I wonder how much better efficiency this technology might have in california, nevada, or arizona. In the southwest you're gonna have more sunny days, higher solar intensity, and your energy usage is going to be higher in the summer (air conditioning instead of heating) when the solar energy is more abundant. You couple that with the booming population of pheonix, las vegas, and much of california, and your level of equipment costs might be considerably less.
I also wonder how much of the systems' cost is static, and how much scales with the size of the energy draw. Can I replace 80% of my energy usage for 50% of the cost, or will I have to pay 90% of the cost for a 40% drop in energy usage?
Or shopping mall, tunnels, bridges, etc. There are lots of places where people are vulnerable. The real damage from a terrorist attack isn't killing people, though. It's the fear that is seeded in the rest of the population. Every day in America 500 people die by driving off the road, or falling off roofs. From a statistical perspective, blowing up a plane has a very small effect on the country. From a psychological perspective, it's huge. The airlines, as well as the government, want the US population to FEEL safe.
The hard part is the same in both cases. You need to know when to use the countermeasures, and know quickly. The cost of this system isn't the laser, it's the missile tracking and detection system.
In my office we use telecommuting not to recruit people in different metropolitan areas, but cantidates further away, within our metro area. I usually go into the office 2-3 times a week, and try not to be driving at rush-hour. If I can work at the office 10:00-3:30, a few times a week, and get the rest of my work done remote, that's considered sufficient "face-time". It means I can live a lot further away, and endure the long commute because it's not very often, and not at rush hour.
I don't think we'd put up with complete telecommuting, not unless the employee was phenomenal.
Umm. According to a BLOG. We all know how legitimage and incorruptable those are.
180 days gives paypall time to let the FBI get involved in case this guy and his blog really aren't charity minded friends, but are scam artists trying to make a buck off of someone else's pain. After Katrina the FBI shut down hundreds of fake charities, pretending to offer relief to those affected. It happens, and it's sad.
nope. The Zseries (64bit successor to s/390) still uses a custom processor. It does use powerpc chips to power the I/O channels, and the storage controllers, but they are not the central processors, which have some very demanding requirements. Check out the ars technica and realworldtech articles about power6. There's some evidence that power6 was designed to work in the zseries, though IBM has stated that it will not use power6 in the mainframe. power7 perhapse.
power5+ is dual core, as are power4, power4+, power5, and so too will be power6.
In HPC applications, IBM uses single core power5+ chips, allowing the core to get greater memory and cache bandwidth, and a larger L3 cache, in so much as it need not share this with the other core on the chip.
Low end power5 systems pack 2 power5 chips and 2 L3 cache chips onto a single ceramic MCM, make a pseudo quad-core, though it is logically 2 seperate dual-core processors. High end power5 systems pack 4 power5 chips and 4 L3 cache chips onto a single multichip-module, making it an 8-core module. On the high-end they call them "books".
I agree completely, and will point out that there's a 10billion dollar a year business for mainframe systems to run bizzare ibm system/390 (31bit cisc) and burroughs 2200 (36bit cisc) green-screen applications. IT's not fast, or sexy, but there are tens of thousands of businesses who depend on these legacy systems.
The same, of course, is true in the PC world, where many corporations still depend DOS and 16bit windows applications. If I buy a machine with a whole new architecture, and have to replace all my software, the software and training costs are going to be several times the cost of the hardware.
It would require a huge performance and usability jump before anyone would undertake a change away from x86, and the x86 processors out there are good enough that any advantage a particular architecture might provide is very small. In fact, today's x86 processors are some of the best performing chips available. A better question might be why did we all use x86 processors in the mid-90's. At that time Alpha or Mips processors were often performing at five-ten times the rate of 486 or pentium processors. If the economics of switching didn't make sense back then, they sure don't make any sense now.
