The cool thing about MgB2 is that people are hoping they can "tweak" it to superconduct above the magic 77K (the cupric oxide semiconductors were also discovered with a Tc around 40K, and quickly improved), while still being easier to work with than ceramic superconductors and more importantly, it points to new areas of research that might improve our understanding of superconductors in general, possibly leading to mugh higher temperature supercondutors
Well, like any electric line, the person who cut it is in a world of hurt. Presumably the lN2 system has some sort of safety shutoff the prevents it from pumping too much lN2 out of a rupture, but even so, there could be enough cold nitrogen gas in the area to suffocate someone nearby.
It sounds from the article like this isn't likely to be a problem in their location: They are doing this because it is so hard to dig there.
No, he is right. I don't know where you took your thermodynamics classes, but you were misled. All materials have a "critical point" above which there is no liquid state. To be more precise, there is no liquid-gas phase transition -- in fact what you get at high pressures is a mush that transforms smoothly from "mostly liquid like" to "mostly gas like", but there is never any 2 phase seperation.
For N2, the critical point is below room temperature.
I suspect that there was at the very least a pressure release valve somewhere on your ln2 tank.
Finally, I suppose they could use N2 based vapor cooling along the length of this pipe, but they would have to be screwed in the head to want to. You would have to build the whole system to withstand several thousands of psi with controlled bleed valves the whole way. It is much, much simpler to pump low temperature lN2 the whole way where your equipment only has to handle a small pressure over 1 atm.
Actually, it should be substantially cheaper. High power underground lines, including these ones, are usually oil cooled. Oil cooling is pretty expensive, since you have to somewhere dissipage quite a bit of heat. Liquid nitrogen is cheaper than water, and the superconductor doesn't produce any heat.
People are also looking at using this kind of wire in high power electric motors and transformers for the same reason -- not efficiency, but size and cost.
First, as the poster above said, that is a hardware error, and exactly the kind of thing that affects MTBF numbers for various hardware devices.
Second, cosmic rays do not actually cause memory errors. What can cause errors are alpha particles, usually given off by decay of trace radioactive elements in the ceramic casing of ICs. This is a problem for any IC, and they have to be designed to withstand it. I don't know if that is actually a significant source of errors in modern memory or not, though.
Of course, the #1 cause of memory errors is defective memory, but that is another issue entirely:)
It isn't that you are paying for rarity, but a company has to be able to recoup their R&D costs in a reasonable time. Even if the engine is a "stock" helicopter component, there has to be some considerable custom engineering on this. That costs money.
A typical target "time to recover initial investment" is 5 years. So if they sell 25 over the next 5 years, that is $3.75 million, which has to cover materials, labor, development, and still hopefully turn a profit. That seems unlikely to me. Hopefully, they know more about their financial situation than I do:)
Still, it is way cool. Anyone know if this is a record for power/weight ratio? The specs say 320 HP/ 460 lb.
Hardware can be meaningfully rated with a MTBF value because the errors are random, physical (often mechanical) defects. It is a measure of how likely a given operation is to fail.
With software, usually the same operation always fails. Software errors are design errors, not random failure.
While measuring the frequency of breakins is perhaps a useful metric, it shouldn't be confused with something like a MTBF for hardware. Also, the frequency of breakins due to script kiddies that scan and more-or-less target systems at random and just want a shell, or to deface your webpage, versus a deliberate and directed attack against you to steal/corrupt data are completely unrelated. The latter may have access to more sophisticated tools, better knowledge of your network and software, etc. Trying to apply numbers gained from random attacks to indicated your defendability against directed attacks is severely misguided.
Also, attempting an MTBF rating doesn't take into account visibility. If I drop most incoming connections through my cable modem and run a port scan detector, most people scanning my whole ISP will not even notice I am there. This doesn't work for a public website that many people know exists, even if it does drop their traffic to port 31337. Hardware MTBF is usually given in "operating hours" or some other well specified metric. I don't see how to do that for software. A better metric might be the ratio of intrusions/attempts, but since I would wager the majority if intrusion attempts, and even many successful ones are never discovered, that isn't a really good metric, either.
Photons are pretty stable against these kinds of interference, since they don't interact strongly with matter. So, as long as you use high quality optical elements, you can do quantum computation experements at room temperature. Unfortunately, the lack of interactions makes it hard to implement quantum gates, making systems larger than 2 qubits difficult.
