I heard Adams speak, in 1984, about the television series and, in particular, the HHG sequences within the series. Those sequences were all hand-animated (since computer graphics were actually expensive then!). He talked about how he saw the initial draft animations, and figured that that they were too complex for an American audience (in keeping with conventional Hollywood wisdom). Then he happened to be in the U.S. (maybe to cut a deal for the series, I forget why) and spotted some teenagers playing video games. That was the time when Xevious, Defender, and Sinistar were all the rage. Adams went back to the studio, he told us, and insisted that they redo the animations to make them more complicated.
Here's hoping the creative/adaptive tam here has the same spirit of humorous overload!
... where his towel was. It was in a large auditorium at the University of California (San Diego). He was talking about his new book, "So Long, and Thanks for All The Fish". It turned out that he didn't know where it was. But about half of the audience happened to have brought theirs, and started waving them about...:-)
The towel story, he said, came from being on an extended vacation in Greece and never having a towel with him when his buddies wanted to go to the beach. The cooler folks always seemed to have theirs along.
Agree agree.
The IDC report assumes that the Linux networking server requires $940 in software, while the Microsoft server only requires $211. I see no mention of the Windows license itself, and I know that most common networking tools for Linux are free. Similarly, there's a mysterious line "Outsourced" that claims a Linux networking server will require $946 of outsourced charging, while the Windows server will require $26. (IDC report, p. 10, Table 3).
On page 12, they account for the higher Linux software and outsourcing (custom software development) cost by acknowledging that many shops build custom management software to tie together existing components, and custom software costs more than off-the-shelf stuff. But there's no mention at all of the cost of the Windows license to run the server. Perhaps they're assuming it's bundled with the hardware?
They also claim that Linux shops tend to buy commercial application software for functionality that is included in Windows. The example they use is a directory server: while Windows 2000 comes bundled with a directory server, they claim that a Linux user would have to buy one from Sun or IBM. But freeware solutions do come bundled with Linux: LDAP, YP, and MySQL are a few. Hence, they're comparing apples and oranges -- both Windows and Linux shops have existing solutions bundled with their OS, but the study only charges the Linux shops for 3rd party software. Furthermore, the poor Linux shop appears to be being charged for both a commercial product (the Software line) and custom development (the Outsourcing line). If either of these were eliminated, the costs of Windows and Linux would be within $200 of one another for this solution; if both were eliminated or the cost of the Windows server license were included, Linux would appear cheaper.
Similarly, the File Sharing description (p. 13) shows that Linux has less downtime and lower cost of acquisition (p. 14); but higher staff training costs (probably true) and much higher staffing costs. The only supporting evidence is "...IDC believes that this is another example of how a relatively new operating environment is unable to offer the same ease of management that is available for an incumbent, well-known operating environment" (p. 13). I smell the fine, crisp smell of bullshit.
If the staffing costs were equalized (assuming that you hire someone at the MCSE level and then spend the extra money to train her, as accounted for in the Training column) then Linux would be head and shoulders above Windows for this application.
I stopped reading there -- the smell of numbers being cooked became overpowering.
The term ``plasma'' is often stretched and abused by the low-temperature community. It is sometimes used to refer to a gas that consists only of ions, or only of electrons, even though the term was originally meant to describe charge-neutral clouds. Some Bose-Einstein condensates consist mainly of ions, since the electromagnetic field can then be used to confine them (so they don't hit the floor of the vacuum chamber that holds them). Colloquially, these clouds of cool ions are often referred to as plasma even though free electrons would rapidly neutralize the ions. Likewise, other physicists have captured clouds of electrons (which are fermions and hence can't directly form a bose-einstein condensate, absent some sort of pairing mechanism) and referred to them as a "pure-electron plasma" despite the fact that the cloud is clearly not charge-neutral.
A gas doesn't have to be fully ionized to be plasma. The transition comes gradually as more and more of the gas is ionized. The crucial parameter is the ratio of the average spacing of the molecules, and something called the "Debye radius", that measures the distance over which charge neutrality holds (that is to say, plasmas are charge-neutral mixes of positively and negatively charged particles; so if you add, say, some extra positive charges to a small region you will attract a cloud of negative charges to cancel it out. The Debye radius tells you the size of the cloud).
