Shannon's law says that the number of bits per second (bps) you can get down a channel per Hertz of bandwidth equals the log (base 2) of the signal-to-noise ratio. In the article they bandy about numbers like 90 bps/Hz. There's nothing impossible about this, as long as your signal power is 2^90 bigger than the power of the background noise. 2^90 = 1237 trillion trillion. So if you're using VPSK/2 to modulate the world's brightest searchlight on a moonless night in a coal mine, you could probably get this signal-to-noise ratio. But with real radio signals in the real world? Not likely.
This whole topic is some combination of genuinely good technology, hype, exaggerations by imperfect journalists, and fraud. From where I'm sitting, I don't know which of the above causes is the true explanation. But I doubt Shannon's law has been repealed. Alas...
Contrary to what it says at the top of this page, Tera is not a 'new kid on the supercomputing block'. They've been around since 1987, and their design has been around since before that.
I was one of the designers of the CM-5 supercomputer at Thinking Machines Co. back in the '80s. One of the terrible things about designing a supercomputer is that it takes so long, Moore's law has moved the underlying technology out from under you before you're done. We had to throw away a mostly-completed microprocessor design when it became clear that we would be beaten by commodity microprocessors, and re-engineer the rest of the system to use a big pile of commodity SPARC chips instead of a big pile of our custom whiz-bangs. It was an enormous waste and really bad for morale (mine, anyway). So we ended up slipping one generation.
I find it hard to imagine what it must be like to work at Tera. I read Burton Smith's first paper, proposing what has now become the MTA, in 1985. Computing has gotten a thousand times faster since then, and they're still plugging away. Imagine the number of designs they have had to throw away! The've probably had to redesign the system every 18 months, just to stay even, without ever being able to build it, until they managed to deploy one machine, last July. It's an incredibly long slog, espescially for an industry like computers, that rewards sprinters, not sloggers.
Iridium, unable to find other suitors, asked the bankruptcy court to approve a plan that would crash the Iridium satellites into the atmosphere and let them burn up, potentially causing a rain of molten metal across the globe and one heck of a tax write off for all involved.
Anybody that could write this paragraph isn't competent to run a satellite network (not that many of us are, but most of us aren't claiming we can.) They don't know much about either the economics or the mechanics of the business.
Economic bug #1: They're not going to get a tax write-off for this. They're losing money, so they don't pay any taxes anyway. The problem is that they owe more money than they can pay, and a Judge has commanded them to stop running up more debts, and pay the ones they have already. They'll probably end up paying pennies on the dollar.
Economic bug #2: Iridium spent years getting permission to broadcast in 167 separate countries. Are all those foriegn telecommunications agencies going to be happy about handing those permissions to a bunch of geeks that advertises itself as "beyond the reach of any govt"? Fat chance. And without the approval of the local governments, the network won't be allowed to operate.
Mechanical bug #1: They're not burning the satellites for a tax write-off, they're buring them to keep them from becoming space junk, that would present a traffic hazard to future spacecraft. If you leave them up there, they will run into something eventually. Guaranteed. Good citizens deorbit their sats before they run out of fuel.
Mechanical bug #2: And they won't come down as a rain of molten metal. There isn't anything in a communication satellite solid enough to survive re-entry. That takes hefty chunks of metal or ceramic. Remember how the only chunk of Skylab that came down in one piece was the lead-lined film safe? And they can steer the satellites to re-enter them in the middle of the ocean so there's no hazard even if they did reach ground level.
None of this is rocket science (well, some of it is-- how about "None of this is brain surgery"). I'm not even in the satellite business, and even I can tell these people are making it up as they go along.
I got an aluminum attache case from Zero Halliburton for by birthday, and it is loads of fun. It's the slimline Z2 with silver satin finish. The important thing about aluminum attache cases is that they are too cool for the real world. I've never seen anyone else carry one, but they show up all the time in movies and books where people do dramatic stuff. However, they let me buy one despite not being a fictional character.
