Indeed, it generally means you aren't too clever. The smarter brains tend to demonstrate less activity due to increased efficiency (or at least that's the theory).
Ahhhhh I like stories like this that tell me I'm smarter than everbody else!
I'm not sure, but it seems like you may have misinterpreted the findings. Keep in mind that more intelligent people tend to demonstrate less brain activity when problem solving. The belief is that their brains are simply more efficient, rather than "working harder". So, if the findings suggest anything about intelligence (the article is so vague it's really hard to tell one way or the other), its that introverts are on average less intelligent.
Actually, the completely appropriate description is that it's a *cryptopgrahic* hash algorithm. The key difference been that a cryptographic hash algorithm is difficult to invert (the "intentional" part of your statement) whereas other hash algorithms need not be (indeed many of the best algorithms for small hash tables are pretty easy to invert). They are critical components in most mechanisms for digital signing. Basically, with MD5 broken, you no longer have a secure way of verifying that the RPM's and.deb's installed on your system are the actual binaries and were distributed by the author of the packages. If you are using something like PGP with MD5 as the message digest, you no longer can trust that the messages you receive are actually from the sender baed on the cryptosystem .
So yeah, this has huge implications in the world of digital security and cryptography.
It looks like the article is describing something somewhat different from symlinks. That said, NTFS since Windows 2000 and on has already had the capability of doing symlinks through Reparse Points. Indeed, this is how mount points were done. Perhaps not the most efficient way to do things, but it works just great, and it's far more flexible than symlinks.
I'm not sure how you reach that conclusion. In fact, I've found the experience is quite the opposite. Because Yahoo uses humans to review ads before posting them, it tends to be very hard to spam. With Google it's much easier. Sure, you'll quickly get weighted down, but in the short term you get your ad out there, and you can always resubmit it under a different keyword later.
I read your paper, and I do think there is some interesting research there. However, it seems that the way you went about doing things is going to limit the relevance of these findings. In particular, even when you are comparing malloc/free, the free calls are generated automatically. So, what you are comparing here is the advantages of having amortized free operations vs. the disadvantage of having to compute when such operations are done at runtime. So, the end result is not a very suprising conclusion: you need to amortize a lot of free operations to make up for the cost of having to compute when memory can be freed.
Even that conclusion is somewhat limited, as the cost of malloc/free and GC'd memory management are limited to the specific tools you evaluated, which are clearly not the best available (although they are decent). However, there are lots of ways that I'd imagine the GC'd memory management could be improved.
It also ignores the fact in a multithreaded environment, that the cost of computing the root set can be off-loaded to another thread (or threads). Particularly in a multiprocessing and I/O intensive context, this could all but eliminate the impact of these calculations on application thoroughput.
The big question that crosses my mind though is this: your paper clearly demonstrates that at least with certain constraints, there are times when at *compile time* a compiler or JIT can and should determine when certain objects can and should be freed. This is kind of what Mustang's escape analysis achieves (although in a more limited way). An even more clever compiler can determine the set of objects which *might* be freeable. If done right, this can be integrated with the allocation of said objects to further simply things and improve performance. The end result would still qualify as an automatic memory managed system, and assuming some degree of non-determinism for the timing of deallocations, it would still be a garbage collected environment. Such a system should in theory be able to combine the performance benefits of your malloc/compiler-assigned free case with the amortized free benefits of your GC case.
I totally got the point of your comment. The comment is totally irrelevant to the article. If you'd read the article, you would note the section titled "Stack Allocation", which actually takes up roughly half of the article. They do compare against manually doing stack allocation in C++. They cite it as being faster than heap allocation and of course infinitely faster than heap deallocation. Then they talk about how Mustang now does stack allocation where possible.
On the other hand, it's even quicker to not read the article and just make off-hand generalizations based on the Slashdot summary.;-)
The "fatness" perception of JVM's is actually pretty misleading. A great deal of that memory is not part of the working set for the JVM. For more details check out this excellent article which mentions how Mustang now reduces the "perceived" memory requirements of the JVM by as much as half by simply changing how jar's are loaded.
That said, Netbeans does tend to be a hideous memory pig.
Actually, you got the idea behind group wisdom totally wrong. In fact group wisdom tends to be poorer when the group is influenced by "leaders" or "experts".
Groups/information markets aren't always "wise". The circumstances where they perform best is when there are no experts because the relevant information is spread diffusely amongst the members of the group.
Yahoo has launched one of these to the public back at in March of 2005 at the O'Reilly Etech conference. They actually had a contest where the top performer got a Mac-mini.