To be a plasma, the gas should have many free electrons (or ions) in each Debye length. There could be many more neutrals, just along for the ride, in the same space.
Most molecular gases become more or less fully ionized at around 10,000 degrees Kelvin (give or take a factor of four or so, depending on composition) since that's the temperature at which the collision energy becomes significant compared to valence electron binding energies, so most collisions can make new ions. So anything hotter than that is definitely plasma.
But even a fraction of a percent ionization is often enough to give you the nice bulk behavior of a plasma, because the ionized particles do their thing and drag along the neutral ones by collision. Depending on the density, it's probably reasonable to call the 8,000F (3800K) gases "plasma".
I own the Sony Vaio Z1A. It is a rockin' machine in most respects: built-in wireless with hard "off" switch, nice keyboard and video, 6 hour (nominal) battery life with the standard battery, yadda yadda. Unfortunately it's a (spit) Centrino machine, which means:
No sleep support (fu^H^Hthank you, Intel)
No wireless support (FMH, Intel). Though the miniPCI slot is accessible, it's getting hard to find wireless miniPCI cards that are supported. I got mine from the Dell parts desk, pretending to own one of their late-2002 model laptops and ordering it as a replacement part.
Poor throttling support. ACPI works moderately well on the hacked 2.6.0 kernel I'm running, so I can throttle the CPU -- but the main bus throttled modes don't work. This may be due to Intel weirdness or Linux weirdness, I don't know.
The upshot is that the battery life is effectively halved, and with no sleep capability that's a real pain. Swsusp is not ready for prime time, and anyhow it takes longer than it should.
I like the Vaio line, but given the opportunity for another purchase I would steer FAR away from Scumtrino.
Sprint's PCS cellphone system includes a data connection. It seems to work at about 150 kbps. You get a USB adapter cable that plugs right into the phone (which can then charge off the USB power). The computer sees it as a 150kbps Hayes modem, and you run ppp over it. Ping times are usually in the 100-200 ms range: too slow for good gaming or remote X applications, but quite usable for typing, browsing, etc.
The data connection comes free with their Vision service, which in turn comes free with the larger plans. For about $150 or $200/month you can get enough minutes to keep your phone connected 24/7.
I was rather impressed with the service on a recent road trip (the first time I tried it). If you're in range of a cell phone tower, it might be worth trying as a remote ISP. It's not that fast for the price, but it is completely mobile -- you get the same data service from anywhere in their coverage area.
Unfortunately, Matlab is still a category killer for certain kinds of pipelining. But the various open-source data analysis languages are coming on strong. Perl Data Language, Numeric Python, Octave, R -- they're all worth a look, though at least the first three fit the IDL niche a little better than the MatLab one. I'm not as familiar with R as I probably should be.
Unfortunately, all of 'em (including MatLab) suck if you're working with chunks of data that are bigger than your cache, because you end up pumping stuff out over the main bus.
I posted this somewhere else -- but the problem with Matlab and all the other vectorized languages is that they work in the wrong order to preserve cache. Ideally you want to arrange your instructions so that your list of operations is in the innermost loop, and pixels are in a larger loop (because looping over pixels usually requires you to access RAM, while each pixel's working memory can usually fit in cache). But the list of operations is handled by the slow interpreter, rather than fast compiled code -- so you're hosed.
One day someone will write a properly optimized vector language. PDL started going that way (with lazy evaluation and queued dataflow) but it hasn't materialized.
Are you sure you're testing the chip and not the memory bus? Matlab, IDL, PDL, Numeric Python, and (for all I know) APL all have the same general problem that you try to `vectorize' your instructions to minimize your use of the slow interpreter -- but vectorized order is usually the wrong way to use your machine.
If you have a bunch of steps to do on each of a million pixels, your best bet is to do them all on the current pixel, then advance to the next one -- which keeps everything in cache. But the vectorized languages tend to do the first step to all million pixels, then do the second step to all million pixels, etc. That swaps everything out to RAM every time, so you're running at the main bus fetch/write rate, not at the CPU's clock speed.