It's light and nigh indestructible. My model is too thin for a laptop, but the thinness is important to improve the coolness factor. They're available in all sizes.
Around $200. Do a web search on "zero halliburton".
People have known for years that some visual processes occur in parallel, because they take constant time regardless of the amount of input. For example, if I ask you to pick one red square out of a scattering of many green squares, the time required does not depend on the number of squares. Other tasks require times proportional to the number of objects. For example, finding one red square in a scattering of red circles, green circles, and green squares, is a task requiring time proportional to the number of items you have to sort through. Everybody assumes that this is a serial process. All this has been known for years-- the description of the tasks that can be done in parallel, and hence the properties of the hardware that computes them, was pretty much settled in the late '80s.
No doubt the research reported in this article is important for some reason, because I saw the technical paper it was based on in the most recent issue of Nature, which is a pretty major journal. Unfortunately I don't have it with me, so I can't read the paper and tell you why it is important. Certainly it's not just the fact that some kinds of visual perception are serial.
The person who wrote this article clearly hasn't thought about crossbars very much. It's all about how hard it is to put a big crossbar on a single chip. But the author has apparently never dealt with an actual crossbar. I used to design supercomputers back in the '80s, and we had it even worse than they do nowadays, as far as how many pins you could get on a chip. And yet, we managed to design and even build big crossbars. How did we do it?
We split the data path across multiple chips. The way this applies to the example at hand is to build a chip that can switch among 14 ports, each only 16 bits wide. This will require a failrly reasonable 224 pins on the chip. Then you gang these chips up, with one switching data bits 0 to 15, one switching 16 through 32, etc. To switch a 128-bit-wide bus, you need 8 of these chips. You also need to design a control chip to look at the address and control lines to decide which processor to connect to which memory on each clock cycle. The control chip broadcasts identical switching instructions to all the data chips.
This solution keeps you from having to multiplex pins, and it keeps you from having to build stupendous 1000-pin packages, and it keeps you from having to run busses at 800 MHz and turn your computer into a microwave oven. The only downside is that you need a set of nine chips, but on a motherboard that aleady has eight processors, that's not too bad.
>``The OSR8000 has addressed the market demand for a flexible, high >performance enterprise solution that increases network bandwidth on demand >and preserves mission critical application requirements for enterprise and >service provider networks,'' said Noam Lotan, president and CEO of MRV >Communications. >
Do you think he actually *said* that? Wow. It seems like it would take a commitee to say something like that.
Slashdot Mirror
This whole topic is some combination of genuinely good technology, hype, exaggerations by imperfect journalists, and fraud. From where I'm sitting, I don't know which of the above causes is the true explanation. But I doubt Shannon's law has been repealed. Alas...
I was one of the designers of the CM-5 supercomputer at Thinking Machines Co. back in the '80s. One of the terrible things about designing a supercomputer is that it takes so long, Moore's law has moved the underlying technology out from under you before you're done. We had to throw away a mostly-completed microprocessor design when it became clear that we would be beaten by commodity microprocessors, and re-engineer the rest of the system to use a big pile of commodity SPARC chips instead of a big pile of our custom whiz-bangs. It was an enormous waste and really bad for morale (mine, anyway). So we ended up slipping one generation.
I find it hard to imagine what it must be like to work at Tera. I read Burton Smith's first paper, proposing what has now become the MTA, in 1985. Computing has gotten a thousand times faster since then, and they're still plugging away. Imagine the number of designs they have had to throw away! The've probably had to redesign the system every 18 months, just to stay even, without ever being able to build it, until they managed to deploy one machine, last July. It's an incredibly long slog, espescially for an industry like computers, that rewards sprinters, not sloggers.
Economic bug #1: They're not going to get a tax write-off for this. They're losing money, so they don't pay any taxes anyway. The problem is that they owe more money than they can pay, and a Judge has commanded them to stop running up more debts, and pay the ones they have already. They'll probably end up paying pennies on the dollar.
