I presume it means that the application can directly access the shared data through the NUMA memory rather than using mpi to access the data from whatever node it thinks the data is on. Data coherency gets moved down into the hardware/kernel layer rather than up at the application layer. The communication of the data would be done by a low latency interconnect either way.
This article is news to me. My impression was that HPC programmers preferred mpi over shared memory multi-threading because they found the former somehow more intuitive and less error prone.
the entire tech economy is in a race for the bottom. All the companies are living off of their seed grain because they're waiting for someone else to make the first move. McNealy should be right because long term thinking should be the best strategy. But even 7 billion may not be enough to tough it out. Even Microsoft is getting worried and it has 30 billion or so.
Well, it used to be true at one time. Apparently the crt calibration controls can handle it now. CRT calibration used to be quite primative at one time and even require field techs to do. See here for current status.
For compasses, see... dammit I have the compass in front of me and I was going to provide a link to the mfgr's web site but the webmonkeys that built the site insist that you use flash7 or nothing and I'll be damned if I going to provide free publicity to a bunch of cretins. I'll take the karmic hit.
if they weren't already. What most people don't know is that CRT monitors come in northern hemisphere and southern hemisphere models due to the the effect of the vertical component of the magnetic field on the electron beam in the CRT.
Same reason there are northern and southern hemisphere compasses except it's a needle balancing issue. In the northern hemisphere, the "north" end of the needle gets pulled down, and it gets pulled up in the southern hemisphere. There are global compasses that work by allowing you to readjust the balance or by using a gimbaled disk magnet.
I've used stuff that gave you a realtime display of i/o perfmance stats and you could see when the hotpoints were happening. The problem was how to fix it. It's not like the old days when you had maybe 3 disks and you just moved a filesystem or database partition to another disk. Now you have huge amounts of raid storage with logical volume managers and your logical volumes (partitions) are all over the place. It's not so easy. You could see a hot spot and determine that it's caused by two logical volumes sharing a physical drive. Except when you move one to a different drive, you could create different hot spots and they could be much worse. Kind of like wack-a-mole except you're playing it while you're on call. Not fun.
It's a hugely complicated problem and it doesn't have a simple solution except in the minds of markedroids.
C is especially bad with multi-threading. It doesn't help that the ANSI C and Posix thread committees refuse to acknowledge each other's existence.
Java has a cleaner thread model but is not as extendable as C. In Java you find that what you want to extend is either not an abstraction point or somebody arbitrarily specified the final keyword to stroke his ego.
So basically, you will never get rid of C unitl you replace it with something as easy to extend as C.
1% or even 3% (what I heard) of layoffs (turnover) doesn't mean that outsourcing is not a problem. It could be 0% and outsourcing could still be a problem. It's the amount of new jobs not being created locally because they're being created overseas. So you really want to take a look at the number of those jobs.
The percent of layoffs due to outsourcing is only significant if layoffs are still a problem. So, you have a roughly 1% increased chance of being laid off due to outsourcing. But the other 99% who were laid off due to other reasons might still be concerned about outsourcing's effect on their getting a new job.
This one needs to be watched carefully. It's claimed the ads are valid because the movie is a documentary. They're going to have to be careful how they define documentary. Otherwise advertizements about a 2 minute documentary about some people watching a policial ad can be used as a loophole on political advertising. I think criteria, such as Miramax has to distribute your documentary and it has to be shown in theaters, needs to be set.
The current trend for companies to require experience rather than train you (it didn't always use to be this way) and the rapid pace of technology change means that you will be in this situation again and again. Unless you are lucky enough to be trying out new technologies as part of your job. Doesn't count otherwise.
For now you can take advantage of internships (basically unpaid training). Later on that is not an option. Ever try supporting a family on an internship even if it was offered to you.
Long term, you are probably screwed. If you are lucky you might be able to pay off that college loan before you have to take a McJob like all those other experienced workers you envy. Oh, wait. Sorry, you weren't supposed to be told this yet.
as there could be a reason they got that last batch of motherboards real cheap. It's not like those suppliers make sure they only carry the latest and most current version of anything.
Actually, considering the trend to not provide any kind of warrently, i.e. all returns to orig. manufacturer, you have to wonder if reputable vendors have any incentive to carry the most current versions either.
You can't patent anything that's obvious to practioners in the art. If scores of people independently come up with the idea without having heard of the patent, then that's pretty much the definition of obvious.
I have a suspicion that any advanced galatic civilization, realizing the nature of expansionistic species, broadcasts instructions on how to blow yourself up, knowing full well that any sufficiently aggressive species will not be able to resist following the instructions. The tiny note at the end "Do not attempt this on your home planet" just indicates a puckish sense of humor.
Standard business practice means that you do business with some pretty sleazy and shady characters without asking too many questions. So being DRM certified won't mean using such is safer. The fact of this can be verified by reading the contractual documentation which will limit the liability of the DRM service providers in case anything bad happens to you from those sleazy and shady characters.
But it will be good for the established sleazy and shady characters as it will limit competition from the upstart sleazy and shady characters who will find higher barriers to competition. "legitimate" spyware vs. illegitmate worms and virii. Did you think spyware would have any problem getting a DRM certification?