"Crew launch vehicles can be made safer at less cost if they aren't also being asked to carry heavy cargo loads" The Ares 1 rocket, which will launch the crew capsule of future moon missions is, by most standards, a heavy launch vehicle. It has a low-earth-orbit payload comparable to the delta IV - Heavy, titan 4, and Atlas 5 Heavy. It is also not a cheap rocket. The Atlas 5 on which this test vehicle will be launched, costs a couple hundred million dollars to launch.
While there are efforts to make space cheaper, I'm not sure that this is one of those. This used to be a join air force & nasa project. Now that Nasa is putting it's bets on CEV and Ares, it's interesting that the air force is funding this alone. Whatever the motive is, it's something military, not cheap-space access. Skip-bombers maybe?
Someone else will probably license Sun's proximity communication technology at some point, but it might be the graphics card makers. The proximity communication stuff lets you hook together multiple chips, almost as if they were part of the same die, using a bunch of capacitor-like plates in the chips. This could be very useful for putting some amount of memory (almost) on the cpu die, and putting a very wide bus between that memory and the processor. Both IBM and Cray currently use very expensive ceramic multichip modules to connect multiple dies together, and they are still somewhat limited in the number of connections that can be attached through the modules.
Apart from that, I don't really know what advances they had. Solaris can scale to 100 processors fairly well, but both IBM and Cray have been working on scalable operating systems for systems with tens of thousands of CPUs. The rock processor would likely be a lot faster than sun's current processors, but it's an incremental advance for microprocessors. Both IBM and Cray are working on more radical technology with FPGAs, vector CPUs, highly treaded designs, and sophisticated coprocessors, and very scalable interconnects.
Interesting notion. For Cray it's a no brainer. They only exist in the upper-end of the HPC market.
For IBM, I'd still say it's a win. If it weren't for some of the early work with blue gene, would there have ever been a cell processor? If their project is more related to power6/7, well those processors are used for the low end of HPC, and for high-end database servers too. Even if the DARPA system requires CPU modifications (VIVA) that don't help the general business user, any advances in memory technology, primary or tertiary storage systems, or networks will help the entire product line. Sun, I'm not sure.
Google's advances are very far from traditional HPC applications like fluid dynamics, weather forcasting, solid body simulations, waveguides, thermal reactions, particle dispersion, oil discovery, etc. Google does data mining, and transactional processing. The very problem that the darpa HPCS program addresses, is that the bulk of the HPC systems sold in the US are just clusters of off-the-sheld database/web-optimized servers. It turns out that these clusters don't deliver very high levels of efficiency, either computationally, or from a power/cooling perspective. Google rolls their own servers, but they still fit into the database/web-optimized server camp. Their software acheivements are all in the data-mining category.
This is not to say that the defense department doesn't need lots of high-end database servers. They use them by the truckload. However, the need for advances in this area are being met by the hardware and software markets. Market forces were not, however, stimulating truly interesting research at the high end of the HPC marketplace. Thus the DoD needed to put together this competition.
cluster, MPP; these two things are really quite similar.
I would phrase it differently: Nobody can approach BG's low cost, both in purchase price, and in TCO. There's lots of scalable systems out there, they just cost too much.
It is worth noting that blue gene and red storm are very similar architectures. Both are 3D torus topologies. Both use powerpc 440 coprocessors to handle communications on the interconnect. Both use a microkernel for compute nodes, and linux on i/o nodes, with lustre as the parallel filesystem. The only tangible difference is the compute node. In blue gene, it's a second 700mhz ppc 440 with 512MB of memory. In Red storm, it's a dual-core 2.4ghz opteron with 4GB of ram. The former uses a lot less electricity, can be packed more densely, and is very inexpensive. The later is faster, and can run a larger problem set on each node.
Well, what Hitachi calls a processor, in the SR11000 series of computers, is actually made up of 8 IBM power processors. They use some special syncronization hardware to make it act like a big vector processor. Thus, if you want the cpu count to actually measure the number of chips, multiply their processor counts by 8. It's sort of the reverse of multi-core.