The big problem (which, IMAO, was a real issue that got glossed over in the propaganda wars about the launch) was hypervelocity impact if the probe accidentally hit Earth during the Earth-flyby gravitational boost.
Cassini, if I recall right, was to go inward to Venus for a gravitational assist, then fly by Earth again for another boost before leaving for the outer solar system. Because the trajectory was only marginally possible to begin with, they had to come rather deep in the gravitational well -- only 200 or 300 miles above the top of the Earth's atmosphere.
During that flyby, Cassini was traveling well above Earth's escape velocity of 10 km/sec. I never saw anyone seriously claim that the plutonium would have remained contained in case of impact.
NASA's response to that point was, essentially, "We don't hit planets by mistake". That was good enough to avoid the various court orders and injunctions that were being cooked up, but it might not suffice today. A few months after the Cassini flyby, NASA (or JPL or Lockheed, depending on whom you ask) did hit a planet by mistake, when the mars probe impacted instead of aerobraking.
On the other hand, the protestors' argument that there was enough plutonium on board to kill half of the Earth's population, if properly distributed, is sheer alarmism. Almost every Slashdot reader generates weekly enough of a certain other substance to, if properly distributed, impregnate half of the Earth's population. Yet only a tiny fraction of children are descended from slashdotters.
I worked at the General Atomic D3D facility in San Diego, the 1980s. The biggest limitation on the rate at which they could explore the experimental parameter space was the number of neutrons that the machine would create. The ultimate end of all modern tokamaks is to be turned into low-level radioactive waste when the machine itself becomes activated by the free neutrons liberated by the fusion process.
The more conventional gamma rays, alpha radiation (helium nucleii), and beta rays (fast moving electrons) are dangerous enough but at least they aren't infectious: you can irradiate food with gamma rays and it doesn't turn radioactive. Neutrons get absorbed by nearby nuclei, which then themselves become unstable and radioactive. Ick.
That's not to say we shouldn't explore nuclear fusion as a power source -- just that it is not the perfectly clean energy source that it is often made out to be.
I've been running 2.6.0-test5 on my laptop and one of two dual-processor "cycle hogs", and it has been rock solid for the magnetohydrodynamic simulations and telescope data reduction I've been working on. Then again, number crunching doesn't really exercise the more esoteric stuff like ACPI support.
It's not hard in principle. As they say, ``the Devil's in the details''. You've got a very hot, combusting mixture under high pressure, right next to large tanks of explosive rocket fuel, and everything has to be light, light, light to fly well -- so you use the lightest, thinnest metal you think you can get away with. And, of course, the metal has to operate at much higher temperatures than you normally encounter, and still have enough strength to avoid blowing up during thrust.
If the rocket didn't have to fly, you could just put loads of engineering margin into every part, and end up with something big and heavy but reliable. But you can't, because "big and heavy" won't get off the ground.
The sheer amount of power that has to converted from chemical to mechanical energy is staggering. In a liquid-fueled rocket engine, you have to push fuel into the chamber against the pressure of combustion. That turns out to be very hard, since you have to move a LOT of fuel and the pressure has to be HIGH for good efficiency. Just the pumping requires a major engineering effort to handle the power required to drive the pumps.
If you have cryogenic liquid propellants (the most efficient for tankage), you have all kinds of material-science problems from the temperature extremes. If you fly less exotic materials, like nitrous oxide, you have less mass margin because the tank is heavier.
Then there are all kinds of weird pitfalls like uneven distribution within the combustion chamber; uneven fuel/oxidiser mixing; choked fuel flow; accumulation of large volumes of fuel mix (which have an alarming tendency to explode later if they don't burn instantly); quenching of the burn by the amazing volume of stuff flowing into the chamber; eddies and cavitation in the turbulent flow out the throat of the engine; detonation (makes your car engine knock, makes your rocket explode); things shaking loose because of the engine's vibration; the nozzle itself starting to combust, ablate, or burn-through; and making a poorly designed nozzle that limits your thrust.
None of those things is unsurmountable -- it's having to get everything more or less right the first time that is the real kicker.