For milk, yes. They want a guaranteed locked in revenue stream. They can't be subject to the arbitrary whims of a consumer ordering milk only when he thinks he needs it.
It's patent 5,924,098. It's owned by Sun and it's basically on using the Boehm style GC to do well known lock-free programming techniques, one of which RCU is based on. I even commented on the obviousness of the technique before Sun was issued the patent here. It's not prior art but it does show obviousness to someone (me) versed in the art of lock-free programming. It's a standard technique that depends on some mechanism to delay deallocation of data nodes until they are no longer referenced. Which is by definition Garbage Collection. Specifying a known form of GC in conjunction with this technique is not an invention. Coming up with a new form of GC or proxy GC is an invention.
No, but isn't this like one of those "when did you stop being a troll" questions? Oh, right. You're AC.
But to explain it in more detail to the comceptually challenged. The algorithm was a binary search and was O(log n) and was PDQ now matter how you did it. The thing you can get from assembly programming that the article strongly hints at is an O(n) improvement in code path lenghts. But in this case the linear improvement in for such a small execution time was not enough to outweight the startup overhead of the binary on this particular system (not unix). If the startup overhead of the binary is greater than the total runtime of the script, it doesn't matter how fast the assember program would have run. We had extensive performance profiling data so we knew where the system spent most of it's exection time.
But if you are writing some kind of calculation that has a loop of a gazillion or so, than a linear improvment might be worth going to assembler for.
I've probably been assembler programming longer than you've been alive but I rarely use it now. Just in C inline code where there is no C language facility available to do what is needed. I used to be a kernel developer for an OS written in millions of lines of assembler code but I maintained that less than 1% of that code was actually performance sensitive enough to justify writting assembler and even there more a compiler can't do it rather than a performance issue. Most of our performance bang for the buck was gotten from algorithmic improvements, not from code tweaking.
Thanks for this opportunity to make some additional clarifying points.
I used to have two binary search utilities. One written in assembler and one written in a shell script. Same logic. The script was faster. Performance isn't just a matter of number of lines of code.
In this case I think the reason was the system having to load the binary vs. the script interpreter already being resident in memory. The start up overhead dominated the actual runtime overhead - binary searchs are very quick.
at this point since the basic cracking techniques have pretty well be refined by now and the bad guys aren't going to stop. You probably could have made an argument about not escalating in the first place which would have places a higher cost to engaging in cracking.
Nope, unfortunately Darwinian selection is now going to start applying to software and computers. Bandages aren't going to help at this point.
The adoption analogy doesn't quite correspond to being unemployed. That should have been, put your children out on the street rather than with another family.
The number I'm seeing is 4,633 jobs lost to offshoring in the first quarter. Okay. So assuming a 6:1 US to offshore salary ration, then we should have seen only a 27,798 increase in offshore positions. If the increase was more then the US has had a net loss of jobs. And we definitely have not see enough of an increase in the US job level to maintain current emmployment levels in light of new workers entering the workforce. I'm talking about tech jobs mainly. So this may be good for the US economy but the economy and the workforce are not the same thing. It's sort of like saying that putting your children up for adoption is good. It may be good for your family's budget but it's definitely not good for your family.
The ability to catch speeders with older and even much more primative technology has been there for a while. There was once a proposal to use the timestamped toll tickets to automatically catch speeders on the NY State Thruway until it was realized that state legislature representatives would also automatically get them on their way to Albany. That's why your state reps have special plates. It's not for vanity, it's so the police can recognize them and not ticket them for speeding. The process has to remain manual because that's the only way unofficial exemptions can work.
Slashdot Mirror
This article is news to me. My impression was that HPC programmers preferred mpi over shared memory multi-threading because they found the former somehow more intuitive and less error prone.
the entire tech economy is in a race for the bottom. All the companies are living off of their seed grain because they're waiting for someone else to make the first move. McNealy should be right because long term thinking should be the best strategy. But even 7 billion may not be enough to tough it out. Even Microsoft is getting worried and it has 30 billion or so.
Suunto M2 compass. See the nice map of the 5 compass zones. Even comes with an optional "global" needle. The one I have is an older model, the MC-1G.
For compasses, see ... dammit I have the compass in front of me and I was going to provide a link to the mfgr's web site but the webmonkeys that built the site insist that you use flash7 or nothing and I'll be damned if I going to provide free publicity to a bunch of cretins. I'll take the karmic hit.
Same reason there are northern and southern hemisphere compasses except it's a needle balancing issue. In the northern hemisphere, the "north" end of the needle gets pulled down, and it gets pulled up in the southern hemisphere. There are global compasses that work by allowing you to readjust the balance or by using a gimbaled disk magnet.
It's a hugely complicated problem and it doesn't have a simple solution except in the minds of markedroids.
I like the old fashioned way of doing it. Just watering down the gas. No gimicky reprogramming of gas pump electronics.
Remember, your browser is only secure as the least secure plug-in.
Java has a cleaner thread model but is not as extendable as C. In Java you find that what you want to extend is either not an abstraction point or somebody arbitrarily specified the final keyword to stroke his ego.