NMR quantum computer experements are also at room temperature, but have their own problems.
It isn't that a QC couldn't conveivably replace your desktop -- while it currently doesn't look feasable, a QC can emulate anything a classical computer can do.
However, there are some things that a QC doesn't do any better than a classical computer. In fact, most things, a classical computer does just as well as a QC. So the likelyhood of ever wanting to replace your computer with a QC is pretty small.
QCs have some real downsides that make implementing a general purpose one impractical or expensive. For one things, because of the no-cloning theorem (which states that you can't duplicate a quantum state without destroying the orignial), you cannot do a fan-out (connect one gate output to several gate inputs). There are ways around this, but I doubt it will ever be worth it.
Note that a classical computer is one that implements a turing machine, and a quantum computer is one that has a time-evolution operator given by the Schroding equation. "classical computers" implemented with silicon are obviously quantum mechanical devices. Classical computers will eventually have to face the quantum indeterminism of the extremely small scale, but they will still be classical comptuers -- even if they aren't implemented as silicon based semiconductors.
I just did some similar benchmarks on XFS vs. ReiserFS with and without the notail option, and ext2 (for reference -- I wanted to switch to a journaled filesystem).
My tests were 1) copy a several gigabyte file tree (/home) to a temp filesystem, 2) run du on the tree, 3) run Bonnie++, 4) build a kernel, and 5) rm -rf the whole tree.
My results were that ext2 and xfs had high read/write throughput, resierfs-notail was slightly lower, and resierfs without notail was a fair bit slower.
All metadata intensive operations (rm -rf, du, bonnie++ small files, etc) were blazing fast on ReiserFS, slow as molasses on xfs, and fast on ext2, but scaling poorly to high numbers of files.
xfs could do slightly more random seeks than either ext2 which in turn beat out Reiserfs, by about the same amount.
reiserfs doesn't have an fsck -- at least the one I have is a noop.
I switched to reiserfs-notail, and was proptly annoyed by the lack of fsck, but otherwise it is running well. I ran a pair of make -j6's, some directory copies, and an updatedb, then hit the power and it is running fine.
My personal preference would be to have ext3 in 2.4 along with Daniel Phillips' hash-tree directory patch stabalized--I think that would smoke reiserfs and still give me journaling.
Well, that is you. But some people would rather be a captain than a tourist.
I am not planning on following either of these guys, but I have more respect for this guy than Tito.
Tito is basically a spoiled rich guy who has figured out that for enough money you can get almost anyone to do almost anything. But designing and building the rocket yourself on another level.
They should make a Mastercard commercial about it:)
The article seemed to have an identiy crisis over what this was, but at least part of it was not talking about watermarking in the traditional sense. They only wanted to embed a single two-bit code every second. probably like CDs have a "copy-protected" and "is-a-copy" bits.If properly coded (ie, not a flashing white pixel), it should be well below the noise floor of the MPEG2 encoding to begin with.
However, later on they talked about identifying particular sources, which is begining to sound a lot more like watermarks ala digimark.
In either case, how resilient they are to tampering remains to be seen.
The funny thing is, this might keep someone who buys a DVD player in the future from copying a DVD to VHS, but as long as someone has a DVD-ROM that predates this, it will be possible to find perfect digital copies online.
Heh. Obviously everyone reads that as "If we try to make people upgrade to something that doesn't have significantly better quality, it will go the way of DIVX".
The problem with systems like this is that they rarely do anything to counteract large-scale piracy, but they usually *do* cause problems for individuls who aren't trying to break the law.
Honestly, this particular proposal, which basically sounds like the SCMS from DAT/CD isn't so bad. What pisses me off is CSS and region coding. Honestly, I don't understand why region coding is legal, and I think CSS ought to be illegal.
Also, while from the description in the article doesn't sound so bad, given the recent history of these IP organizations, I am going to reserve judgement until I actually see it. The fact is, they are a bunch of greedy, arrogant, unethical bastards that should all be thrown in jail.
Depending on your drive, you may need to set the region code of your DVD drive before you can read encrypted DVDs at all. This can be an issue if you have never used your drive under Windows. This was the case with my Pioneer.