Because, unlike any of the JIT or scripting languages, Lisp requires turning your head inside out to use it. I like evaluative syntax as much as the next guy, and it's certainly the right tool for lots of things -- but using it requires a mindset that appears only to exist in a tiny fraction of the hacker population, sort of like an RPN calculator. Those who could adapt to RPN really, really loved it; but those who couldn't (which, unfortunately, was most people) found it cumbersome. So, while RPN was arguably better than the expression syntax that calculators use now, it faded away due to lack of mass appeal.
No, actually, it's not nearly so bad. The point is that business users will generally get more hyped up about minor glitches in service, and require more attention, than ``recreational'' users. Same story with telephone lines. Businesses use the phone differently than households, so businesses get charged more for a phone line. (Never mind that your teenage daughter spends six hours a day on the darned thing...)
Speakeasy rocks. Everyone should send them a wheelbarrow of money.
I don't know about "realtime". Audacity is an interactive sound editor that, I believe, has a GUI pitch-shifter. PDL can certainly shift pitches on existing files (just fft, pad or clip, and fft back), but the interface is a little lower level than you seem to be looking for.
No standard audio CDs contain digital author or title information. It was an oversight or stupidity on the part of the people who invented the CD standard. When you fire up Grip or WMP and pop in your CD, the track listing you get is retrieved from a database, based on a hash made from (among other things) the precise length of each track that is found.
I gave my family a vanity site in 1997. It was fun to set up, but then it got boring and then it sat, idle, while my youngest brother grew up. You have to be a little weird (like most of the slashdot crowd) and have a little time on your hands to maintain a personal web site. Most people aren't weird enough and/or don't have the time.
Actually, essentially all of it is turned into heat in the room (except for the microscopic amount of power that gets sent out of the room as Ethernet pulses). Yep, technically 1 is "at least 1/3", but the second law of thermodynamics is too oft forgotten.
When I moved into my dorm room at Stanford, nearly 15 years ago, I was shocked to discover that the university imposed a surcharge per quarter for the power used by certain appliances -- e.g. if you brought a mini-fridge or a microwave, you were were supposed to pay an extra $10/month or something to account for the power you used. The catch? The dorms were heated with ELECTRICAL HEATERS. Hence it was practically impossible to waste electricity in the cool half of the year -- using your fridge or microwave would just reduce the duty cycle of the elctric heater...
That's not an aircraft... it's a spacecraft.
on
Son of Concorde
·
· Score: 1
Tokyo to Paris in 2 hours is over a third of orbital speed. Going that fast would require getting 'way above what we normally consider "atmosphere" and skimming the boundary of space.
Slashdot Mirror
Heh. Small world. On the other hand, I can't be the only person ever to ask him that...
Here's hoping the creative/adaptive tam here has the same spirit of humorous overload!
The towel story, he said, came from being on an extended vacation in Greece and never having a towel with him when his buddies wanted to go to the beach. The cooler folks always seemed to have theirs along.
On page 12, they account for the higher Linux software and outsourcing (custom software development) cost by acknowledging that many shops build custom management software to tie together existing components, and custom software costs more than off-the-shelf stuff. But there's no mention at all of the cost of the Windows license to run the server. Perhaps they're assuming it's bundled with the hardware?
They also claim that Linux shops tend to buy commercial application software for functionality that is included in Windows. The example they use is a directory server: while Windows 2000 comes bundled with a directory server, they claim that a Linux user would have to buy one from Sun or IBM. But freeware solutions do come bundled with Linux: LDAP, YP, and MySQL are a few. Hence, they're comparing apples and oranges -- both Windows and Linux shops have existing solutions bundled with their OS, but the study only charges the Linux shops for 3rd party software. Furthermore, the poor Linux shop appears to be being charged for both a commercial product (the Software line) and custom development (the Outsourcing line). If either of these were eliminated, the costs of Windows and Linux would be within $200 of one another for this solution; if both were eliminated or the cost of the Windows server license were included, Linux would appear cheaper.
Similarly, the File Sharing description (p. 13) shows that Linux has less downtime and lower cost of acquisition (p. 14); but higher staff training costs (probably true) and much higher staffing costs. The only supporting evidence is "...IDC believes that this is another example of how a relatively new operating environment is unable to offer the same ease of management that is available for an incumbent, well-known operating environment" (p. 13). I smell the fine, crisp smell of bullshit.
