He benchmarked gcc 3.2.1 and did not benchmark gcc 2.95.x...
Why should he have benchmarked gcc 2.95? What critical information would that provide? Other reports seem to indicate that compared to gcc 3.2, 2.95 is slightly slower, especially with processor-specific optimizations, and it has poor support for C99 and C++98. Nothing new or especially interresting here.
Plus, parallelization should not have been a part of such a benchmark. Parallelization for a cluster _or_ a supercomputer has many issues...
He didn't claim to test parallalization for supercomputers, he just tested to see if the automatic parallelization and openmp directives helped with the intel compiler on the p4 hyperthreading cpu. Automatic parallelization has a lot to do with the compilers ability to analyze the structure of the code and determine which parts can be run in parallel and which parts cannot. And as you can see from the results, in this case at least, there was little benefit from it.
OO blah blah..
I agree with you that OO is beneficial (for real life projects). OTOH, if you look at his benchmark code it's just a few pages. For a code which is that small there is probably little benefit from OO. And thus the c++ and java versions use no OO features, rightly so IMHO.
Indias government is democratically elected, just like most nuclear states like the US, Russia, etc. The same argument applies to any of them. Remember the cold war?
Most interresting problems are parallelizable, to some extent. In fact, all supercomputers today are parallel computers. The only thing that separates what you call "general purpose supercomputers" and a cluster is the bandwidth and latency of the interconnect. And in some cases, the programming model (MPI vs. threads vs. vector compilers/libraries).
The entire idea of supercomputers has always been to do things in parallel. It's not like supercomputer chips are magically faster at doing serial work than other chips. Whether parallelization is achieved by having lots of chips or by having vector chips (or both), supercomputers calculate things faster than serial computers because they calculate many things at once.s has always been to do things in parallel. It's not like supercomputer chips are magically faster at doing serial work than other chips. Whether parallelization is achieved by having lots of chips or by having vector chips (or both), supercomputers calculate things faster than serial computers because they calculate many things at once.
Uh.. My understanding is that SATA is point-to-point only. Well, at least they got rid of that master/slave stuff, and you get the whole bandwidth for a single drive. At least in theory. In practice most SATA chipsets will probably share the bandwidth anyway so the manufacturers can save a few cents.:)
Any actual analysis is so horrible and incomplete as to make the review worthless. Very fiew hardware review sites do an actual good job of doing hardware reviews. Storagereview I view as the best. Too bad they only do hard drives!
Well, what did you expect then? Most of these "hardware review sites" seem to be run by 14-year olds with way too much free time and who drool over the latest-and-greatest because, uh oh, it's the latest-and-greatest. And their mommys are probably proud of them too: "Look daddy, Jimmy is doing all this important uh, community work instead of doing hard drugs and shooting people on the streets. I'm so proud of him!"
Mercifully, when I had to install Excel on Wine because OpenOffice doesn't do something as fscking simple as a polynomial regression, the damned paperclip didn't work.
Well, when you need math there's for example R and Octave instead of that excel crap.
I read evaCuation.. stuff the patient in a "briefcase" and throw him out of the plane. Briefcase has a satellite beacon (and hopefully a parachute) so personnel on the ground can find him... Oh well...:)
Weren't they planning to develop some kind of box-wing plane for a next generation tanker/transport? Google around for box-wing and you find a few articles with pictures of them... Looks like a good concept, which I think could succeed in a commercial airliner as well. Too bad that Boeing and Airbus seem content with only improving their current designs instead of coming up with some real innovation.
I really hope that building ancillary tools like nice clients (wincvs) and useful add-ons (bonsai) is easy enough to do, because that's really where the critical mass is wrt widespread adoption.
Most of the subversion functionality is actually in a library, which makes it a lot easier and more robust to build gui clients and such for subversion. CVS, by comparison, is only accessed through the command-line client, so cvs gui clients have to execute the cvs binary and then parse the output, which as you certainly can imagine, is cumbersome.
Except that http is even more lightweight, in terms of bandwidth overhead and it also requires only one tcp connection.
Re:Using PHP on a professional site
on
Professional PHP4
·
· Score: 2
One of the main reasons yahoo went with php instead of java is that freebsd currently has lousy thread support, which is quite essential for java apps.