They didn't say it would eliminate the bandwidth problem, they said it would eliminate the spectrum (the fact that we're running out of it) problem.
Well, the reason we are running out of spectrum is because high-bandwidth applications take a lot of spectrum.
Now part of the problem is that you always lose some of the spectrum when you divide it into channels finely. Upon further reading on their website, it looks to me like they aren't really going for particularly high bandwidth, but just being able to have many, many channels. After reading a large chunk of their whitepaper, I am skeptical that it is as scalable as they say, but it comes off as a lot more plausable than the articles linked to.
As a bit of historical context, every few years someone comes up with a brand new revolutionary way of dividing up the spectrum into channels that is incredibly efficient. Two that come to mind are CDMA and FHSS (used by cell phones and some wireless lans, respectively). In each case, once they make it to actual production and deployment it turns out they vastly overestimated the number of channels that can work simultaneously without causing severe degredation. I doubt this will be any different.
My best guess for the deadly factor is multi-pathing. While the pulsed nature of the radio should prevent destructive interference that causes cell phone fadeout in doors, it will probably substantially reduce the number of channels they can use before they start causing problems.
I am reading the information on timedomain's website, which includes a frequency spectrum analysis, and what they have actually demonstrated. First of all, their Their prototypes are not nearly as ridiculous sounding as the press articles I have seen, and aren't really that great except for their extremly low power consumption:
A full duplex 1.3 GHz system with an average output power of 250 microWatts, and a variable data rate of either 39 kbps or 156 kbps. The radio has been tested to beyond 16 kilometers (10 miles).
A full duplex 1.7 GHz walkie-talkie with an average output power of 2 milliWatts, a data rate of 32 kbps and a range of 900 meters. The unit was also capable of measuring the distance between radios with an accuracy of 3 cm (0.1 ft).
A simplex 2.0 GHz data link with an effective average output power of 50 microWatts, a data rate of 5 Mbps at bit error rate (BER) of 0 with no forward error correction (FEC) and a range of 10 meters (32 ft) through two walls inside an office building.
Also, their pulses are not remotely square, but gausian monocycles of the form: V=t*e^(-t^2). In addition to low power, they claim that they can do really dense channelization by using different clock sequences -- very similar to the way FHSS works, only by varying the time base, rather than frequency hopping.
Another plausable advantage is that since they don't use continious waves, multi-pathing isn't a big problem. The wave packets from the two paths are completely distinguishable, and therefore do not interfere. However, this makes each path look like a seperate transmitter on a different channel. So you sacrifice total bandwidth (by reducing the number of channels availbable) in exchange for reducing fade-out from point of destructive interference.
In any case, anyone interested should check out
this whitepaper
more info. I doubt this is a scalable as they claim, but they do have some interesting ideas, and the single-chip positioning and radar sounds cool, too.
I think this is a really important as well as scientifically responsible experement.
The fact is, one day we are going to have to come head to head with massive genetic manipulation. Eventually, people with genetic diseases will not accept that their children must be born with the same disease despite the technology to prevent it being available. Whether this is a Good Thing in the long run for the human race as a whole is unclear, and will likely not factor into the debate at all.
In any case, one day this will happen. Slight tinkering with mitochandrial DNA by transplanting whole healthy mitochondria is a relatively low-risk way to gain experience and knowledge about genetic engineering, since it doesn't involve gene splicing or removal of any of the original parents DNA.
Well, it is possible what he is doing is equivelant to a single-sideband radio, using the phase locked oscillators to reconstruct the signal. This mights justify their low power claims. Doesn't really sound like it to me, though. I should read the whitepaper on their website.
But the fact remains--to resolve 1ps pulses, you need electronics that can handle 1THz. That doesn't seem so likely to me:)
Not to mention that keeping two physically seperated clocks in lock-step with sub-picosecond accuracy is not exactly easy.
Not to be a naysayer, but this really isn't that remarkable. In particular, it basically has exactly the same constraints as "normal" frequency domain signals. In particular, it has to take up the same sized region of the spectrum as conventional broadcasting, and all of your electronic components have to work at the same high frequency to resolve data.
If you want to resolve picosecond pulses, you need electronics that can pass frequencies up to 1 THz.
