Erm. I beg to differ. Nvidia GPUs are "SIMT" (Single Instruction, Multiple Threads). There are "tricks" to avoid threads in a warp from waiting for other threads (basically, don't use if (condition)... else..., but if(condition)... and if(!condition)...). AMD GPUs are based on VLIW processors, and are closer to your assertion of SIMD, but it's not quite the same thing either.
I am against cheating, and ideally, only the better candidates should get in. But at the same time, I find it interesting that, in the end, someone slightly less good than his/her neighbor can end up graduating from the university he/she should not have been able to get into in the first place — possibly without even cheating once he/she got in.
It's VLIW (very long instruction word) [wikipedia.org] architecture in case anyone wants to look it up
No it's not. There is a VLIW-like processor made by Intel, and it's the Itanium processor (well, Itanium 2 nowadays). It is not a "pure" VLIW processor though, as it is both VLIW and superscalar. VLIW instructions have a fixed size (in the case of the Itanium, each bundle/long word can feature as much as three instructions, although two is the average).
On the other hand, x86 processors have really two parts: one which I would call a "front-end", which receives the CISC instruction. Its only purpose is to decode them, and decompose them into micro-operations/micro-instructions. The latter are the "RISC-like" instructions which are really executed and fill the pipeline.
[Talking about doing "burger jobs" for a year] I quit reading the article at this point.
This is a shame, because the interview does address this issue through questions from the journalist.
Copyright was a legal construct the printers (not the writers!) lobbied for in order to increase their profits
I know how the "copyright" equivalent was created in the late 1700s in France. Beaumarchais was tired of printers and publishers ripping him off. He created the "author's right" (which is slightly different from the copyright US and UK countries use) to be protected against publishers and printers who would sell his work without giving him a cut of the profits. When it came to public execution of his plays however, he had no problem, because he felt a play should be... well, played, and that it somehow belongs to the public.
Later on, this right given to authors to control how they want their work to be distributed became more and more distorted.
Don't forget that many systems use a "first touch" allocation policy. It means even though you might actually allocate only at the beginning of your program, it could actually start to really allocate right in the middle of your program (take a look at how gnu sort is implemented, for example).
Quite simply, this is bullshit. Some of the greatest (sorry, "High value") music and film was produced in an era when there was no DRM. The Beatles, the Rolling Stones, Charlie Chaplin, B.B King, Billie Holiday, Miles Davis, Martin Scorcese, Stanley Kubrick, even Steven Spielberg created their work in a pre-DRM era and somehow managed to sell their work.
Disclaimer: I really think DRM stinks.
But even then, you can't compare an era when people just couldn't copy a record, and thus had to buy it, with ours where everyone can copy almost everything, as long as it exists digitally. DRM are needed not because of (lack of) talent from some artists; they are needed because without them, movie and record companies can't make more money. By having digital equipment, by being able to produce one's own music/film/whatever, the value of a work of art decreases, which is quite a problem when your job is to make money off it.
I have a hard time thinking humans came from apes.
Well, if you're uncomfortable with it, don't worry : humans started out of tiny little thingies in water (unicellular stuff, you know).
I guess I often think of something I heard someone say: "If humans evolved from apes...why are there still apes?"
Well, picture it that way : why not only one species for each branch that occurs ? Why not one single kind for cats, dogs, etc. ? The answer is always the same : different kind of evolutions, different abilities to survive, different places where to survive. I guess an ape is much more apt to survive to living in the jungle barehanded than a human being is. And because one species is more adaptive than another doesn't mean the other one can't survive anyway, as long as the ecosystem it lives in still exists.
Name me one time violence DIDN'T solve a problem
Bombing Afghanistan because of AL QAIDA's acts of terrorism helped prevent Spain from being attacked, right ?
When you pass a C file to a compiler, it generates an object file. It has absolutely no way of knowing where functions declared in the header are defined. You can hack around this; pass multiple source files to the compiler at once and have it treat them as a single one, for example, but this falls down completely when the function is declared in a library (e.g. libc) and you don't have access to the source.
ICC (Intel C Compiler) does what they call "IPO" - InterProcess Optimization, and yes, they use inlining between files. But on the other hand, you can't use the object files anymore, because they're not "real" object files (they contain symbols to indicate where to inline stuff, for instance). So you have in-file inlining, and then, at link time, ICC is able to inline functions that can be.