If the staffing costs were equalized (assuming that you hire someone at the MCSE level and then spend the extra money to train her, as accounted for in the Training column) then Linux would be head and shoulders above Windows for this application.
I stopped reading there -- the smell of numbers being cooked became overpowering.
The parent article is describing a superheated froth, not a plasma -- which is just an ionized gas.
The term ``plasma'' is often stretched and abused by the low-temperature community. It is sometimes used to refer to a gas that consists only of ions, or only of electrons, even though the term was originally meant to describe charge-neutral clouds. Some Bose-Einstein condensates consist mainly of ions, since the electromagnetic field can then be used to confine them (so they don't hit the floor of the vacuum chamber that holds them). Colloquially, these clouds of cool ions are often referred to as plasma even though free electrons would rapidly neutralize the ions. Likewise, other physicists have captured clouds of electrons (which are fermions and hence can't directly form a bose-einstein condensate, absent some sort of pairing mechanism) and referred to them as a "pure-electron plasma" despite the fact that the cloud is clearly not charge-neutral.
No, actually, the transition to plasma behavior is not sharp. It's gradual, as the ionization fraction increases.
To be a plasma, the gas should have many free electrons (or ions) in each Debye length. There could be many more neutrals, just along for the ride, in the same space.
Most molecular gases become more or less fully ionized at around 10,000 degrees Kelvin (give or take a factor of four or so, depending on composition) since that's the temperature at which the collision energy becomes significant compared to valence electron binding energies, so most collisions can make new ions. So anything hotter than that is definitely plasma.
But even a fraction of a percent ionization is often enough to give you the nice bulk behavior of a plasma, because the ionized particles do their thing and drag along the neutral ones by collision. Depending on the density, it's probably reasonable to call the 8,000F (3800K) gases "plasma".
The upshot is that the battery life is effectively halved, and with no sleep capability that's a real pain. Swsusp is not ready for prime time, and anyhow it takes longer than it should.
I like the Vaio line, but given the opportunity for another purchase I would steer FAR away from Scumtrino.
The data connection comes free with their Vision service, which in turn comes free with the larger plans. For about $150 or $200/month you can get enough minutes to keep your phone connected 24/7.
I was rather impressed with the service on a recent road trip (the first time I tried it). If you're in range of a cell phone tower, it might be worth trying as a remote ISP. It's not that fast for the price, but it is completely mobile -- you get the same data service from anywhere in their coverage area.
Unfortunately, all of 'em (including MatLab) suck if you're working with chunks of data that are bigger than your cache, because you end up pumping stuff out over the main bus.
One day someone will write a properly optimized vector language. PDL started going that way (with lazy evaluation and queued dataflow) but it hasn't materialized.
If you have a bunch of steps to do on each of a million pixels, your best bet is to do them all on the current pixel, then advance to the next one -- which keeps everything in cache. But the vectorized languages tend to do the first step to all million pixels, then do the second step to all million pixels, etc. That swaps everything out to RAM every time, so you're running at the main bus fetch/write rate, not at the CPU's clock speed.
Cassini, if I recall right, was to go inward to Venus for a gravitational assist, then fly by Earth again for another boost before leaving for the outer solar system. Because the trajectory was only marginally possible to begin with, they had to come rather deep in the gravitational well -- only 200 or 300 miles above the top of the Earth's atmosphere.
During that flyby, Cassini was traveling well above Earth's escape velocity of 10 km/sec. I never saw anyone seriously claim that the plutonium would have remained contained in case of impact.
NASA's response to that point was, essentially, "We don't hit planets by mistake". That was good enough to avoid the various court orders and injunctions that were being cooked up, but it might not suffice today. A few months after the Cassini flyby, NASA (or JPL or Lockheed, depending on whom you ask) did hit a planet by mistake, when the mars probe impacted instead of aerobraking.
On the other hand, the protestors' argument that there was enough plutonium on board to kill half of the Earth's population, if properly distributed, is sheer alarmism. Almost every Slashdot reader generates weekly enough of a certain other substance to, if properly distributed, impregnate half of the Earth's population. Yet only a tiny fraction of children are descended from slashdotters.
