not compared to entire chips, but single transistors.
Viruses and especially bacteria are huuge compared to microchips.
This is a cross section of the pmos transistors in one of Intels 45nm high-k metal gate CPUs. As you can see there are many layers with a horizontal and lateral extend far below 10 nm. In fact the thinnest layers are in the order of 1-2nm - The gate stack itself consists of a multilayer stack of SiO2/HfO2/TiN, where each of the layers is only 1-3 nm thick.
How is this not nanotechnology?
Most of the known bottom up approaches that are hyped and studied at universities, such as nanoparticles and nanowires, lead to significantly larger structures.
Top down beat bottom up years ago. Sorry guys, it's a nice phd topic but the industry is already there.
not been physically realized and connected with that theory until recently.
Actually that is not true. Resistive switching had been demonstrated even before 1971 (there are some examples on Wikipedia).
Let's see, what happened here?
1) Someone found resistive switching
2) Someone developed a trivial algebraic model of a resistor with a memory effect
3) Someone with good marketing skills and connections combines 1) and 2) and manages to ram a paper into Nature.
4) Lots of press hype ensues
5) Profit? nope. Novel? nope. Publicity? Way too much, given for the nonevent. Application? Was already there at 1) in form of RRAM.
Did anybody notice that a new class of high temperature superconductors was found recently? Or multiferroics? Or commercial availability of phase change memory (PCRAM), which may render RRAM obselete before it went anywhere?
s/MTI/MIT
Sorry to be so harsh, but the specific experiment reported here is of little to none value outside of science. Why?
Hysteretic resistive switching in metal oxide systems is a well known phenomenon (RRAM) and occurs in all transition metal oxides with noble eletrodes. This is what has been recristened as "Memrestor" by HP. It is widely agreed upon that this switching mechanism is due to a redox reaction where oxygen is added or removed from the insulator. The specifics (filament, interfacial barrier lowering etc.) are still subject of current research though.
The experiment in the paper takes a slightly different approach: vanadium oxide has a very interesting property where its resistance switches apruptly by orders of magnitude at a certain temperature due to a reorganisation of its electronic structure. This phenomenon is known as metal to insulator (MTI) transition and has been research for at least 50 years.
The MTI has a hysteretic behavior which means that it retains its state if you vary the temperature only a little above or below the critical MTI temperature Tc. The researchers have now shown that if you keep the temperature of the system close to Tc, you can use an additional electric current to switch the resistivity of the system. A possible explanation could be self heating.
Why is it useless for practical application?
1) The phenomenon instrinsically only works at a certain temperature. Deviations by fractions of degrees K will destroy all information.
2) As far as I can see they only demonstrated electrical switching into one direction. To erase the memory both would be required.
All in all a nice experiment, but again with typical university style hype, piggybacking on the Memristor craze.
I am also relatively certain that current driven MTI switching has been reported before. I am aware of a couple of experiments where a field switched MTI transition was proposed for transistors. Those devices should exhibit exactly the same hysteresis and "memory" properties.
I think it is quite useful of you want to work while travelling in an airplane on a train. The 9" netbooks are not really good for anything that involves a lot of typing.
A bought a DELL latitude x200 off ebay a couple of years ago for exactly that reason and I have never regretted it. Back then this was still a business notebook and costed $3000+ (I paid $250, years later). The $999 price point is not too bad.
The main drawback seems to be the battery. But did you know they had outlets in many european trains?
Ok, so they are able to extract some information from a ciphertext that was created from a low entropy plaintext. It is quite obvious that redundancy in data creates vulnerabilities. This is exactly the reason why any serious encryption software applied data compression (entropy coder) before encrypting.
Actually simply storing images in any format with data compression would completely invalidate this method. (eg. JPG, TIFF, PNG, GIF)
There is an obvious flaw in your assertion: The people who don't cope with linux today are the 95% of the population that did not own a computer at all in the 80ies and early 90ies.
Computers are not targetted at professionals and enthusiasts anymore.
Sorry, but Origins spread sheet functions are horrible and rudimentary. There is a reason why it offers embedded Excel spreadsheets.
Working with Matlab is only helpful if you work with a large amount of similar data. In many experimental fields you have to deal with small datasets of varying formats. Especially if you are fiddling with the set up of new characterization methods. Excel is really quite useful for that.
Indeed, they have not even demonstrated working devices yet. The press release is nothing but the announcement of the utilization of one specific technique.
09/19/2008, The internet. Slashdot user Bender_ announces to leapfrog IBM and Intel by intending to build 10 nm structures in his garage.