To Apple, letting a relatively small population of niche scientific users "slim down" the Mac OS X kernel is massively outweighed by preventing the Mac OS X on x86 hacking community from being able to easily and quickly deliver an extremely polished distribution of Mac OS X for non-Apple Intel hardware, instead of the ugly hack they have now.
Well, so all those Xserve computers that were sold to scientific people are just uninteresting machines, then. And the fact that people could alter the kernel to suit their needs was just marketing BS... Ok, I get it now.
By 2010 I suspect at least desktops are 4-CPU systems and as the numbers of cores increase one of the large drawbacks of microkernels raises it's ugly head: microkernels turn simple locking algorithms into distributed computing-style algorithms.
Wake up, distributed algorithms exist at the hardware level since there's been a problem with cache coherent behaviour in multi-processor machines. I don't see what's the problem with them, really. Well, to be honest, I do see, but I'm not sure that the problem is that important with the actual (and future) hardware we can get.
It's fairly clear that I have a moral right to keep my work secret. The moment I make it public and people start reading or viewing or using it, however, it becomes part of the heritage of the society,
Mmmh. There's a slight difference between the copyright as you depict it here, and the droit d'auteur (author's right) upon one of one's own creation in France. As the author of a book, for instance, I can decide that, no, finally, the ending of my book is not the good one, and that the 50,000 copies already printed are to be destroyed. My publisher can't force me not to do that. Of course, in exchange, I have to make up for the loss since I made my publisher spend money uselessly - which is why authors tend to change things with the next editions.:-)
I've begun programming with Learning Perl 1st Ed, almost 8 years ago. I was a highschool student then, and had almost no programming experience (which is to say : I had learned how to program a bit of ASM for a Motorola 6809 chip, which had 2 registers only, so you couldn't do much complex programming with it).
I litterally learnt how to program and - even more important - how to think before beginning to code thanks to the Llama Book.
When later I began my CS studies, programming wasn't a problem at all.
The question whether "God designed evolution" is totally out outside of science. It's a matter of faith, not knowledge. The only academic activity when you can seriously consider this is philosophy.
Philosophy has got nothing to do with religion. It isn't a science nor a religion, nor even a way of studying religions. Of course, there are philosophical topics in various religions, and of course these are discussed by Philosophy, but the question about God creating the world, God creating the process of Evolution, etc, is by definition a non-philosophical question, since the very existence of God (or its non-existence) can't be proved.
As you said, this is a matter of faith, which has nothing to do with philosophy.
just so you know, Jobs earns his life well, but allows himself to be paid a symbolic dollar a year from time to time, just like last year... So this guys not only is a good CEO, he is also clever enough to show that he, contrary to many other CEOs, isn't overpaid (ok, so he is only multimillionaire and not multibillionaire...:-) )
beside taking advantage of availbe libraries (and this is a HUGE good reason) I dont see any other reason to get excited over Perl.
I see one. It has already been said in a previous comment : Perl is fun. You can really express a solution to a problem with ease, and still, you could have found a solution with the same effects with totally different means. How does that sound ? I can't talk about Ruby, nor Tcl, nor Python. I've seen a little bit of the latter for some project I had, and I find it interesting. So this is not about having the best language.
This is all about having a language fun to program with. And Perl is such a language. Strangely enough, that is what makes Perl difficult to handle too, sometimes : that freedom to program as you like make it that much harder to read previous code sometimes, if one hasn't taken some preliminary coding standards.
Re:Maintainability of Perl code?
on
Perl Medic
·
· Score: 1
ok, my mistake... Then replace the first die with die "couldn't open the file $ARGV[0]" and the second with "couldn't close..." what else ? My point was that readability is very subjective a topic. I know of people who read C and C++ idioms like I read Dilbert. But the big mistake people make is that they expect Perl to be a C-like language. This is not so. Perl is far richer : of course one can do C-like programming, but then this would really shorten one's possibilities with the language. This is really a languange that should be read aloud !:-) (I hope I am clear enough now...)
Re:Maintainability of Perl code?
on
Perl Medic
·
· Score: 1
ok.
Tell me, between those two programs (that do exactly the same thing), which is easier to read ? (I had a mycat.c file to show too, but it didn't pass the filters...)
That was exactly one of my points:-)
It is impossible to keep up with Java's improvements quickly enough, unless you're Sun. And I'm not even talking about deprecated classes where you have to change your old code, etc.