Another thing is that they still use their proprietary backend code, they are not going to rewrite that in php (or java for that matter).
I appreciate this man's writing, he is thorough and insightful....
Keep posting articles from this man, whoever is reading, I would like to see more of his work.
Well, considering that he (and two other guys) got the Nobel prize in physics a few years ago (for the discovery and explanation of the fractional quantum hall effect), you'd kinda expect him to be able to write insightful stuff..:)
Umm, this really is quantum mechanics 101 stuff...
Think about when you have for example an electron in a potential barrier. Or for a macroscopic analogy, a ball in a bowl. If the energy of the electron is less than the barrier height, there's no way, according to classical physics, that the electron (ball) is going to escape from that barrier (bowl).
Enter quantum mechanics, where it is entirely possible that the electron will escape from the barrier. Actually, as long as the barrier is finite, the probability of the electron tunneling through the barrier is non-zero. And as the barrier gets smaller, the probability increases.
As you perhaps have guessed by now, in transistors there is a potential barrier preventing the electrons from traveling the wrong way. So when transistors get smaller, the probability of the electron tunneling through the barrier increases.
2003 won't be a problem, chips which will debut then are probably running fine in labs already, they're just doing fine tuning on them.
If you had read the article you would have seen it mentioned "by the end of the decade". As the parent poster mentioned, the problem is basic physics; when transistors get small enough, quantum effects will start to dominate. There's no way around that.
Cellular networks are pretty much the only good way to deploy cheap, scalable wireless data services.
While not ready yet, the HAPS (high altitude platform station) concept looks cool. Basically, stick a phased array antenna on either an autonomous solar powered aircraft (like the NASA helios) or an airship. Cheaper than satellite, and thanks to array antennas capacity is huge. For more info, check for example this overview.
I'd say the main reason for Linux development continuing rapidly despite the economy is that the Linux market is orders of magnitude larger than the *BSD market, so the distro makers (and other companies who employ Linux kernel hackers) have the money to keep Linux kernel hackers employed.
Well, about kde3, all I can say is that since I upgraded kde on my woody box to kde3, konqueror (and the few other kde apps I use) crashes a lot less than konqueror in kde 2.2.2. Not that I blame the packagers though, when the decision was made not to package kde3 not much was known about how stable it would be (if it even was released at the time, I don't remember). It could have been full of bugs, so it was certainly a better and safer choice to go with kde 2.2.2, which was known to be relatively bug-free.
Anyway, hopefully they get to release sarge by the end of summer 2003. Or if they at least could get gcc 3.2 as the default compiler in sid by then...*sigh*..:)
*dons asbestos kit* Related to the above, the weird thing about debian is their stubborn refusal to use a schedule. "Release when ready", what kind of mantra is that? Every human endeavour requiring cooperation among many individuals, related to computing, warfare, whatever, for pay or voluntary, benefits from a schedule. Hell, most people schedule their own lives too, for good or bad. What makes the debian project so special, that they can't use the same basic tool that almost all other projects in the world use?
Yes, of course I know that debian developers are volunteers, they can't be forced to do anything. So what? It's not like debian is the only volunteer project in the world. Most schedule their activities somehow. It gives everyone a common goal to reach for.
Looks like HP is changing from an engineering company to a bunch of vacuum cleaner salesmen, like say, Dell. They spun off their measurement systems stuff (Agilent), killing PA-RISC and alpha, what do they have left? Reselling Intel boxes and perhaps some consulting.
On the other hand, chip development costs seem to grow exponentially, so keeping on developing two high-end architectures for a very small market doesn't really make sense economically.
I'd say a lot of joe sixpack's software can be parallelized to some extent. But as very few consumers have multi-cpu computers, and win9x and xp home don't support SMP, the extra hassle of writing multithreaded programs is just not worth it. Perhaps when hyperthreading becomes more common, developers will start to use multiple threads. Perhaps xp home support it, I don't know.
Slashdot Mirror
He benchmarked gcc 3.2.1 and did not benchmark gcc 2.95.x...
Why should he have benchmarked gcc 2.95? What critical information would that provide? Other reports seem to indicate that compared to gcc 3.2, 2.95 is slightly slower, especially with processor-specific optimizations, and it has poor support for C99 and C++98. Nothing new or especially interresting here.