So basically, you will never get rid of C unitl you replace it with something as easy to extend as C.
Cool. Faster than light communication. How are they doing this? Quantum entaglement?
The percent of layoffs due to outsourcing is only significant if layoffs are still a problem. So, you have a roughly 1% increased chance of being laid off due to outsourcing. But the other 99% who were laid off due to other reasons might still be concerned about outsourcing's effect on their getting a new job.
This one needs to be watched carefully. It's claimed the ads are valid because the movie is a documentary. They're going to have to be careful how they define documentary. Otherwise advertizements about a 2 minute documentary about some people watching a policial ad can be used as a loophole on political advertising. I think criteria, such as Miramax has to distribute your documentary and it has to be shown in theaters, needs to be set.
For now you can take advantage of internships (basically unpaid training). Later on that is not an option. Ever try supporting a family on an internship even if it was offered to you.
Long term, you are probably screwed. If you are lucky you might be able to pay off that college loan before you have to take a McJob like all those other experienced workers you envy. Oh, wait. Sorry, you weren't supposed to be told this yet.
Actually, considering the trend to not provide any kind of warrently, i.e. all returns to orig. manufacturer, you have to wonder if reputable vendors have any incentive to carry the most current versions either.
You can't patent anything that's obvious to practioners in the art. If scores of people independently come up with the idea without having heard of the patent, then that's pretty much the definition of obvious.
I have a suspicion that any advanced galatic civilization, realizing the nature of expansionistic species, broadcasts instructions on how to blow yourself up, knowing full well that any sufficiently aggressive species will not be able to resist following the instructions. The tiny note at the end "Do not attempt this on your home planet" just indicates a puckish sense of humor.
But it will be good for the established sleazy and shady characters as it will limit competition from the upstart sleazy and shady characters who will find higher barriers to competition. "legitimate" spyware vs. illegitmate worms and virii. Did you think spyware would have any problem getting a DRM certification?
For milk, yes. They want a guaranteed locked in revenue stream. They can't be subject to the arbitrary whims of a consumer ordering milk only when he thinks he needs it.
It's patent 5,924,098. It's owned by Sun and it's basically on using the Boehm style GC to do well known lock-free programming techniques, one of which RCU is based on. I even commented on the obviousness of the technique before Sun was issued the patent here. It's not prior art but it does show obviousness to someone (me) versed in the art of lock-free programming. It's a standard technique that depends on some mechanism to delay deallocation of data nodes until they are no longer referenced. Which is by definition Garbage Collection. Specifying a known form of GC in conjunction with this technique is not an invention. Coming up with a new form of GC or proxy GC is an invention.
But to explain it in more detail to the comceptually challenged. The algorithm was a binary search and was O(log n) and was PDQ now matter how you did it. The thing you can get from assembly programming that the article strongly hints at is an O(n) improvement in code path lenghts. But in this case the linear improvement in for such a small execution time was not enough to outweight the startup overhead of the binary on this particular system (not unix). If the startup overhead of the binary is greater than the total runtime of the script, it doesn't matter how fast the assember program would have run. We had extensive performance profiling data so we knew where the system spent most of it's exection time.
But if you are writing some kind of calculation that has a loop of a gazillion or so, than a linear improvment might be worth going to assembler for.
I've probably been assembler programming longer than you've been alive but I rarely use it now. Just in C inline code where there is no C language facility available to do what is needed. I used to be a kernel developer for an OS written in millions of lines of assembler code but I maintained that less than 1% of that code was actually performance sensitive enough to justify writting assembler and even there more a compiler can't do it rather than a performance issue. Most of our performance bang for the buck was gotten from algorithmic improvements, not from code tweaking.
Thanks for this opportunity to make some additional clarifying points.
In this case I think the reason was the system having to load the binary vs. the script interpreter already being resident in memory. The start up overhead dominated the actual runtime overhead - binary searchs are very quick.
Nope, unfortunately Darwinian selection is now going to start applying to software and computers. Bandages aren't going to help at this point.
The adoption analogy doesn't quite correspond to being unemployed. That should have been, put your children out on the street rather than with another family.
The number I'm seeing is 4,633 jobs lost to offshoring in the first quarter. Okay. So assuming a 6:1 US to offshore salary ration, then we should have seen only a 27,798 increase in offshore positions. If the increase was more then the US has had a net loss of jobs. And we definitely have not see enough of an increase in the US job level to maintain current emmployment levels in light of new workers entering the workforce. I'm talking about tech jobs mainly. So this may be good for the US economy but the economy and the workforce are not the same thing. It's sort of like saying that putting your children up for adoption is good. It may be good for your family's budget but it's definitely not good for your family.
The ability to catch speeders with older and even much more primative technology has been there for a while. There was once a proposal to use the timestamped toll tickets to automatically catch speeders on the NY State Thruway until it was realized that state legislature representatives would also automatically get them on their way to Albany. That's why your state reps have special plates. It's not for vanity, it's so the police can recognize them and not ticket them for speeding. The process has to remain manual because that's the only way unofficial exemptions can work.