I worked at the General Atomic D3D facility in San Diego, the 1980s. The biggest limitation on the rate at which they could explore the experimental parameter space was the number of neutrons that the machine would create. The ultimate end of all modern tokamaks is to be turned into low-level radioactive waste when the machine itself becomes activated by the free neutrons liberated by the fusion process.
The more conventional gamma rays, alpha radiation (helium nucleii), and beta rays (fast moving electrons) are dangerous enough but at least they aren't infectious: you can irradiate food with gamma rays and it doesn't turn radioactive. Neutrons get absorbed by nearby nuclei, which then themselves become unstable and radioactive. Ick.
That's not to say we shouldn't explore nuclear fusion as a power source -- just that it is not the perfectly clean energy source that it is often made out to be.
I've been running 2.6.0-test5 on my laptop and one of two dual-processor "cycle hogs", and it has been rock solid for the magnetohydrodynamic simulations and telescope data reduction I've been working on. Then again, number crunching doesn't really exercise the more esoteric stuff like ACPI support.
If the rocket didn't have to fly, you could just put loads of engineering margin into every part, and end up with something big and heavy but reliable. But you can't, because "big and heavy" won't get off the ground.
The sheer amount of power that has to converted from chemical to mechanical energy is staggering. In a liquid-fueled rocket engine, you have to push fuel into the chamber against the pressure of combustion. That turns out to be very hard, since you have to move a LOT of fuel and the pressure has to be HIGH for good efficiency. Just the pumping requires a major engineering effort to handle the power required to drive the pumps.
If you have cryogenic liquid propellants (the most efficient for tankage), you have all kinds of material-science problems from the temperature extremes. If you fly less exotic materials, like nitrous oxide, you have less mass margin because the tank is heavier.
Then there are all kinds of weird pitfalls like uneven distribution within the combustion chamber; uneven fuel/oxidiser mixing; choked fuel flow; accumulation of large volumes of fuel mix (which have an alarming tendency to explode later if they don't burn instantly); quenching of the burn by the amazing volume of stuff flowing into the chamber; eddies and cavitation in the turbulent flow out the throat of the engine; detonation (makes your car engine knock, makes your rocket explode); things shaking loose because of the engine's vibration; the nozzle itself starting to combust, ablate, or burn-through; and making a poorly designed nozzle that limits your thrust.
None of those things is unsurmountable -- it's having to get everything more or less right the first time that is the real kicker.
Speakeasy rocks. Everyone should send them a wheelbarrow of money.
I don't know about "realtime". Audacity is an interactive sound editor that, I believe, has a GUI pitch-shifter. PDL can certainly shift pitches on existing files (just fft, pad or clip, and fft back), but the interface is a little lower level than you seem to be looking for.
No standard audio CDs contain digital author or title information. It was an oversight or stupidity on the part of the people who invented the CD standard. When you fire up Grip or WMP and pop in your CD, the track listing you get is retrieved from a database, based on a hash made from (among other things) the precise length of each track that is found.
We're in an exciting time when many of the scripting languages are being augmented to be able to handle Real Data (Numpy is another example).
I gave my family a vanity site in 1997. It was fun to set up, but then it got boring and then it sat, idle, while my youngest brother grew up. You have to be a little weird (like most of the slashdot crowd) and have a little time on your hands to maintain a personal web site. Most people aren't weird enough and/or don't have the time.
Actually, essentially all of it is turned into heat in the room (except for the microscopic amount of power that gets sent out of the room as Ethernet pulses). Yep, technically 1 is "at least 1/3", but the second law of thermodynamics is too oft forgotten.
When I moved into my dorm room at Stanford, nearly 15 years ago, I was shocked to discover that the university imposed a surcharge per quarter for the power used by certain appliances -- e.g. if you brought a mini-fridge or a microwave, you were were supposed to pay an extra $10/month or something to account for the power you used. The catch? The dorms were heated with ELECTRICAL HEATERS. Hence it was practically impossible to waste electricity in the cool half of the year -- using your fridge or microwave would just reduce the duty cycle of the elctric heater...
Tokyo to Paris in 2 hours is over a third of orbital speed. Going that fast would require getting 'way above what we normally consider "atmosphere" and skimming the boundary of space.