You are correct that the absolute number of processors does not always indicate how fast the real problem gets solved. For tasks that don't parallelize very well, #1 is going to perform less well than on other systems with a smaller number of faster processors. Of course you have to also account for memory performance, interconnect latency, interconnect bandwidth, mean-time-to-system failure, programming environment, system software, filesystem performance. Linpack is just one measure of system usability.
A lot of what the NSA does is not floating-point math. In all likelihood, most of their needs are data-mining, automatic translation, and other database-intensive applications. I'm sure they have a lot of very expensive computers, but they may not be the kind that end up on the top500 list.
Just today Cray pre-announced the XMT machine a href="http://www.cray.com/products/xmt/" which is the next generation of their machine for graph-tree algorithms. The product line has been basically funded by the NSA. It won't, however, make the top500 list anytime soon.
Well #1 needs a lot of asterixes next to it. The Blue Gene architecture uses an increadible number of relatively underpowered compute nodes, each with relatively little memory, and strings them together into a cluster. It's a system architecture designed around VERY LOW COST. It works quite well for a few problems, but is difficult to use for many real world problems. Because it costs so little to build, those Department of Energy guys with the big pockets can build a VERY fast computer, at least on paper.
#2 is a more general purpose supercomputer, with a better balance of processor count, processor performance, and memory. The DOE spent a LOT of money on this machine, and thus it has a very high level of performance.
After that, you see a mix of high and low efficiency machines, but few people have the can fork over the hundreds of millions of dollars necessary for a machine that powerful. It's all about the $$$.
I'll point out, however, that the Earth Simulator is still ranked #14, 5 years after it came on-line. Of course it also cost hundreds of millions of dollars at the time.
First of all, the gf8800 has the same deficiency that the cell has, in that both are really good at performing single precision floating point math. This is great for video processing and the like, but real science has been using 64bit floats since the mid 70's. It might be hard to convince users that they can get the wrong answer, but it'll be really cheap and really fast.
secondly, the bandwidth to memory is very high, but the amount of addressable memory is very very low. 768MB of memory, divided by 128 processing units means that the entire problem set for each PE needs to fit in 6MB, otherwise you're bottlenecked going to main memory. Game rendering, conveniently tends to reuse a lot of data, and that data compresses very well in memory. Not so with real science data. This is quite analagous to the problems a lot of scientists are having with Blue Gene, which has 256MB of memory available to each PE.
This is not to say that doing HPC computing on the GPU won't happen, it will just be fairly limited in the number of problems that will port well to that environment. For those that do, however, you can't beat the bang for the buck. I suspect that this is mostly for game physics and video transcoding, as those are things that nvidia/amd can sell as an added value. Anything else just doesn't seem to provide much additional revenue, so I can't imagine them putting a lot of effort into supporting it.
you could have surfed over to your local newspaper's webpage and become reasonable informed.
It's not that hard to find out a cantidate's position on ten or fifteen topics. No matter who you elect, they will do something you don't like, but you can get a pretty good idea in fifteen minutes.
You know, as much as we feel good bashing the patent troll, the patent process is really built to protect the little guy from the market Gorillas. Lets say I come up with a really clever design for a widget that is really clever and useful. Because I don't have a supply chain, or much manufacturing capacity, it costs me a thousand bucks a unit to manufacture. Then General Amalgamated Industries sees I'm selling them as fast as I can make them, copies the design, and can build them for a hundred bucks, undercutting me and putting me out of business. The idea of a patent, is that the little guy, if he comes up with something that really is a unique invention, has a short period of years to sell the product, without competing with unlicensed knock-offs. It's really a very progressive idea.
The real challenge with patent law in the IT industry is that the patent laws were written when the pace of invention, and of the market were slower. 15 years without competition is a very, very long time in the computer world. Much moreso than competing designs of steam engine.
That's pretty much how patents work. It costs millions of dollars to sue someone for a patent claim. You have to know that winning the case will net you tens of millions, or it's not worth the legal fees.