Plus, parallelization should not have been a part of such a benchmark.
Parallelization for a cluster _or_ a supercomputer has many issues...
He didn't claim to test parallalization for supercomputers, he just tested to see if the automatic parallelization and openmp directives helped with the intel compiler on the p4 hyperthreading cpu. Automatic parallelization has a lot to do with the compilers ability to analyze the structure of the code and determine which parts can be run in parallel and which parts cannot. And as you can see from the results, in this case at least, there was little benefit from it.
OO blah blah..
I agree with you that OO is beneficial (for real life projects). OTOH, if you look at his benchmark code it's just a few pages. For a code which is that small there is probably little benefit from OO. And thus the c++ and java versions use no OO features, rightly so IMHO.
is there a native MATLAB for Linus now?
Matlab has been available for Linux since 1995.
I'm no php man, but I hear that phrame is a struts clone for php.
You see - India has FAIRLY stable government.
Indias government is democratically elected, just like most nuclear states like the US, Russia, etc. The same argument applies to any of them. Remember the cold war?
Most interresting problems are parallelizable, to some extent. In fact, all supercomputers today are parallel computers. The only thing that separates what you call "general purpose supercomputers" and a cluster is the bandwidth and latency of the interconnect. And in some cases, the programming model (MPI vs. threads vs. vector compilers/libraries).
The entire idea of supercomputers has always been to do things in parallel. It's not like supercomputer chips are magically faster at doing serial work than other chips. Whether parallelization is achieved by having lots of chips or by having vector chips (or both), supercomputers calculate things faster than serial computers because they calculate many things at once.s has always been to do things in parallel. It's not like supercomputer chips are magically faster at doing serial work than other chips. Whether parallelization is achieved by having lots of chips or by having vector chips (or both), supercomputers calculate things faster than serial computers because they calculate many things at once.
Uh.. My understanding is that SATA is point-to-point only. Well, at least they got rid of that master/slave stuff, and you get the whole bandwidth for a single drive. At least in theory. In practice most SATA chipsets will probably share the bandwidth anyway so the manufacturers can save a few cents. :)
Any actual analysis is so horrible and incomplete as to make the review worthless. Very fiew hardware review sites do an actual good job of doing hardware reviews. Storagereview I view as the best. Too bad they only do hard drives!
Well, what did you expect then? Most of these "hardware review sites" seem to be run by 14-year olds with way too much free time and who drool over the latest-and-greatest because, uh oh, it's the latest-and-greatest. And their mommys are probably proud of them too: "Look daddy, Jimmy is doing all this important uh, community work instead of doing hard drugs and shooting people on the streets. I'm so proud of him!"
Mercifully, when I had to install Excel on Wine because OpenOffice doesn't do something as fscking simple as a polynomial regression, the damned paperclip didn't work.
Well, when you need math there's for example R and Octave instead of that excel crap.
I read evaCuation.. stuff the patient in a "briefcase" and throw him out of the plane. Briefcase has a satellite beacon (and hopefully a parachute) so personnel on the ground can find him... Oh well...:)
Weren't they planning to develop some kind of box-wing plane for a next generation tanker/transport? Google around for box-wing and you find a few articles with pictures of them... Looks like a good concept, which I think could succeed in a commercial airliner as well. Too bad that Boeing and Airbus seem content with only improving their current designs instead of coming up with some real innovation.
I really hope that building ancillary tools like nice clients (wincvs) and useful add-ons (bonsai) is easy enough to do, because that's really where the critical mass is wrt widespread adoption.
Most of the subversion functionality is actually in a library, which makes it a lot easier and more robust to build gui clients and such for subversion. CVS, by comparison, is only accessed through the command-line client, so cvs gui clients have to execute the cvs binary and then parse the output, which as you certainly can imagine, is cumbersome.
Except that http is even more lightweight, in terms of bandwidth overhead and it also requires only one tcp connection.
One of the main reasons yahoo went with php instead of java is that freebsd currently has lousy thread support, which is quite essential for java apps.
Another thing is that they still use their proprietary backend code, they are not going to rewrite that in php (or java for that matter).