For the most part, the ISA remains unchanged, and to the extent that it is improved, it's important to keep in mind that internally neither Intel nor AMD's chips are actually executing x86 code, as it's translated into an internal instruction set. The same can be said for the registers. While the x86 ISA leaves you with hardly any registers, Intel and AMD's chips do register renaming to hundreds of hardware registers.
Sure, there is some overhead in all that translation, and having a broken instruction set does make the job for compilers and CPU's more difficult, no matter what you do, but you'd be surprised at how little impact that has on x86 performance.
In reality, going from 7 to 16 GPR's is not nearly the win you might think it is, To get really excited you need to be talking about 32 or 64 registers. Even then, because the number of registers in the ISA no longer correspond to registers in the hardware, real performance advantages may not be what they seem.
- STORAGE: Everything goes on a SAN, if not more than one. Don't waste your time with anything less.
I'm sorry, but I gotta disagree with this. I'd say don't waste your money on the SAN unless you know you need it. A cheaper approach a mail solution is to have a good replication setup.
Concluding that Yahoo's index has to be smaller because they return fewer results seems a bit overzealous.
No, it's accurate. They're testing Yahoo's claim of how many pages they've indexed, which just means that all indexed pages that contain the requested words should be returned from the search request. If yahoo returns fewer unique pages, yahoo has indexed fewer pages.
No. If they were using the same algorithm to determine which pages to return, you'd be correct. They don't. It's entirely possible that Yahoo's algorithm is just being more discerning. This is particularly likely as in order to preserve relevance it is usually necessary to make your algorithm more discerning when you increase your index.
Given that the queries were made up of random pairings of words from Ispell which returned less than 1000 results (we're talking less than a millionth of either index), it's quite likely that most of the queries were nonsensical and even in the result sets that did come back, it'd be hard to argue any of the pages are relevant. A perfect search engine would probably return nothing.
General Public License, the most popular license for free software that gives users the freedom... In contrast, an open-source license, like IBM's, is copyrighted.
Let's see all the ways that that sentence is poorly worded:
Whether the licenses themselves are copyrighted isn't what he's talking about, but whether the licensed code is copyrighted.
For the GPL to work, the GPL'd code has to be copyrighted.
The difference between "free software" and "open source software" has nothing to do with licenses or legalese, but rather the rational behind said licenses.
The really big one: The GPL is an open source license!!
Now, you might blame try to blame the writer for some of the mistakes here, but the real crime was committed by the editor. That's just pathetic.
Ah this is the point. Sigh. Actually, Yahoo did it several months before Google did it. However, by posting this dupe Slashdotters can choose to believe that Google did it first.
Actually, it's got lots of Yahoo specific features beyond that (although that is a good feature). You have the "Save to My Web" button, the translation button (uses Babelfish), the Yahoo bookmarks stuff, the Yahoo Mail support. It also has a whole host of links to Yahoo services, although I'd just as soon just use bookmarks for them. The only non-Yahoo specific stuff it has are some of the reference links (and actually they are all under Yahoo Reference, even if some are from 3rd partis).
Not only is this a dupe that of a story from like four months ago, but the story submitter appears to have not noticed the dates and therefore confused who is in who's footsteps.
You should try the toolbar. It has a number of handy features. The "Save to My Web" button is very handy, as are the Yahoo Bookmarks. I also like having the "Translate Page" button for pages in languages I don't know. Also nice to have an "Add to My Yahoo" button for RSS feeds. The integrated Calendar support is handy as Sunfire still isn't quite there yet.
I gotta say, the GPL strikes me as being a far better bargain for whomever is producing copyrighted works (which might be a developer, might be a business, who knows). It gives them the one thing that might be as valuable as the copyrights to their own code: access to the source of derivative works.
Yahoo posts a news story about their biggest competitor's protection mechanism being broken less than 24 hours after release.
Sigh.... welcome to slashdot. Enjoy the show.
Actually, as is shown at the top of the article, the story was actually published on PC Magazine. Yahoo doesn't have it's own news gathering service, they simply aggregate content from others.
Sigh, that is exactly the problem. They claim they are "integrated" the same way Microsoft claims you can't have a web browser outside of the core operating system....and the same way both Google and Yahoo claim to have integrated pop-up blocking in with their toolbars. Go ahead. Scream about it.
Ultimately, all computer programs are the results of a combination of instructions. I suspect that for almost any piece of software you don't actually use all of it's functionality. You might not like the combination of one particular package. That doesn't make it "junk".
Slashdot Mirror
More Brain activity != Clever.
Indeed, it generally means you aren't too clever. The smarter brains tend to demonstrate less activity due to increased efficiency (or at least that's the theory).