That isn't to say there aren't applications, and I am going to read their whitepapers to see how they get such phenomenally low power usage, even when constrained by the inverse square law, but it isn't a revolutionary technology that is going to eliminate the bandwidth problem.
Short-range transmitters with closely spaced receivers connected by fibers could solve the bandwidth problem, but A) would require massive investment in infrastructure and B) be no better or worse than Bluetooth type proposals using more conventional radio technology.
Hmm... Some probjects may be like that, but there are others that aren't.
A very common requirement is to add a boolean or a timestamp field that is used by a maintenence batch process to determine if it needs to do something to a record. In the RDBMS world, you don't need to interupt your interactive application to replace a batch process that is likely running on a seperate machine and only at night.
In fact, with Oracle, you can, on the fly, add a boolean flag, plus an trigger to be run on update to set the flag to 1. Then your modified batch process can set it to 0 when it does it's thing.
Now, I would like to solve the "impedence mismatch" between 3GL code and RDBMSs, but I don't know if it is possible w/o sacrificing the generic nature of RDBMS. I would rather switch languages to something better suited to database work (PERLs DBI is pretty nice, compared to C/C++/Java interfaces, at least for what I used it for)
I know this was a joke, but I kind of agree with it. There are some things you can learn from books, and there are others you need to be taught. Unfortunately, our education system is moving rapidly towards teaching mostly things that can be learned from books.
It amazes me that so many people getting a CS degree take "database classes" where they learn not the theory of relational logic, nor algorithms for implementing databases, but instead SQL, the Dummy's guide to normalization (if they are lucky), and how to use JDBC to make DB backed web services.
These types of things are important, but they are all things that someone properly educated can learn from a book. What the educational system needs to do is give people a really excellent background in math, the sciences, and design theory or whatever is approrpriate in a given field. Specific applications and implementations should be viewed as test-beds to understand the abstract design concepts that underly the application.
A while ago people were making disturbing claims like "an engineering degree becomes obsolete in 5 years". Bullshit. A good engineering education should last a lifetime. Not only does a theoetical/abstract education resist obsolecense, it makes it easier for someone to change fields, or work on interdiciplinary areas of study. It gives you the background needed to learn whatever you need on your own.
Part of the problem is that universities, particularly state univiersities whose funding is dependent on the number of students enrolled go too far (IMO) torwards "treating the student as a customer." Students bitch and moan about how they don't care about calculus or physics, they want to get on to the interesting/practical stuff. And so the schools let them. Even if they don't most of it is ignored or immediately forgotten.
I could go on about what is wrong with our education system, going all the way back to kindergarden, but I have work to do...
The "original" PIII was designed to run (and only ever did run) up to ~600 MHz. The CuMine was a die-shrink (I believe) and migrated the cache on-chip. Anyway, a die shrink increases the transistor density, though it doesn't increase the number of transistors on a chip. Obviously the on-chip cache increases the number of transistors quite a bit, though memory usually has a very high transistor density compared to a CPU because of the high regularity in layout.
The CuMine also runs at a lower voltage, and therefore lower power.
These factors combine to allow the CuMine to run at speeds higher than the 600 MHz design speed of the original PIII, which was a core revision on the PII that I believe used the same process as the 100 MHz FSB PII's (ie, the 350, 400, and 450 MHz models).
How this factors in to whether intel was trying to track Moore's law rather than make the fastest processors they could, I am not really in a position to comment. I haven't really paid attention to this much recently.
This article is not talking about protecting news articles from nytimes (though if it actually worked, that might be a possible application). They are talking about protecting trade secrets or classified information from potential espionage. That is an environment where people would go to extrodinary lengths to copy data without it being recorded.
Many years ago now, the DOD tried to push the Orange Book as a solution to this problem, and IMO, it was a dramatic failure. But in any case, any implementation requires a trusted client terminal, either a tamperproof PC or preferably a terminal in a secure facillity (where you can observe to make sure people don't take pictures, copy down notes, etc). They you just have to worry about people remembering everything well enough to copy it down later.
As content protection for copyrighted material (music, nytimes articles, pr0n), just making it "to painful" to reproduce might be good enough to prevent the majority of casual or unintentional copying. However, once again, people forget the primary attribute of the virtual world: "All marginal costs are zero". Once someone discovers how to circumvent the plugin, the process can be automated and provided as a patch and you will never have to worry about it again.
