Taking the opportunity for a moment to troll, flame bait and be an annoying Apple user, I think it's worth commenting how piss-poor the P4's LinPack performance is.
The AltiVec processor on the G5's is a vector coprocessor. If your compiler/library is set up to use it, that's good for a 4-5x increase in floating-point speed. Essentially the CPU does a block of mathematical operations in parallel--Cray mainframes work the same way, only more so. This is different from pipelining in that it's a true parallel operation. I think the AltiVec can do vector integer operations as well, but that won't change the LinPack performance.
Note too that the boost from a vector processor only works on specific types of floating point operations, most notably matrix math, so it's not a magic cure-all. Also, the data has to be in the right format and loaded into appropriate registers, so it helps to have code written specifically to use vector operations (although a good optimizing compiler can still do a lot of the work for you)
Since the machines can be reprogrammed it has a lot to do with the tech.
If they were just the old style punch cards sitting on a table and someone altered prior to voting starting for the day then it would be a bit obvious when they were passed out to the voters be the election judges.
Stacks of punch cards sitting on a table are an open invitation to ballot-box stuffing. Just grab a handful, mark them all for your preferred candidate, and then [the only hard part] stuff the ballots into the ballot boxes before the votes are counted. There are steps top revent these things, of course, (e.g., official observers from both parties present at all times), but vote fraud still happens.
Second, I doubt you can write any clear rules that will not penalize some parties. Say you have a rule that primaries must be held at the same day in all states. Then how about parties that are too small to have the resources to do so?
Elections are run (and funded) by the state governments, not the individual parties. That's why the state gets to set the rules about how primaries and caucuses are run and how to avoid penalizing small parties.
4. Core files on Linux are set to 0 bytes by default. The only reason this would have been changed, is so that you can debug crashes. If this is the case, Mainsoft was porting Windows applications to Linux as well as Unix.
This depends on the distribution. The system default is easily changed
3. Core files don't contain "lists of files or directories" on Linux. That information is completely irrelevant to the purpose of a core file... diagnosing the reason for a crash. Lists of files in a directory or on the filesystem are completely irrelevant to WHY the app involved crashed.
They don't always, but they can. If you have, say, an opendir() directory handle that hasn't been closed, that directory will be sitting in memory somewhere. Dump core, and it will be written out with everything else.
I remember a few times, way back in the day, where some FTP daemons would leave you a copy of the local shadow password file in the core file if you could overflow any random buffer. The programmer had dropped root privledges by the point the overflow occurred, but they had forgotten to call endpwent(), so the information was still in memory. Rather elegant attack, actually.
No-one _needs_ Disney movies: we need to eat, we need to drink, we don't _need_ to watch 'Disney's Little Aardvark'... any demand that exists is pure hype and marketing. You're making the fatal mistake of assuming that just because Disney have a monopoly on Disney movies, that people don't have a choice as to whether they watch them or not.
No need to throw the Fortran libraries away, though, just wrap them in a higher level language. Chances are it'll be fast enough, and it'll almost certainly be a lot easier to use.
Why bother with a wrapper? I've never seen a Fortran compiler that didn't spit out standard object files. Just call the Fortran code directly. All you have to know is whether your compiler adds underscores before function names (some do, some don't) and that the usual Fortran calling convetion is to pass non-aliased pointers.
One simple #include "foo.h" for convenience and you're in business.
Points are Points, there are 72 of them in an inch.
Except there aren't. True typography points
are 0.013838 inches/point (or roughly 72.26478 points/inch) and not 1/72 inch. This is true for the US and UK. The rest of Europe, naturally, uses yet another size point (Didot points, 0.376 mm or about 0.01476 points per US inch).
Postscript, from its beginning back some time in the 1980's, however, has used points that were by definition 1/72 inch and that usage has become fairly standard for computer applications.
The TeXbook has a very
good description of the whole point mess.
Actually, there is a good possibility that plasma is not a new state of matter per se, but rather a transference state between gas and Bose-Einstienian condensate... much as water at boiling point. Although as we push to further thermal extremes, it's possible that we'll discover more energy states or methods of creating different forms of matter without relying solely on temperature.