The real difference, is that sgi is now out of the graphics card industry, so they don't have to worry about ati coming back and counter-suing them for using some technology Ati has a pattent for.
This would be the normal and expected course for technology patents.
I'm going to have to agree that paper isn't dead anytime soon. The internet is pretty good for transitory, unimportant data. I read the newspaper online. I pay my electric bills. I even submit my taxes online, but not before printing out two copies on paper, and putting one in the firebox, and one in the safety-deposite box. PAper burns at 451 degrees, CDs melt at about 300 degrees, hard drives too. Also: In property disputes, the story goes, possession is 9 / 10ths of the law. Well with data on paper, a good padlock is 9 / 10ths of good data security. Noone in Nigeria is going to hack my filedrawer with some virus on an mp3 file.
Most all of the compositing, editing, and formatting software is for windows. For some very limited sorts of things, you could probably roll your own, but it would take a lot of development time, and probably be hard to use.
IT seems useless, unless we radically change the way we use TV. I've seen quite a lot of high-def video where the end product looks worse than regular definition television. When I'm watching some news reporter talk to me from the whitehouse lawn: regular television shows me a picture of a news reporter in front of an iron fence. HDTV shows me a news-reporter, with smudged makeup and lint on his colar, in front of an iron-fence that has bird droppings on it. Sometimes you don't want the extra detail. At a minimum, it increases the cost of making content that looks good.
Obviously high-resolution images are already made for static content, like magazines. I can imagine a world in which you use you TV to view movies at HDTV resolution, more cheaply made content at TV resolution, video games at HDTV+ resolution, and publishing-style static content at HDTV+++ resolution, all on the same display. You can stream up to HDTV over cable or high-speed wireless, static content gets downloaded to a local cache of some sort, probably on a subscription model.
Slashdot Mirror
I wonder how much better efficiency this technology might have in california, nevada, or arizona.
In the southwest you're gonna have more sunny days, higher solar intensity, and your energy usage is going to be higher in the summer (air conditioning instead of heating) when the solar energy is more abundant. You couple that with the booming population of pheonix, las vegas, and much of california, and your level of equipment costs might be considerably less.
I also wonder how much of the systems' cost is static, and how much scales with the size of the energy draw. Can I replace 80% of my energy usage for 50% of the cost, or will I have to pay 90% of the cost for a 40% drop in energy usage?
Or shopping mall, tunnels, bridges, etc. There are lots of places where people are vulnerable. The real damage from a terrorist attack isn't killing people, though. It's the fear that is seeded in the rest of the population. Every day in America 500 people die by driving off the road, or falling off roofs. From a statistical perspective, blowing up a plane has a very small effect on the country. From a psychological perspective, it's huge. The airlines, as well as the government, want the US population to FEEL safe.
The hard part is the same in both cases. You need to know when to use the countermeasures, and know quickly. The cost of this system isn't the laser, it's the missile tracking and detection system.
In my office we use telecommuting not to recruit people in different metropolitan areas, but cantidates further away, within our metro area. I usually go into the office 2-3 times a week, and try not to be driving at rush-hour. If I can work at the office 10:00-3:30, a few times a week, and get the rest of my work done remote, that's considered sufficient "face-time". It means I can live a lot further away, and endure the long commute because it's not very often, and not at rush hour.
I don't think we'd put up with complete telecommuting, not unless the employee was phenomenal.
Umm. According to a BLOG. We all know how legitimage and incorruptable those are.
180 days gives paypall time to let the FBI get involved in case this guy and his blog really aren't charity minded friends, but are scam artists trying to make a buck off of someone else's pain. After Katrina the FBI shut down hundreds of fake charities, pretending to offer relief to those affected. It happens, and it's sad.
nope. The Zseries (64bit successor to s/390) still uses a custom processor. It does use powerpc chips to power the I/O channels, and the storage controllers, but they are not the central processors, which have some very demanding requirements. Check out the ars technica and realworldtech articles about power6. There's some evidence that power6 was designed to work in the zseries, though IBM has stated that it will not use power6 in the mainframe. power7 perhapse.
power5+ is dual core, as are power4, power4+, power5, and so too will be power6.