Java evolves a little too quickly IMO. It would be OK if it was a really "new" language, but it is almost ten years old (Yes, I know, C was even older when the ISO normalization came, and so was C++ -- probably) ! And as long as the language is not normalized "officialy" (by the ISO or something else), the specs for the language are likely to increase again and again, forcing all vendors to submit their new JVM to Sun for their approval, etc...
... And forcing me to wait for an "official" JVM for my own platform.
"Currently, Java seems to be close to, if not the lead in cross-compatibility"
Could you tell me how much cross platform Java is ? the language is specified, I grant you that. The only problem is that there is no Sun JVM on MacOS X (yes I know, there is one Apple gives with its OS, but it's still a 1.4.x version, not a 1.5 for instance). I'm not even speaking of a Sun JVM for Linux PPC, or even for a UN*X which is not solaris...
Sure, there are various implementations (from IBM, etc.), but tell me : if Java is so cross-platform, why isn't there a JVM for every "famous-enough" OS ? (meaning : AIX, Linux{PPC|other}, MacOS X, etc).
And while I'm writing that, I can only think of a language truly cross-platform : C. Of course, you're still dependant on the compiler, but not that much compared to the JVM when it's about Java. And OSes like NetBSD or Linux prove damn well that if one engineers well one's programs, C is very portable... And certainly more than Java right now.
Java is a language I like, because when there is a correct JVM available, you can really do incredible stuff with it; but since Sun and a board of developers, lead architects, etc... make Java evolve so much, so quickly, all JVMs that are not by Sun are ALWAYS behind it.
Slashdot Mirror
Erm. I beg to differ. Nvidia GPUs are "SIMT" (Single Instruction, Multiple Threads). There are "tricks" to avoid threads in a warp from waiting for other threads (basically, don't use if (condition) ... else ..., but if(condition) ... and if(!condition) ...). AMD GPUs are based on VLIW processors, and are closer to your assertion of SIMD, but it's not quite the same thing either.
I am against cheating, and ideally, only the better candidates should get in. But at the same time, I find it interesting that, in the end, someone slightly less good than his/her neighbor can end up graduating from the university he/she should not have been able to get into in the first place — possibly without even cheating once he/she got in.
And yet, here is this 3-month-old game that in the end sells more than the first week it was launched.
It's VLIW (very long instruction word) [wikipedia.org] architecture in case anyone wants to look it up
No it's not. There is a VLIW-like processor made by Intel, and it's the Itanium processor (well, Itanium 2 nowadays). It is not a "pure" VLIW processor though, as it is both VLIW and superscalar. VLIW instructions have a fixed size (in the case of the Itanium, each bundle/long word can feature as much as three instructions, although two is the average). On the other hand, x86 processors have really two parts: one which I would call a "front-end", which receives the CISC instruction. Its only purpose is to decode them, and decompose them into micro-operations/micro-instructions. The latter are the "RISC-like" instructions which are really executed and fill the pipeline.
[Talking about doing "burger jobs" for a year] I quit reading the article at this point.
This is a shame, because the interview does address this issue through questions from the journalist.
Copyright was a legal construct the printers (not the writers!) lobbied for in order to increase their profits
I know how the "copyright" equivalent was created in the late 1700s in France. Beaumarchais was tired of printers and publishers ripping him off. He created the "author's right" (which is slightly different from the copyright US and UK countries use) to be protected against publishers and printers who would sell his work without giving him a cut of the profits. When it came to public execution of his plays however, he had no problem, because he felt a play should be... well, played, and that it somehow belongs to the public.
Later on, this right given to authors to control how they want their work to be distributed became more and more distorted.
Don't forget that many systems use a "first touch" allocation policy. It means even though you might actually allocate only at the beginning of your program, it could actually start to really allocate right in the middle of your program (take a look at how gnu sort is implemented, for example).
Quite simply, this is bullshit. Some of the greatest (sorry, "High value") music and film was produced in an era when there was no DRM. The Beatles, the Rolling Stones, Charlie Chaplin, B.B King, Billie Holiday, Miles Davis, Martin Scorcese, Stanley Kubrick, even Steven Spielberg created their work in a pre-DRM era and somehow managed to sell their work.
Disclaimer: I really think DRM stinks.
But even then, you can't compare an era when people just couldn't copy a record, and thus had to buy it, with ours where everyone can copy almost everything, as long as it exists digitally. DRM are needed not because of (lack of) talent from some artists; they are needed because without them, movie and record companies can't make more money. By having digital equipment, by being able to produce one's own music/film/whatever, the value of a work of art decreases, which is quite a problem when your job is to make money off it.