Well, no.
First, Bose-Einstein condensates only occur at low temperatures when a significant fraction of the atoms sit in the ground state of the system. Second, Bose-Einstein condensates only occur in Bosonic systems. Particles that obey Fermi-Dirac statistics (e.g., ionized Hydrogen) are forbidden by the rules of quantum mechanics from forming Bose-Einstein condensates. There is no way that plasmas could be a precursor to Bose-Einstein condensates.
Here's where education is important. Do you understand what "going critical" is? Very specifically, it's a build up of heat from a "melt-down".
Well, no. "Going critical" means no more than you have initiated a self-sustaining reaction. A controlled critical reaction is routinely used for power generation. Uncontrolled critical reactions are used in weapons.
How do you know that the shape of the universe does not include holes?
We don't know that. There's nothing in physics to rule out topological holes, but the solutions to GR are so messy that no one really wants to go there (and there's no compelling reason to expect the univers has a a topologocal hole).
Because something works (which is contestable, pending evidence) does not mean it is the best tool for the job. The pragmatic fallacy is alluring, but it must be avoided:)
When my life is on the line, if I know the item works then it is the best tool for the job.
Also check out its companion Dark Sun Rising, also by Richard Rhoads, which is about the making of the hydrogen bomb, but also gets into a number of other historically relevant topics (such as Teller getting Oppenhiemer's security clearance yanked because he was insufficiently in favor of thermonuclear weapons development).
The CWRU cash fund is supposedly something in the billions (though this may just be heresay), and I can't imagine that CFN really sucked that many resources out. SO...
That's the endowment (it passed the $1B mark in the mid- to late-90's when I was in grad school there).
On the other hand, as anyone who's been to the school know, this has nothing to do with public access, nothing to do with student access, and everything to do with marking to prospective students. This is, after all, the university that upgraded its entire network (with great fanfare and major user-level headaches) to ATM to the desktop, and then put a firewall up with a pair of 10Mb/s network cards through which all university traffic had to pass in between our OC-3 network connection and the on-campus ATM network since that's all the IT folks could get money for that sesmester (firewalls don't generate headlines, ATM does).
there is nothing wrong with his pronunciation; it is infact the first (ie preferred) one.
I'm not sure what dictionary you're looking in, but shouldn't it be either gi-ga or ji-ja? After all, in Greek they're both "gamma" [which, I note, is also a hard g], so the two g's should be pronounced the same way in both syllables, no?
How finely split can the spectrum be before there's danger of overlap? Is it possible for, say, one phone to send signals at 800.0001 MHz while another does so at 800.0002 MHz? Where is the precision cutoff for neighboring frequencies before things start to interfere?
It depends how much noise is present in the communications channel and how much bandwidth (i.e., what data rate) your application requires. Look at Claude Shannon's classic paper A Mathematical Theory of Communication for a complete treatment--he developed information theory to answer precisely this question.
But if you want a versitile, general-purpose supercomputer, why not go with the clustering solution?
Because some problems don't work on clusters--things like large-scale molecular dynamcis simulations with long-range spatial interactions.
Problems that require the nodes to share massive
amounts of data between nodes (gigabytes per second and up--these problems often have N^2 behaviors) don't do so well on a cluster since they tend to saturate the network. A shared-memory system, like a supercomputer, on the other hand, can provide much better memory access times (top of the line Cray's have a peak memory transfer rate of 204 GB per sec per node [yes, 204 gigabytes per second]) and since there's only one copy of the memory, there can often be a lower peak bandwidth requirement.
In short, it all depends on the problem you need to solve. Some problems work very well on clusters, others do not.
A brain wave pattern called P300 ( "positive wave" 300 ms after onset ) has been discovered to be activated when a person looks at a familiar object ( the P300 hypothesis has gained very solid evidence since a few years).
These are extremely controversial findings. I've had discussions with cognitive neuroscience researchers I work with and not one of them is totally convinced this is real yet.