In HPC applications, IBM uses single core power5+ chips, allowing the core to get greater memory and cache bandwidth, and a larger L3 cache, in so much as it need not share this with the other core on the chip.
Low end power5 systems pack 2 power5 chips and 2 L3 cache chips onto a single ceramic MCM, make a pseudo quad-core, though it is logically 2 seperate dual-core processors. High end power5 systems pack 4 power5 chips and 4 L3 cache chips onto a single multichip-module, making it an 8-core module. On the high-end they call them "books".
I agree completely, and will point out that there's a 10billion dollar a year business for mainframe systems to run bizzare ibm system/390 (31bit cisc) and burroughs 2200 (36bit cisc) green-screen applications. IT's not fast, or sexy, but there are tens of thousands of businesses who depend on these legacy systems.
The same, of course, is true in the PC world, where many corporations still depend DOS and 16bit windows applications. If I buy a machine with a whole new architecture, and have to replace all my software, the software and training costs are going to be several times the cost of the hardware.
It would require a huge performance and usability jump before anyone would undertake a change away from x86, and the x86 processors out there are good enough that any advantage a particular architecture might provide is very small. In fact, today's x86 processors are some of the best performing chips available. A better question might be why did we all use x86 processors in the mid-90's. At that time Alpha or Mips processors were often performing at five-ten times the rate of 486 or pentium processors. If the economics of switching didn't make sense back then, they sure don't make any sense now.
"Crew launch vehicles can be made safer at less cost if they aren't also being asked to carry heavy cargo loads"
The Ares 1 rocket, which will launch the crew capsule of future moon missions is, by most standards, a heavy launch vehicle. It has a low-earth-orbit payload comparable to the delta IV - Heavy, titan 4, and Atlas 5 Heavy. It is also not a cheap rocket. The Atlas 5 on which this test vehicle will be launched, costs a couple hundred million dollars to launch.
While there are efforts to make space cheaper, I'm not sure that this is one of those. This used to be a join air force & nasa project. Now that Nasa is putting it's bets on CEV and Ares, it's interesting that the air force is funding this alone. Whatever the motive is, it's something military, not cheap-space access. Skip-bombers maybe?
Someone else will probably license Sun's proximity communication technology at some point, but it might be the graphics card makers. The proximity communication stuff lets you hook together multiple chips, almost as if they were part of the same die, using a bunch of capacitor-like plates in the chips. This could be very useful for putting some amount of memory (almost) on the cpu die, and putting a very wide bus between that memory and the processor. Both IBM and Cray currently use very expensive ceramic multichip modules to connect multiple dies together, and they are still somewhat limited in the number of connections that can be attached through the modules.
Apart from that, I don't really know what advances they had. Solaris can scale to 100 processors fairly well, but both IBM and Cray have been working on scalable operating systems for systems with tens of thousands of CPUs. The rock processor would likely be a lot faster than sun's current processors, but it's an incremental advance for microprocessors. Both IBM and Cray are working on more radical technology with FPGAs, vector CPUs, highly treaded designs, and sophisticated coprocessors, and very scalable interconnects.
Interesting notion. For Cray it's a no brainer. They only exist in the upper-end of the HPC market.
For IBM, I'd still say it's a win. If it weren't for some of the early work with blue gene, would there have ever been a cell processor? If their project is more related to power6/7, well those processors are used for the low end of HPC, and for high-end database servers too. Even if the DARPA system requires CPU modifications (VIVA) that don't help the general business user, any advances in memory technology, primary or tertiary storage systems, or networks will help the entire product line. Sun, I'm not sure.