I have a hard time thinking humans came from apes.
Well, if you're uncomfortable with it, don't worry : humans started out of tiny little thingies in water (unicellular stuff, you know).
I guess I often think of something I heard someone say: "If humans evolved from apes...why are there still apes?"
Well, picture it that way : why not only one species for each branch that occurs ? Why not one single kind for cats, dogs, etc. ? The answer is always the same : different kind of evolutions, different abilities to survive, different places where to survive. I guess an ape is much more apt to survive to living in the jungle barehanded than a human being is. And because one species is more adaptive than another doesn't mean the other one can't survive anyway, as long as the ecosystem it lives in still exists.
Name me one time violence DIDN'T solve a problem
Bombing Afghanistan because of AL QAIDA's acts of terrorism helped prevent Spain from being attacked, right ?
When you pass a C file to a compiler, it generates an object file. It has absolutely no way of knowing where functions declared in the header are defined. You can hack around this; pass multiple source files to the compiler at once and have it treat them as a single one, for example, but this falls down completely when the function is declared in a library (e.g. libc) and you don't have access to the source.
ICC (Intel C Compiler) does what they call "IPO" - InterProcess Optimization, and yes, they use inlining between files. But on the other hand, you can't use the object files anymore, because they're not "real" object files (they contain symbols to indicate where to inline stuff, for instance). So you have in-file inlining, and then, at link time, ICC is able to inline functions that can be.
Thank you. At last, someone who knows what he's speaking about.
There never was any "iTunes law".
To Apple, letting a relatively small population of niche scientific users "slim down" the Mac OS X kernel is massively outweighed by preventing the Mac OS X on x86 hacking community from being able to easily and quickly deliver an extremely polished distribution of Mac OS X for non-Apple Intel hardware, instead of the ugly hack they have now.
Well, so all those Xserve computers that were sold to scientific people are just uninteresting machines, then. And the fact that people could alter the kernel to suit their needs was just marketing BS... Ok, I get it now.
By 2010 I suspect at least desktops are 4-CPU systems and as the numbers of cores increase one of the large drawbacks of microkernels raises it's ugly head: microkernels turn simple locking algorithms into distributed computing-style algorithms.
Wake up, distributed algorithms exist at the hardware level since there's been a problem with cache coherent behaviour in multi-processor machines. I don't see what's the problem with them, really. Well, to be honest, I do see, but I'm not sure that the problem is that important with the actual (and future) hardware we can get.
It's fairly clear that I have a moral right to keep my work secret. The moment I make it public and people start reading or viewing or using it, however, it becomes part of the heritage of the society, :-)
Mmmh. There's a slight difference between the copyright as you depict it here, and the droit d'auteur (author's right) upon one of one's own creation in France. As the author of a book, for instance, I can decide that, no, finally, the ending of my book is not the good one, and that the 50,000 copies already printed are to be destroyed. My publisher can't force me not to do that. Of course, in exchange, I have to make up for the loss since I made my publisher spend money uselessly - which is why authors tend to change things with the next editions.
53% of Monaco
Err, you do know that Monaco is no bigger than a "big" town in France, right ?
Four letters. MSDN. Now show me something comparable in the OS world.
Three letters. MAN
I've begun programming with Learning Perl 1st Ed, almost 8 years ago. I was a highschool student then, and had almost no programming experience (which is to say : I had learned how to program a bit of ASM for a Motorola 6809 chip, which had 2 registers only, so you couldn't do much complex programming with it).
I litterally learnt how to program and - even more important - how to think before beginning to code thanks to the Llama Book.
When later I began my CS studies, programming wasn't a problem at all.
The question whether "God designed evolution" is totally out outside of science. It's a matter of faith, not knowledge. The only academic activity when you can seriously consider this is philosophy.
Philosophy has got nothing to do with religion. It isn't a science nor a religion, nor even a way of studying religions. Of course, there are philosophical topics in various religions, and of course these are discussed by Philosophy, but the question about God creating the world, God creating the process of Evolution, etc, is by definition a non-philosophical question, since the very existence of God (or its non-existence) can't be proved.