Ahhhhh I like stories like this that tell me I'm smarter than everbody else!
I'm not sure, but it seems like you may have misinterpreted the findings. Keep in mind that more intelligent people tend to demonstrate less brain activity when problem solving. The belief is that their brains are simply more efficient, rather than "working harder". So, if the findings suggest anything about intelligence (the article is so vague it's really hard to tell one way or the other), its that introverts are on average less intelligent.
Actually, the completely appropriate description is that it's a *cryptopgrahic* hash algorithm. The key difference been that a cryptographic hash algorithm is difficult to invert (the "intentional" part of your statement) whereas other hash algorithms need not be (indeed many of the best algorithms for small hash tables are pretty easy to invert). They are critical components in most mechanisms for digital signing. Basically, with MD5 broken, you no longer have a secure way of verifying that the RPM's and .deb's installed on your system are the actual binaries and were distributed by the author of the packages. If you are using something like PGP with MD5 as the message digest, you no longer can trust that the messages you receive are actually from the sender baed on the cryptosystem .
So yeah, this has huge implications in the world of digital security and cryptography.
It looks like the article is describing something somewhat different from symlinks. That said, NTFS since Windows 2000 and on has already had the capability of doing symlinks through Reparse Points. Indeed, this is how mount points were done. Perhaps not the most efficient way to do things, but it works just great, and it's far more flexible than symlinks.
Hence, Yahoo is for spammers
I'm not sure how you reach that conclusion. In fact, I've found the experience is quite the opposite. Because Yahoo uses humans to review ads before posting them, it tends to be very hard to spam. With Google it's much easier. Sure, you'll quickly get weighted down, but in the short term you get your ad out there, and you can always resubmit it under a different keyword later.
If you remember GMail, when it was first released, didn't have very good support for other browsers but over time they worked on support for them.
If you remember Yahoo Mail, when it was first released, it already supported IE and Firefox. I'm pretty sure the same was true of GMail.
I read your paper, and I do think there is some interesting research there. However, it seems that the way you went about doing things is going to limit the relevance of these findings. In particular, even when you are comparing malloc/free, the free calls are generated automatically. So, what you are comparing here is the advantages of having amortized free operations vs. the disadvantage of having to compute when such operations are done at runtime. So, the end result is not a very suprising conclusion: you need to amortize a lot of free operations to make up for the cost of having to compute when memory can be freed.
Even that conclusion is somewhat limited, as the cost of malloc/free and GC'd memory management are limited to the specific tools you evaluated, which are clearly not the best available (although they are decent). However, there are lots of ways that I'd imagine the GC'd memory management could be improved.
It also ignores the fact in a multithreaded environment, that the cost of computing the root set can be off-loaded to another thread (or threads). Particularly in a multiprocessing and I/O intensive context, this could all but eliminate the impact of these calculations on application thoroughput.
The big question that crosses my mind though is this: your paper clearly demonstrates that at least with certain constraints, there are times when at *compile time* a compiler or JIT can and should determine when certain objects can and should be freed. This is kind of what Mustang's escape analysis achieves (although in a more limited way). An even more clever compiler can determine the set of objects which *might* be freeable. If done right, this can be integrated with the allocation of said objects to further simply things and improve performance. The end result would still qualify as an automatic memory managed system, and assuming some degree of non-determinism for the timing of deallocations, it would still be a garbage collected environment. Such a system should in theory be able to combine the performance benefits of your malloc/compiler-assigned free case with the amortized free benefits of your GC case.
I totally got the point of your comment. The comment is totally irrelevant to the article. If you'd read the article, you would note the section titled "Stack Allocation", which actually takes up roughly half of the article. They do compare against manually doing stack allocation in C++. They cite it as being faster than heap allocation and of course infinitely faster than heap deallocation. Then they talk about how Mustang now does stack allocation where possible.
;-)
On the other hand, it's even quicker to not read the article and just make off-hand generalizations based on the Slashdot summary.
The "fatness" perception of JVM's is actually pretty misleading. A great deal of that memory is not part of the working set for the JVM. For more details check out this excellent article which mentions how Mustang now reduces the "perceived" memory requirements of the JVM by as much as half by simply changing how jar's are loaded.
That said, Netbeans does tend to be a hideous memory pig.
Congratulations on posting a comment without RTFA! ;-)
The article talks specifically about this and cites how the Mustang JVM does stack allocation.
Actually, you got the idea behind group wisdom totally wrong. In fact group wisdom tends to be poorer when the group is influenced by "leaders" or "experts".
Groups/information markets aren't always "wise". The circumstances where they perform best is when there are no experts because the relevant information is spread diffusely amongst the members of the group.