I appreciate this man's writing, he is thorough and insightful.
Keep posting articles from this man, whoever is reading, I would like to see more of his work.
Well, considering that he (and two other guys) got the Nobel prize in physics a few years ago (for the discovery and explanation of the fractional quantum hall effect), you'd kinda expect him to be able to write insightful stuff..
If you prefer java to php/perl there is Roller, made with open source java stuff (mostly jakarta). It has become quite popular in the java community.
Umm, this really is quantum mechanics 101 stuff...
Think about when you have for example an electron in a potential barrier. Or for a macroscopic analogy, a ball in a bowl. If the energy of the electron is less than the barrier height, there's no way, according to classical physics, that the electron (ball) is going to escape from that barrier (bowl).
Enter quantum mechanics, where it is entirely possible that the electron will escape from the barrier. Actually, as long as the barrier is finite, the probability of the electron tunneling through the barrier is non-zero. And as the barrier gets smaller, the probability increases.
As you perhaps have guessed by now, in transistors there is a potential barrier preventing the electrons from traveling the wrong way. So when transistors get smaller, the probability of the electron tunneling through the barrier increases.
2003 won't be a problem, chips which will debut then are probably running fine in labs already, they're just doing fine tuning on them.
If you had read the article you would have seen it mentioned "by the end of the decade". As the parent poster mentioned, the problem is basic physics; when transistors get small enough, quantum effects will start to dominate. There's no way around that.
Cellular networks are pretty much the only good way to deploy cheap, scalable wireless data services.
While not ready yet, the HAPS (high altitude platform station) concept looks cool. Basically, stick a phased array antenna on either an autonomous solar powered aircraft (like the NASA helios) or an airship. Cheaper than satellite, and thanks to array antennas capacity is huge. For more info, check for example
this overview.
Well, not for emacs, but you can have a paperclip for vi:
Vigor
I'd say the main reason for Linux development continuing rapidly despite the economy is that the Linux market is orders of magnitude larger than the *BSD market, so the distro makers (and other companies who employ Linux kernel hackers) have the money to keep Linux kernel hackers employed.
With that kind of price difference, he could even buy a beowulf cluster of those lapt... *smack*
*ouch*
*wince*
Wired has introduced a new layout, which is xhtml compliant (and looks quite sophisticated too). See this interview for more info.
Well, about kde3, all I can say is that since I upgraded kde on my woody box to kde3, konqueror (and the few other kde apps I use) crashes a lot less than konqueror in kde 2.2.2. Not that I blame the packagers though, when the decision was made not to package kde3 not much was known about how stable it would be (if it even was released at the time, I don't remember). It could have been full of bugs, so it was certainly a better and safer choice to go with kde 2.2.2, which was known to be relatively bug-free.
Anyway, hopefully they get to release sarge by the end of summer 2003. Or if they at least could get gcc 3.2 as the default compiler in sid by then...*sigh*..:)
*dons asbestos kit*
Related to the above, the weird thing about debian is their stubborn refusal to use a schedule. "Release when ready", what kind of mantra is that? Every human endeavour requiring cooperation among many individuals, related to computing, warfare, whatever, for pay or voluntary, benefits from a schedule. Hell, most people schedule their own lives too, for good or bad. What makes the debian project so special, that they can't use the same basic tool that almost all other projects in the world use?
Yes, of course I know that debian developers are volunteers, they can't be forced to do anything. So what? It's not like debian is the only volunteer project in the world. Most schedule their activities somehow. It gives everyone a common goal to reach for.
Looks like HP is changing from an engineering company to a bunch of vacuum cleaner salesmen, like say, Dell. They spun off their measurement systems stuff (Agilent), killing PA-RISC and alpha, what do they have left? Reselling Intel boxes and perhaps some consulting.
On the other hand, chip development costs seem to grow exponentially, so keeping on developing two high-end architectures for a very small market doesn't really make sense economically.
I'd say a lot of joe sixpack's software can be parallelized to some extent. But as very few consumers have multi-cpu computers, and win9x and xp home don't support SMP, the extra hassle of writing multithreaded programs is just not worth it. Perhaps when hyperthreading becomes more common, developers will start to use multiple threads. Perhaps xp home support it, I don't know.