Slashdot Mirror
The cool thing about MgB2 is that people are hoping they can "tweak" it to superconduct above the magic 77K (the cupric oxide semiconductors were also discovered with a Tc around 40K, and quickly improved), while still being easier to work with than ceramic superconductors and more importantly, it points to new areas of research that might improve our understanding of superconductors in general, possibly leading to mugh higher temperature supercondutors
Well, like any electric line, the person who cut it is in a world of hurt. Presumably the lN2 system has some sort of safety shutoff the prevents it from pumping too much lN2 out of a rupture, but even so, there could be enough cold nitrogen gas in the area to suffocate someone nearby.
It sounds from the article like this isn't likely to be a problem in their location: They are doing this because it is so hard to dig there.
No, he is right. I don't know where you took your thermodynamics classes, but you were misled. All materials have a "critical point" above which there is no liquid state. To be more precise, there is no liquid-gas phase transition -- in fact what you get at high pressures is a mush that transforms smoothly from "mostly liquid like" to "mostly gas like", but there is never any 2 phase seperation.
For N2, the critical point is below room temperature.
I suspect that there was at the very least a pressure release valve somewhere on your ln2 tank.
Finally, I suppose they could use N2 based vapor cooling along the length of this pipe, but they would have to be screwed in the head to want to. You would have to build the whole system to withstand several thousands of psi with controlled bleed valves the whole way. It is much, much simpler to pump low temperature lN2 the whole way where your equipment only has to handle a small pressure over 1 atm.
Actually, it should be substantially cheaper. High power underground lines, including these ones, are usually oil cooled. Oil cooling is pretty expensive, since you have to somewhere dissipage quite a bit of heat. Liquid nitrogen is cheaper than water, and the superconductor doesn't produce any heat.
People are also looking at using this kind of wire in high power electric motors and transformers for the same reason -- not efficiency, but size and cost.
First, as the poster above said, that is a hardware error, and exactly the kind of thing that affects MTBF numbers for various hardware devices.
:)
Second, cosmic rays do not actually cause memory errors. What can cause errors are alpha particles, usually given off by decay of trace radioactive elements in the ceramic casing of ICs. This is a problem for any IC, and they have to be designed to withstand it. I don't know if that is actually a significant source of errors in modern memory or not, though.
Of course, the #1 cause of memory errors is defective memory, but that is another issue entirely
It isn't that you are paying for rarity, but a company has to be able to recoup their R&D costs in a reasonable time. Even if the engine is a "stock" helicopter component, there has to be some considerable custom engineering on this. That costs money.
:)
A typical target "time to recover initial investment" is 5 years. So if they sell 25 over the next 5 years, that is $3.75 million, which has to cover materials, labor, development, and still hopefully turn a profit. That seems unlikely to me. Hopefully, they know more about their financial situation than I do
Still, it is way cool. Anyone know if this is a record for power/weight ratio? The specs say 320 HP/ 460 lb.
Hardware can be meaningfully rated with a MTBF value because the errors are random, physical (often mechanical) defects. It is a measure of how likely a given operation is to fail.
With software, usually the same operation always fails. Software errors are design errors, not random failure.
While measuring the frequency of breakins is perhaps a useful metric, it shouldn't be confused with something like a MTBF for hardware. Also, the frequency of breakins due to script kiddies that scan and more-or-less target systems at random and just want a shell, or to deface your webpage, versus a deliberate and directed attack against you to steal/corrupt data are completely unrelated. The latter may have access to more sophisticated tools, better knowledge of your network and software, etc. Trying to apply numbers gained from random attacks to indicated your defendability against directed attacks is severely misguided.
Also, attempting an MTBF rating doesn't take into account visibility. If I drop most incoming connections through my cable modem and run a port scan detector, most people scanning my whole ISP will not even notice I am there. This doesn't work for a public website that many people know exists, even if it does drop their traffic to port 31337. Hardware MTBF is usually given in "operating hours" or some other well specified metric. I don't see how to do that for software. A better metric might be the ratio of intrusions/attempts, but since I would wager the majority if intrusion attempts, and even many successful ones are never discovered, that isn't a really good metric, either.