Google's advances are very far from traditional HPC applications like fluid dynamics, weather forcasting, solid body simulations, waveguides, thermal reactions, particle dispersion, oil discovery, etc. Google does data mining, and transactional processing. The very problem that the darpa HPCS program addresses, is that the bulk of the HPC systems sold in the US are just clusters of off-the-sheld database/web-optimized servers. It turns out that these clusters don't deliver very high levels of efficiency, either computationally, or from a power/cooling perspective. Google rolls their own servers, but they still fit into the database/web-optimized server camp. Their software acheivements are all in the data-mining category.
This is not to say that the defense department doesn't need lots of high-end database servers. They use them by the truckload. However, the need for advances in this area are being met by the hardware and software markets. Market forces were not, however, stimulating truly interesting research at the high end of the HPC marketplace. Thus the DoD needed to put together this competition.
cluster, MPP; these two things are really quite similar.
I would phrase it differently: Nobody can approach BG's low cost, both in purchase price, and in TCO. There's lots of scalable systems out there, they just cost too much.
It is worth noting that blue gene and red storm are very similar architectures. Both are 3D torus topologies. Both use powerpc 440 coprocessors to handle communications on the interconnect. Both use a microkernel for compute nodes, and linux on i/o nodes, with lustre as the parallel filesystem. The only tangible difference is the compute node. In blue gene, it's a second 700mhz ppc 440 with 512MB of memory. In Red storm, it's a dual-core 2.4ghz opteron with 4GB of ram. The former uses a lot less electricity, can be packed more densely, and is very inexpensive. The later is faster, and can run a larger problem set on each node.
Well, what Hitachi calls a processor, in the SR11000 series of computers, is actually made up of 8 IBM power processors. They use some special syncronization hardware to make it act like a big vector processor. Thus, if you want the cpu count to actually measure the number of chips, multiply their processor counts by 8. It's sort of the reverse of multi-core.
You are correct that the absolute number of processors does not always indicate how fast the real problem gets solved. For tasks that don't parallelize very well, #1 is going to perform less well than on other systems with a smaller number of faster processors. Of course you have to also account for memory performance, interconnect latency, interconnect bandwidth, mean-time-to-system failure, programming environment, system software, filesystem performance. Linpack is just one measure of system usability.
A lot of what the NSA does is not floating-point math. In all likelihood, most of their needs are data-mining, automatic translation, and other database-intensive applications. I'm sure they have a lot of very expensive computers, but they may not be the kind that end up on the top500 list.
Just today Cray pre-announced the XMT machine a href="http://www.cray.com/products/xmt/" which is the next generation of their machine for graph-tree algorithms. The product line has been basically funded by the NSA. It won't, however, make the top500 list anytime soon.
Well #1 needs a lot of asterixes next to it. The Blue Gene architecture uses an increadible number of relatively underpowered compute nodes, each with relatively little memory, and strings them together into a cluster. It's a system architecture designed around VERY LOW COST. It works quite well for a few problems, but is difficult to use for many real world problems. Because it costs so little to build, those Department of Energy guys with the big pockets can build a VERY fast computer, at least on paper.
#2 is a more general purpose supercomputer, with a better balance of processor count, processor performance, and memory. The DOE spent a LOT of money on this machine, and thus it has a very high level of performance.
After that, you see a mix of high and low efficiency machines, but few people have the can fork over the hundreds of millions of dollars necessary for a machine that powerful. It's all about the $$$.
I'll point out, however, that the Earth Simulator is still ranked #14, 5 years after it came on-line. Of course it also cost hundreds of millions of dollars at the time.
First of all, the gf8800 has the same deficiency that the cell has, in that both are really good at performing single precision floating point math. This is great for video processing and the like, but real science has been using 64bit floats since the mid 70's. It might be hard to convince users that they can get the wrong answer, but it'll be really cheap and really fast.
secondly, the bandwidth to memory is very high, but the amount of addressable memory is very very low. 768MB of memory, divided by 128 processing units means that the entire problem set for each PE needs to fit in 6MB, otherwise you're bottlenecked going to main memory. Game rendering, conveniently tends to reuse a lot of data, and that data compresses very well in memory. Not so with real science data. This is quite analagous to the problems a lot of scientists are having with Blue Gene, which has 256MB of memory available to each PE.