Economic bug #2: Iridium spent years getting permission to broadcast in 167 separate countries. Are all those foriegn telecommunications agencies going to be happy about handing those permissions to a bunch of geeks that advertises itself as "beyond the reach of any govt"? Fat chance. And without the approval of the local governments, the network won't be allowed to operate.
Mechanical bug #1: They're not burning the satellites for a tax write-off, they're buring them to keep them from becoming space junk, that would present a traffic hazard to future spacecraft. If you leave them up there, they will run into something eventually. Guaranteed. Good citizens deorbit their sats before they run out of fuel.
Mechanical bug #2: And they won't come down as a rain of molten metal. There isn't anything in a communication satellite solid enough to survive re-entry. That takes hefty chunks of metal or ceramic. Remember how the only chunk of Skylab that came down in one piece was the lead-lined film safe? And they can steer the satellites to re-enter them in the middle of the ocean so there's no hazard even if they did reach ground level.
None of this is rocket science (well, some of it is-- how about "None of this is brain surgery"). I'm not even in the satellite business, and even I can tell these people are making it up as they go along.
I got an aluminum attache case from Zero Halliburton for by birthday, and it is loads of fun. It's the slimline Z2 with silver satin finish. The important thing about aluminum attache cases is that they are too cool for the real world. I've never seen anyone else carry one, but they show up all the time in movies and books where people do dramatic stuff. However, they let me buy one despite not being a fictional character.
It's light and nigh indestructible. My model is too thin for a laptop, but the thinness is important to improve the coolness factor. They're available in all sizes.
Around $200. Do a web search on "zero halliburton".
People have known for years that some visual processes occur in parallel, because they take constant time regardless of the amount of input. For example, if I ask you to pick one red square out of a scattering of many green squares, the time required does not depend on the number of squares. Other tasks require times proportional to the number of objects. For example, finding one red square in a scattering of red circles, green circles, and green squares, is a task requiring time proportional to the number of items you have to sort through. Everybody assumes that this is a serial process. All this has been known for years-- the description of the tasks that can be done in parallel, and hence the properties of the hardware that computes them, was pretty much settled in the late '80s.
No doubt the research reported in this article is important for some reason, because I saw the technical paper it was based on in the most recent issue of Nature, which is a pretty major journal. Unfortunately I don't have it with me, so I can't read the paper and tell you why it is important. Certainly it's not just the fact that some kinds of visual perception are serial.
The person who wrote this article clearly hasn't thought about crossbars
very much. It's all about how hard it is to put a big crossbar on a
single chip. But the author has apparently never dealt with an actual
crossbar. I used to design supercomputers back in the '80s, and we had it
even worse than they do nowadays, as far as how many pins you could get on
a chip. And yet, we managed to design and even build big crossbars. How
did we do it?
We split the data path across multiple chips. The way this
applies to the example at hand is to build a chip that can switch among 14
ports, each only 16 bits wide. This will require a failrly reasonable 224
pins on the chip. Then you gang these chips up, with one switching data
bits 0 to 15, one switching 16 through 32, etc. To switch a
128-bit-wide bus, you need 8 of these chips. You also need to design a
control chip to look at the address and control lines to decide which
processor to connect to which memory on each clock cycle. The control
chip broadcasts identical switching instructions to all the data chips.
This solution keeps you from having to multiplex pins, and it keeps you
from having to build stupendous 1000-pin packages, and it keeps you from
having to run busses at 800 MHz and turn your computer into a microwave
oven. The only downside is that you need a set of nine chips, but on a
motherboard that aleady has eight processors, that's not too bad.
--Carl Feynman
>``The OSR8000 has addressed the market demand for a flexible, high
>performance enterprise solution that increases network bandwidth on demand
>and preserves mission critical application requirements for enterprise and
>service provider networks,'' said Noam Lotan, president and CEO of MRV
>Communications.
>
Do you think he actually *said* that? Wow. It seems like it would take a commitee to say something like that.