Near-infrared neural imaging is already a very useful technique.
This is a new application whether or not you approve.
I work with some of the authors you cited and I personally know most of the others. Near infrar-red spectroscopy and diffuse optical tomography are very good for measuring localized changes in blood oxygenation and volume (and, by inference, brain activation), but going from "you show activation in your right frontal cortex" to "you're lying" is a huge leap and way beyond the limits of any neuroscience research I've ever run across.
This is a submarine patent-they've got an idea, but I assure you they don't know how to make it work yet. First, collecting the signals will be extremely difficult since you have to get past all the light reflected off the skin. The changes in signal level you're looking for are easily at the PPB (yes, 1e-9) level compared with the light you're putting in, so this is decidedly non-trivial. Second, this "information processing device capable of determining various physiological characteristics" is going to be real interesting to build since no one knows what those characteristics are and (as with conventional polygraph) there are innocent explanations for most everything you can measure. Finally, NIRS doesn't penetrate very far through the brain, so you're limited to areas near the skull. Deep structures (eg, the hippocampus) aren't probed at all. If the brain doesn't do any work in these superficial regions (which may or may not be the case), there's nothing to be measured.
In short, you have nothing to worry about from this patent. Worry instead why the government is so interested all of a sudden in an undetectable remotely operated lie detector. We've had visitors from NIMA in our lab, and various three-letter agencies have sent letters to several of my cognitive science colleagues offering to fund research along these lines as well.
It's like the end of Jurassic Park, when the kids and the scientists are all cornered by the Velociraptors, and suddenly, the T-Rex comes along, and tries to eat one of the Velociraptors, and then the other two attack the T-Rex, and the humans escape.
Except that in the book, the velociraptors manage to get off the island and begin migrating through Costa Rica and living off of lysine rich foods.
We're still in the early evolution of megacorporations.
One hundred years ago, technologies (railroad and telegraph) enabled a few companies to go national. Now, technology (internet and routine intercontinental flights) lets companies go multi-national. Megacorporations aren't evolving, it's the same thing as the robber barons all over again. Only the names have changed.
-JS
Those who fail to learn the lessons of history are doomed to repeat them...
and ignorant. So, I am asking the question. What about few thousands RFID tags in luggages and litterally stuffed in all goods in a not so far future?
RFID tags (the retail versions) are passive - they only emit RF when scanned by an external RF source. Without an external transmitter (which also needs to be quite strong, located nearby, or both), there's no signal coming from the tags.
The article states that RF devices may induce currents in airplane wiring. I'm not sure how much commercial airliners cost, but I know it's probably well over 50 million. How much would it cost to use fiber optics instead of twisted-pair wiring? I'm sure relative to the cost of the airplane, it wouldn't be much, and that would eliminate having interference with the wires that must be run throughout the aircraft.
You have two choices. You can fly in:
a plane where every system has been individually checked, has redundant fail-safe systems, and then confirmed by 30+ years of flying, or
a plane using fancy high-tech fiber network to replace 90% of the internal communications systems, using components that will be obsolete (and thus hard to replace) in 5 years, requires a year of instruction before anyone can service it, and introduces a whole new set of failure mechanisms into the system (e.g., how robust is your router to multiple cosmic ray hits)?
I'll take old-fashioned copper wires, thank you very much.
I realize your joking, but in a twisted sort of way, AT&T Unix is the troff operating system. The original unix killer-app was ROFF and its derivatives (including troff). I've always assumed that that's why the man pages are still done in troff--it was the obvious choice originally and there's been too big an investment (in man-hours) to switch to anything else.
I cannot speak for later versions of Windows since I stopped using them, but I never saw a version of windows that does not force you to completely log off and back on to access privileged functions, encouraging people to run with privileges on all the time, because they cannot just enter the password for privileged activities. Su does not exist, nor does sudo.
No, this is a good thing--it forces you to stop and think about what you're doing. Besides, the only time you should ever need to be logged in as administrator (or root) is to install software for system-wide use or to update device drivers.