As you said, this is a matter of faith, which has nothing to do with philosophy.
just so you know, Jobs earns his life well, but allows himself to be paid a symbolic dollar a year from time to time, just like last year... So this guys not only is a good CEO, he is also clever enough to show that he, contrary to many other CEOs, isn't overpaid (ok, so he is only multimillionaire and not multibillionaire... :-) )
beside taking advantage of availbe libraries (and this is a HUGE good reason) I dont see any other reason to get excited over Perl.
I see one. It has already been said in a previous comment : Perl is fun. You can really express a solution to a problem with ease, and still, you could have found a solution with the same effects with totally different means. How does that sound ? I can't talk about Ruby, nor Tcl, nor Python. I've seen a little bit of the latter for some project I had, and I find it interesting. So this is not about having the best language.
This is all about having a language fun to program with. And Perl is such a language. Strangely enough, that is what makes Perl difficult to handle too, sometimes : that freedom to program as you like make it that much harder to read previous code sometimes, if one hasn't taken some preliminary coding standards.
ok, my mistake... Then replace the first die with ..." :-) (I hope I am clear enough now...)
die "couldn't open the file $ARGV[0]" and the second with "couldn't close
what else ?
My point was that readability is very subjective a topic. I know of people who read C and C++ idioms like I read Dilbert. But the big mistake people make is that they expect Perl to be a C-like language. This is not so. Perl is far richer : of course one can do C-like programming, but then this would really shorten one's possibilities with the language. This is really a languange that should be read aloud !
ok.
...)
;
Tell me, between those two programs (that do exactly the same thing), which is easier to read ?
(I had a mycat.c file to show too, but it didn't pass the filters
/**** MyCat.java ****/
package net.beuarh.lasher;
import java.io.BufferedReader;
import java.io.File;
import java.io.IOException;
import java.io.FileNotFoundException;
import java.io.FileReader;
public class MyCat {
public static void main(String[] args) {
if (args.length != 1) {
System.err.println("Usage : " + MyCat.class + " <file_to_cat>");
System.exit(0);
}
try {
FileReader fr = new FileReader(new File(args[0]));
BufferedReader br = new BufferedReader(fr);
for (String in = br.readLine(); in != null; in = br.readLine())
System.out.println(in);
} catch (FileNotFoundException fnfe) {
System.err.println(fnfe) ; System.exit(0)
} catch (IOException ioe) {
System.err.println(ioe); System.exit(0);
}
}
}
/**** MyCat.java ****/
#---- mycat.pl ----#
#!/usr/bin/perl
use strict;
use warnings;
print STDERR "usage : $0 <file_to_cat>\n" and exit unless($ARGV[0]);
open(IN, "<$ARGV[0]") || die "$! $@";
print while (<IN>);
close(IN) || die "$! $@";
#---- mycat.pl ----#
That was exactly one of my points :-)
... And forcing me to wait for an "official" JVM for my own platform.
It is impossible to keep up with Java's improvements quickly enough, unless you're Sun. And I'm not even talking about deprecated classes where you have to change your old code, etc.
Java evolves a little too quickly IMO. It would be OK if it was a really "new" language, but it is almost ten years old (Yes, I know, C was even older when the ISO normalization came, and so was C++ -- probably) ! And as long as the language is not normalized "officialy" (by the ISO or something else), the specs for the language are likely to increase again and again, forcing all vendors to submit their new JVM to Sun for their approval, etc...
"Currently, Java seems to be close to, if not the lead in cross-compatibility"
Could you tell me how much cross platform Java is ? the language is specified, I grant you that. The only problem is that there is no Sun JVM on MacOS X (yes I know, there is one Apple gives with its OS, but it's still a 1.4.x version, not a 1.5 for instance). I'm not even speaking of a Sun JVM for Linux PPC, or even for a UN*X which is not solaris...
Sure, there are various implementations (from IBM, etc.), but tell me : if Java is so cross-platform, why isn't there a JVM for every "famous-enough" OS ? (meaning : AIX, Linux{PPC|other}, MacOS X, etc).
And while I'm writing that, I can only think of a language truly cross-platform : C. Of course, you're still dependant on the compiler, but not that much compared to the JVM when it's about Java. And OSes like NetBSD or Linux prove damn well that if one engineers well one's programs, C is very portable... And certainly more than Java right now.
Java is a language I like, because when there is a correct JVM available, you can really do incredible stuff with it; but since Sun and a board of developers, lead architects, etc... make Java evolve so much, so quickly, all JVMs that are not by Sun are ALWAYS behind it.
So much for the "cross-platform" attitude, then.