This is not to say that doing HPC computing on the GPU won't happen, it will just be fairly limited in the number of problems that will port well to that environment. For those that do, however, you can't beat the bang for the buck. I suspect that this is mostly for game physics and video transcoding, as those are things that nvidia/amd can sell as an added value. Anything else just doesn't seem to provide much additional revenue, so I can't imagine them putting a lot of effort into supporting it.
you could have surfed over to your local newspaper's webpage and become reasonable informed.
It's not that hard to find out a cantidate's position on ten or fifteen topics. No matter who you elect, they will do something you don't like, but you can get a pretty good idea in fifteen minutes.
Hop to it.
yeah, but the money for the hubble and for its successors come from the same purse. prolonging hubble means delaying the successor.
You know, as much as we feel good bashing the patent troll, the patent process is really built to protect the little guy from the market Gorillas. Lets say I come up with a really clever design for a widget that is really clever and useful. Because I don't have a supply chain, or much manufacturing capacity, it costs me a thousand bucks a unit to manufacture. Then General Amalgamated Industries sees I'm selling them as fast as I can make them, copies the design, and can build them for a hundred bucks, undercutting me and putting me out of business. The idea of a patent, is that the little guy, if he comes up with something that really is a unique invention, has a short period of years to sell the product, without competing with unlicensed knock-offs. It's really a very progressive idea.
The real challenge with patent law in the IT industry is that the patent laws were written when the pace of invention, and of the market were slower. 15 years without competition is a very, very long time in the computer world. Much moreso than competing designs of steam engine.
That's pretty much how patents work. It costs millions of dollars to sue someone for a patent claim. You have to know that winning the case will net you tens of millions, or it's not worth the legal fees.
The real difference, is that sgi is now out of the graphics card industry, so they don't have to worry about ati coming back and counter-suing them for using some technology Ati has a pattent for.
This would be the normal and expected course for technology patents.
I'm going to have to agree that paper isn't dead anytime soon. The internet is pretty good for transitory, unimportant data. I read the newspaper online. I pay my electric bills. I even submit my taxes online, but not before printing out two copies on paper, and putting one in the firebox, and one in the safety-deposite box. PAper burns at 451 degrees, CDs melt at about 300 degrees, hard drives too. Also: In property disputes, the story goes, possession is 9 / 10ths of the law. Well with data on paper, a good padlock is 9 / 10ths of good data security. Noone in Nigeria is going to hack my filedrawer with some virus on an mp3 file.
Most all of the compositing, editing, and formatting software is for windows. For some very limited sorts of things, you could probably roll your own, but it would take a lot of development time, and probably be hard to use.
/
However, you should check out Apple's final cut pro. I've seen it used for small-medium sized TV stations, and it's not too hard to use. I like it.
http://www.apple.com/finalcutstudio/
http://www.apple.com/uk/pro/profiles/tourdefrance
IT seems useless, unless we radically change the way we use TV. I've seen quite a lot of high-def video where the end product looks worse than regular definition television. When I'm watching some news reporter talk to me from the whitehouse lawn: regular television shows me a picture of a news reporter in front of an iron fence. HDTV shows me a news-reporter, with smudged makeup and lint on his colar, in front of an iron-fence that has bird droppings on it. Sometimes you don't want the extra detail. At a minimum, it increases the cost of making content that looks good.
Obviously high-resolution images are already made for static content, like magazines. I can imagine a world in which you use you TV to view movies at HDTV resolution, more cheaply made content at TV resolution, video games at HDTV+ resolution, and publishing-style static content at HDTV+++ resolution, all on the same display. You can stream up to HDTV over cable or high-speed wireless, static content gets downloaded to a local cache of some sort, probably on a subscription model.