Photons are pretty stable against these kinds of interference, since they don't interact strongly with matter. So, as long as you use high quality optical elements, you can do quantum computation experements at room temperature. Unfortunately, the lack of interactions makes it hard to implement quantum gates, making systems larger than 2 qubits difficult.
NMR quantum computer experements are also at room temperature, but have their own problems.
It isn't that a QC couldn't conveivably replace your desktop -- while it currently doesn't look feasable, a QC can emulate anything a classical computer can do.
However, there are some things that a QC doesn't do any better than a classical computer. In fact, most things, a classical computer does just as well as a QC. So the likelyhood of ever wanting to replace your computer with a QC is pretty small.
QCs have some real downsides that make implementing a general purpose one impractical or expensive. For one things, because of the no-cloning theorem (which states that you can't duplicate a quantum state without destroying the orignial), you cannot do a fan-out (connect one gate output to several gate inputs). There are ways around this, but I doubt it will ever be worth it.
Note that a classical computer is one that implements a turing machine, and a quantum computer is one that has a time-evolution operator given by the Schroding equation. "classical computers" implemented with silicon are obviously quantum mechanical devices. Classical computers will eventually have to face the quantum indeterminism of the extremely small scale, but they will still be classical comptuers -- even if they aren't implemented as silicon based semiconductors.
I just did some similar benchmarks on XFS vs. ReiserFS with and without the notail option, and ext2 (for reference -- I wanted to switch to a journaled filesystem).
My tests were 1) copy a several gigabyte file tree (/home) to a temp filesystem, 2) run du on the tree, 3) run Bonnie++, 4) build a kernel, and 5) rm -rf the whole tree.
My results were that ext2 and xfs had high read/write throughput, resierfs-notail was slightly lower, and resierfs without notail was a fair bit slower.
All metadata intensive operations (rm -rf, du, bonnie++ small files, etc) were blazing fast on ReiserFS, slow as molasses on xfs, and fast on ext2, but scaling poorly to high numbers of files.
xfs could do slightly more random seeks than either ext2 which in turn beat out Reiserfs, by about the same amount.
reiserfs doesn't have an fsck -- at least the one I have is a noop.
I switched to reiserfs-notail, and was proptly annoyed by the lack of fsck, but otherwise it is running well. I ran a pair of make -j6's, some directory copies, and an updatedb, then hit the power and it is running fine.
My personal preference would be to have ext3 in 2.4 along with Daniel Phillips' hash-tree directory patch stabalized--I think that would smoke reiserfs and still give me journaling.
2.4.0 has a dramatically improved mm system, most of the benefits of which don't show up on these tests, yet make a world of difference in real life.
Well, that is you. But some people would rather be a captain than a tourist.
:)
I am not planning on following either of these guys, but I have more respect for this guy than Tito.
Tito is basically a spoiled rich guy who has figured out that for enough money you can get almost anyone to do almost anything. But designing and building the rocket yourself on another level.
They should make a Mastercard commercial about it
The article seemed to have an identiy crisis over what this was, but at least part of it was not talking about watermarking in the traditional sense. They only wanted to embed a single two-bit code every second. probably like CDs have a "copy-protected" and "is-a-copy" bits.If properly coded (ie, not a flashing white pixel), it should be well below the noise floor of the MPEG2 encoding to begin with.
However, later on they talked about identifying particular sources, which is begining to sound a lot more like watermarks ala digimark.
In either case, how resilient they are to tampering remains to be seen.
The funny thing is, this might keep someone who buys a DVD player in the future from copying a DVD to VHS, but as long as someone has a DVD-ROM that predates this, it will be possible to find perfect digital copies online.
Heh. Obviously everyone reads that as "If we try to make people upgrade to something that doesn't have significantly better quality, it will go the way of DIVX".
The problem with systems like this is that they rarely do anything to counteract large-scale piracy, but they usually *do* cause problems for individuls who aren't trying to break the law.
Honestly, this particular proposal, which basically sounds like the SCMS from DAT/CD isn't so bad. What pisses me off is CSS and region coding. Honestly, I don't understand why region coding is legal, and I think CSS ought to be illegal.
Also, while from the description in the article doesn't sound so bad, given the recent history of these IP organizations, I am going to reserve judgement until I actually see it. The fact is, they are a bunch of greedy, arrogant, unethical bastards that should all be thrown in jail.