Yahoo has launched one of these to the public back at in March of 2005 at the O'Reilly Etech conference. They actually had a contest where the top performer got a Mac-mini.
For the most part, the ISA remains unchanged, and to the extent that it is improved, it's important to keep in mind that internally neither Intel nor AMD's chips are actually executing x86 code, as it's translated into an internal instruction set. The same can be said for the registers. While the x86 ISA leaves you with hardly any registers, Intel and AMD's chips do register renaming to hundreds of hardware registers.
Sure, there is some overhead in all that translation, and having a broken instruction set does make the job for compilers and CPU's more difficult, no matter what you do, but you'd be surprised at how little impact that has on x86 performance.
In reality, going from 7 to 16 GPR's is not nearly the win you might think it is, To get really excited you need to be talking about 32 or 64 registers. Even then, because the number of registers in the ISA no longer correspond to registers in the hardware, real performance advantages may not be what they seem.
- STORAGE: Everything goes on a SAN, if not more than one. Don't waste your time with anything less.
I'm sorry, but I gotta disagree with this. I'd say don't waste your money on the SAN unless you know you need it. A cheaper approach a mail solution is to have a good replication setup.
No. If they were using the same algorithm to determine which pages to return, you'd be correct. They don't. It's entirely possible that Yahoo's algorithm is just being more discerning. This is particularly likely as in order to preserve relevance it is usually necessary to make your algorithm more discerning when you increase your index.
Given that the queries were made up of random pairings of words from Ispell which returned less than 1000 results (we're talking less than a millionth of either index), it's quite likely that most of the queries were nonsensical and even in the result sets that did come back, it'd be hard to argue any of the pages are relevant. A perfect search engine would probably return nothing.
...that the GPL isn't an open source license?
General Public License, the most popular license for free software that gives users the freedom... In contrast, an open-source license, like IBM's, is copyrighted.
Let's see all the ways that that sentence is poorly worded:
- Whether the licenses themselves are copyrighted isn't what he's talking about, but whether the licensed code is copyrighted.
- For the GPL to work, the GPL'd code has to be copyrighted.
- The difference between "free software" and "open source software" has nothing to do with licenses or legalese, but rather the rational behind said licenses.
- The really big one: The GPL is an open source license!!
Now, you might blame try to blame the writer for some of the mistakes here, but the real crime was committed by the editor. That's just pathetic.What is the news here?
Fair point, as this is a dupe from February.
Yahoo did it because Google did it.
Ah this is the point. Sigh. Actually, Yahoo did it several months before Google did it. However, by posting this dupe Slashdotters can choose to believe that Google did it first.
Actually, it's got lots of Yahoo specific features beyond that (although that is a good feature). You have the "Save to My Web" button, the translation button (uses Babelfish), the Yahoo bookmarks stuff, the Yahoo Mail support. It also has a whole host of links to Yahoo services, although I'd just as soon just use bookmarks for them. The only non-Yahoo specific stuff it has are some of the reference links (and actually they are all under Yahoo Reference, even if some are from 3rd partis).
Not only is this a dupe that of a story from like four months ago, but the story submitter appears to have not noticed the dates and therefore confused who is in who's footsteps.
You should try the toolbar. It has a number of handy features. The "Save to My Web" button is very handy, as are the Yahoo Bookmarks. I also like having the "Translate Page" button for pages in languages I don't know. Also nice to have an "Add to My Yahoo" button for RSS feeds. The integrated Calendar support is handy as Sunfire still isn't quite there yet.
All in all, it's got something for everyone.
...just kidding!
I gotta say, the GPL strikes me as being a far better bargain for whomever is producing copyrighted works (which might be a developer, might be a business, who knows). It gives them the one thing that might be as valuable as the copyrights to their own code: access to the source of derivative works.
I'm surprised noone has mentioned Yahoo's product search.
ROTFL!
Yahoo posts a news story about their biggest competitor's protection mechanism being broken less than 24 hours after release.
Sigh.... welcome to slashdot. Enjoy the show.
Actually, as is shown at the top of the article, the story was actually published on PC Magazine. Yahoo doesn't have it's own news gathering service, they simply aggregate content from others.
Sigh, that is exactly the problem. They claim they are "integrated" the same way Microsoft claims you can't have a web browser outside of the core operating system. ...and the same way both Google and Yahoo claim to have integrated pop-up blocking in with their toolbars. Go ahead. Scream about it.
Ultimately, all computer programs are the results of a combination of instructions. I suspect that for almost any piece of software you don't actually use all of it's functionality. You might not like the combination of one particular package. That doesn't make it "junk".