The real problem is that most Windows software is badly written and requires you to have administrator-level access to do anything useful (including most of the Microsoft Office suite).
Slashdot Mirror
The AltiVec processor on the G5's is a vector coprocessor. If your compiler/library is set up to use it, that's good for a 4-5x increase in floating-point speed. Essentially the CPU does a block of mathematical operations in parallel--Cray mainframes work the same way, only more so. This is different from pipelining in that it's a true parallel operation. I think the AltiVec can do vector integer operations as well, but that won't change the LinPack performance.
Note too that the boost from a vector processor only works on specific types of floating point operations, most notably matrix math, so it's not a magic cure-all. Also, the data has to be in the right format and loaded into appropriate registers, so it helps to have code written specifically to use vector operations (although a good optimizing compiler can still do a lot of the work for you)
.-JS
Stacks of punch cards sitting on a table are an open invitation to ballot-box stuffing. Just grab a handful, mark them all for your preferred candidate, and then [the only hard part] stuff the ballots into the ballot boxes before the votes are counted. There are steps top revent these things, of course, (e.g., official observers from both parties present at all times), but vote fraud still happens.
-JS
Elections are run (and funded) by the state governments, not the individual parties. That's why the state gets to set the rules about how primaries and caucuses are run and how to avoid penalizing small parties.
-JS
In reverse order...
This depends on the distribution. The system default is easily changed
They don't always, but they can. If you have, say, an opendir() directory handle that hasn't been closed, that directory will be sitting in memory somewhere. Dump core, and it will be written out with everything else.
I remember a few times, way back in the day, where some FTP daemons would leave you a copy of the local shadow password file in the core file if you could overflow any random buffer. The programmer had dropped root privledges by the point the overflow occurred, but they had forgotten to call endpwent(), so the information was still in memory. Rather elegant attack, actually.
-JS
Ask me how I know you don't have kids...
-JS
Why bother with a wrapper? I've never seen a Fortran compiler that didn't spit out standard object files. Just call the Fortran code directly. All you have to know is whether your compiler adds underscores before function names (some do, some don't) and that the usual Fortran calling convetion is to pass non-aliased pointers. One simple #include "foo.h" for convenience and you're in business.
-JS
Except there aren't. True typography points are 0.013838 inches/point (or roughly 72.26478 points/inch) and not 1/72 inch. This is true for the US and UK. The rest of Europe, naturally, uses yet another size point (Didot points, 0.376 mm or about 0.01476 points per US inch). Postscript, from its beginning back some time in the 1980's, however, has used points that were by definition 1/72 inch and that usage has become fairly standard for computer applications.
The TeXbook has a very good description of the whole point mess.
-JS
Well, no.
First, Bose-Einstein condensates only occur at low temperatures when a significant fraction of the atoms sit in the ground state of the system. Second, Bose-Einstein condensates only occur in Bosonic systems. Particles that obey Fermi-Dirac statistics (e.g., ionized Hydrogen) are forbidden by the rules of quantum mechanics from forming Bose-Einstein condensates. There is no way that plasmas could be a precursor to Bose-Einstein condensates.
-JS (Yes, IAAP)
Well, no. "Going critical" means no more than you have initiated a self-sustaining reaction. A controlled critical reaction is routinely used for power generation. Uncontrolled critical reactions are used in weapons.
-JS
We don't know that. There's nothing in physics to rule out topological holes, but the solutions to GR are so messy that no one really wants to go there (and there's no compelling reason to expect the univers has a a topologocal hole).
-JS (yes, IAAP)
When my life is on the line, if I know the item works then it is the best tool for the job.
-JS
-JS
That's the endowment (it passed the $1B mark in the mid- to late-90's when I was in grad school there).
On the other hand, as anyone who's been to the school know, this has nothing to do with public access, nothing to do with student access, and everything to do with marking to prospective students. This is, after all, the university that upgraded its entire network (with great fanfare and major user-level headaches) to ATM to the desktop, and then put a firewall up with a pair of 10Mb/s network cards through which all university traffic had to pass in between our OC-3 network connection and the on-campus ATM network since that's all the IT folks could get money for that sesmester (firewalls don't generate headlines, ATM does).