Depending on your drive, you may need to set the region code of your DVD drive before you can read encrypted DVDs at all. This can be an issue if you have never used your drive under Windows. This was the case with my Pioneer.
Well, the reason we are running out of spectrum is because high-bandwidth applications take a lot of spectrum.
Now part of the problem is that you always lose some of the spectrum when you divide it into channels finely. Upon further reading on their website, it looks to me like they aren't really going for particularly high bandwidth, but just being able to have many, many channels. After reading a large chunk of their whitepaper, I am skeptical that it is as scalable as they say, but it comes off as a lot more plausable than the articles linked to.
As a bit of historical context, every few years someone comes up with a brand new revolutionary way of dividing up the spectrum into channels that is incredibly efficient. Two that come to mind are CDMA and FHSS (used by cell phones and some wireless lans, respectively). In each case, once they make it to actual production and deployment it turns out they vastly overestimated the number of channels that can work simultaneously without causing severe degredation. I doubt this will be any different.
My best guess for the deadly factor is multi-pathing. While the pulsed nature of the radio should prevent destructive interference that causes cell phone fadeout in doors, it will probably substantially reduce the number of channels they can use before they start causing problems.
- A full duplex 1.3 GHz system with an average output power of 250 microWatts, and a variable data rate of either 39 kbps or 156 kbps. The radio has been tested to beyond 16 kilometers (10 miles).
- A full duplex 1.7 GHz walkie-talkie with an average output power of 2 milliWatts, a data rate of 32 kbps and a range of 900 meters. The unit was also capable of measuring the distance between radios with an accuracy of 3 cm (0.1 ft).
- A simplex 2.0 GHz data link with an effective average output power of 50 microWatts, a data rate of 5 Mbps at bit error rate (BER) of 0 with no forward error correction (FEC) and a range of 10 meters (32 ft) through two walls inside an office building.
Also, their pulses are not remotely square, but gausian monocycles of the form: V=t*e^(-t^2). In addition to low power, they claim that they can do really dense channelization by using different clock sequences -- very similar to the way FHSS works, only by varying the time base, rather than frequency hopping.Another plausable advantage is that since they don't use continious waves, multi-pathing isn't a big problem. The wave packets from the two paths are completely distinguishable, and therefore do not interfere. However, this makes each path look like a seperate transmitter on a different channel. So you sacrifice total bandwidth (by reducing the number of channels availbable) in exchange for reducing fade-out from point of destructive interference.
In any case, anyone interested should check out this whitepaper more info. I doubt this is a scalable as they claim, but they do have some interesting ideas, and the single-chip positioning and radar sounds cool, too.
I think this is a really important as well as scientifically responsible experement.
The fact is, one day we are going to have to come head to head with massive genetic manipulation. Eventually, people with genetic diseases will not accept that their children must be born with the same disease despite the technology to prevent it being available. Whether this is a Good Thing in the long run for the human race as a whole is unclear, and will likely not factor into the debate at all.
In any case, one day this will happen. Slight tinkering with mitochandrial DNA by transplanting whole healthy mitochondria is a relatively low-risk way to gain experience and knowledge about genetic engineering, since it doesn't involve gene splicing or removal of any of the original parents DNA.
Well, it is possible what he is doing is equivelant to a single-sideband radio, using the phase locked oscillators to reconstruct the signal. This mights justify their low power claims. Doesn't really sound like it to me, though. I should read the whitepaper on their website.
:)
But the fact remains--to resolve 1ps pulses, you need electronics that can handle 1THz. That doesn't seem so likely to me
Not to mention that keeping two physically seperated clocks in lock-step with sub-picosecond accuracy is not exactly easy.
Not to be a naysayer, but this really isn't that remarkable. In particular, it basically has exactly the same constraints as "normal" frequency domain signals. In particular, it has to take up the same sized region of the spectrum as conventional broadcasting, and all of your electronic components have to work at the same high frequency to resolve data.
If you want to resolve picosecond pulses, you need electronics that can pass frequencies up to 1 THz.
That isn't to say there aren't applications, and I am going to read their whitepapers to see how they get such phenomenally low power usage, even when constrained by the inverse square law, but it isn't a revolutionary technology that is going to eliminate the bandwidth problem.
Short-range transmitters with closely spaced receivers connected by fibers could solve the bandwidth problem, but A) would require massive investment in infrastructure and B) be no better or worse than Bluetooth type proposals using more conventional radio technology.
Hmm... Some probjects may be like that, but there are others that aren't.
A very common requirement is to add a boolean or a timestamp field that is used by a maintenence batch process to determine if it needs to do something to a record. In the RDBMS world, you don't need to interupt your interactive application to replace a batch process that is likely running on a seperate machine and only at night.
In fact, with Oracle, you can, on the fly, add a boolean flag, plus an trigger to be run on update to set the flag to 1. Then your modified batch process can set it to 0 when it does it's thing.
Now, I would like to solve the "impedence mismatch" between 3GL code and RDBMSs, but I don't know if it is possible w/o sacrificing the generic nature of RDBMS. I would rather switch languages to something better suited to database work (PERLs DBI is pretty nice, compared to C/C++/Java interfaces, at least for what I used it for)
I know this was a joke, but I kind of agree with it. There are some things you can learn from books, and there are others you need to be taught. Unfortunately, our education system is moving rapidly towards teaching mostly things that can be learned from books.
It amazes me that so many people getting a CS degree take "database classes" where they learn not the theory of relational logic, nor algorithms for implementing databases, but instead SQL, the Dummy's guide to normalization (if they are lucky), and how to use JDBC to make DB backed web services.
These types of things are important, but they are all things that someone properly educated can learn from a book. What the educational system needs to do is give people a really excellent background in math, the sciences, and design theory or whatever is approrpriate in a given field. Specific applications and implementations should be viewed as test-beds to understand the abstract design concepts that underly the application.
A while ago people were making disturbing claims like "an engineering degree becomes obsolete in 5 years". Bullshit. A good engineering education should last a lifetime. Not only does a theoetical/abstract education resist obsolecense, it makes it easier for someone to change fields, or work on interdiciplinary areas of study. It gives you the background needed to learn whatever you need on your own.
Part of the problem is that universities, particularly state univiersities whose funding is dependent on the number of students enrolled go too far (IMO) torwards "treating the student as a customer." Students bitch and moan about how they don't care about calculus or physics, they want to get on to the interesting/practical stuff. And so the schools let them. Even if they don't most of it is ignored or immediately forgotten.
I could go on about what is wrong with our education system, going all the way back to kindergarden, but I have work to do...
The "original" PIII was designed to run (and only ever did run) up to ~600 MHz. The CuMine was a die-shrink (I believe) and migrated the cache on-chip. Anyway, a die shrink increases the transistor density, though it doesn't increase the number of transistors on a chip. Obviously the on-chip cache increases the number of transistors quite a bit, though memory usually has a very high transistor density compared to a CPU because of the high regularity in layout.
The CuMine also runs at a lower voltage, and therefore lower power.
These factors combine to allow the CuMine to run at speeds higher than the 600 MHz design speed of the original PIII, which was a core revision on the PII that I believe used the same process as the 100 MHz FSB PII's (ie, the 350, 400, and 450 MHz models).
How this factors in to whether intel was trying to track Moore's law rather than make the fastest processors they could, I am not really in a position to comment. I haven't really paid attention to this much recently.
This article is not talking about protecting news articles from nytimes (though if it actually worked, that might be a possible application). They are talking about protecting trade secrets or classified information from potential espionage. That is an environment where people would go to extrodinary lengths to copy data without it being recorded.
Many years ago now, the DOD tried to push the Orange Book as a solution to this problem, and IMO, it was a dramatic failure. But in any case, any implementation requires a trusted client terminal, either a tamperproof PC or preferably a terminal in a secure facillity (where you can observe to make sure people don't take pictures, copy down notes, etc). They you just have to worry about people remembering everything well enough to copy it down later.
As content protection for copyrighted material (music, nytimes articles, pr0n), just making it "to painful" to reproduce might be good enough to prevent the majority of casual or unintentional copying. However, once again, people forget the primary attribute of the virtual world: "All marginal costs are zero". Once someone discovers how to circumvent the plugin, the process can be automated and provided as a patch and you will never have to worry about it again.