-JS
-JS
It depends how much noise is present in the communications channel and how much bandwidth (i.e., what data rate) your application requires. Look at Claude Shannon's classic paper A Mathematical Theory of Communication for a complete treatment--he developed information theory to answer precisely this question.
-JS
Because some problems don't work on clusters--things like large-scale molecular dynamcis simulations with long-range spatial interactions.
Problems that require the nodes to share massive amounts of data between nodes (gigabytes per second and up--these problems often have N^2 behaviors) don't do so well on a cluster since they tend to saturate the network. A shared-memory system, like a supercomputer, on the other hand, can provide much better memory access times (top of the line Cray's have a peak memory transfer rate of 204 GB per sec per node [yes, 204 gigabytes per second]) and since there's only one copy of the memory, there can often be a lower peak bandwidth requirement.
In short, it all depends on the problem you need to solve. Some problems work very well on clusters, others do not.
-JS
These are extremely controversial findings. I've had discussions with cognitive neuroscience researchers I work with and not one of them is totally convinced this is real yet.
-JS
I work with some of the authors you cited and I personally know most of the others. Near infrar-red spectroscopy and diffuse optical tomography are very good for measuring localized changes in blood oxygenation and volume (and, by inference, brain activation), but going from "you show activation in your right frontal cortex" to "you're lying" is a huge leap and way beyond the limits of any neuroscience research I've ever run across.
This is a submarine patent-they've got an idea, but I assure you they don't know how to make it work yet. First, collecting the signals will be extremely difficult since you have to get past all the light reflected off the skin. The changes in signal level you're looking for are easily at the PPB (yes, 1e-9) level compared with the light you're putting in, so this is decidedly non-trivial. Second, this "information processing device capable of determining various physiological characteristics" is going to be real interesting to build since no one knows what those characteristics are and (as with conventional polygraph) there are innocent explanations for most everything you can measure. Finally, NIRS doesn't penetrate very far through the brain, so you're limited to areas near the skull. Deep structures (eg, the hippocampus) aren't probed at all. If the brain doesn't do any work in these superficial regions (which may or may not be the case), there's nothing to be measured.
In short, you have nothing to worry about from this patent. Worry instead why the government is so interested all of a sudden in an undetectable remotely operated lie detector. We've had visitors from NIMA in our lab, and various three-letter agencies have sent letters to several of my cognitive science colleagues offering to fund research along these lines as well.
-JS
Except that in the book, the velociraptors manage to get off the island and begin migrating through Costa Rica and living off of lysine rich foods.
-JS
One hundred years ago, technologies (railroad and telegraph) enabled a few companies to go national. Now, technology (internet and routine intercontinental flights) lets companies go multi-national. Megacorporations aren't evolving, it's the same thing as the robber barons all over again. Only the names have changed.
-JS
Those who fail to learn the lessons of history are doomed to repeat them...
Yes. Your average /.'r writes code. He doesn't build cars.
-JS
If you're out of work, it's a recession. If I'm out of work, it's a depression.
RFID tags (the retail versions) are passive - they only emit RF when scanned by an external RF source. Without an external transmitter (which also needs to be quite strong, located nearby, or both), there's no signal coming from the tags.
-JS
You have two choices. You can fly in:
I'll take old-fashioned copper wires, thank you very much.
-JS -JS
I realize your joking, but in a twisted sort of way, AT&T Unix is the troff operating system. The original unix killer-app was ROFF and its derivatives (including troff). I've always assumed that that's why the man pages are still done in troff--it was the obvious choice originally and there's been too big an investment (in man-hours) to switch to anything else.
-JS
No, this is a good thing--it forces you to stop and think about what you're doing. Besides, the only time you should ever need to be logged in as administrator (or root) is to install software for system-wide use or to update device drivers.
The real problem is that most Windows software is badly written and requires you to have administrator-level access to do anything useful (including most of the Microsoft Office suite).
-JS