Well, yeah, they're not claiming they invented radix sort. They're claiming that their GPU implementation of radix sort runs about 4x as fast as the CPU implementation you describe.
Well, by "it", I mean "any software they receive". The GNU position is basically that end users should always be able to modify any software they receive.
The GPL is focused primarily on freedoms of end users rather than developers. From that perspective, the freedom the BSD license gives to developers to put proprietary licenses on their code isn't very pro-freedom for their users, since their users are now prohibited from modifying it.
This is somewhat opposite from the usual argument I hear from BSD advocates. They argue that putting it in the license, like the GPL, is unnecessarily legalistic and problematic, and instead the license should be BSD, and encouragement to contribute back should be done via social pressure, PR, etc. Some BSD advocates, at least, argue that this approach overall results in at least as many return contributions, without being legalistic about it.
Somehow I rarely find them informative either, which is maybe a bigger problem with videogame journalism overall. Not that other kinds of reviews are always great, but I do sometimes learn things about a book, film, or album by reading a review---sometimes from both positive and negative reviews, and sometimes the review even adds an interesting perspective and background information for things I've actually already read/watched/listened-to. I very rarely get that feeling from game reviews.
Of course, this may be partly the games' fault as well. AAA titles focus so much on things like graphics and polish that the reviewers inevitably spend a lot of their time reviewing that, plus basic mechanical things like, "does the damn thing work", and "is multiplayer completely broken", leaving no time/interest for things that would be more interesting to me, like, "does this game break new gameplay ground?", and "in what ways are the core mechanics used in this game an interesting twist on games it draws influence from?"
It discourages the use of media that can't run on all the major platforms. Something that was "works everywhere except Linux" might, unfortunately, have a chance of catching on, due to not enough webdevs caring about Linux.
For formats in particular, though, I see some advantages to supporting the same formats everywhere. If "Firefox on Windows" and "Firefox on Linux" support different formats, that means that some webpages will be Windows-only or Linux-only, which defeats much of the cross-platform purpose of the web.
You can also achieve that, if your platform vendor is willing to put in the customization effort (Firefox is open-source, after all). Debian does this, modifying Firefox so the Debian version uses the already-installed system libraries for things like libpng.
The Gimp is just for porting reasons, I think. They wrote it in GTK+ initially, which is the native widget-set for their original target (GNOME), and porting it to different widget sets is work nobody's been particularly interested in doing.
It's a little broader: it appears to allow any "free-to-view" internet video to use the codecs royalty-free, even if that video is being used to make money through e.g. ads. It does exclude video where users are paying to watch it, like Hulu.
From the license summary [pdf], which hasn't yet been updated to indicate the indefinite moratorium:
In the case of Internet broadcast (AVC video that is delivered via the Worldwide Internet to an end user for which the End User does not pay remuneration for the right to receive or view, i.e., neither title-by-title nor subscription), there will be no royalty during the first term of the License (ending December 31, 2010) and the following term (ending December 31, 2015), after which the royalty shall be no more than the economic equivalent of the royalties payable during the same time for free television.
Commercial usage, even without charging end users, off the internet, e.g. with free television broadcasts, still requires royalties.
That isn't a specifically H.264-related policy: Firefox doesn't use system codecs for anything, because they want the exact same experience on all platforms. For example, they use internal image decoders, rather than relying on OS services like OSX's CoreImage. The downside is that therefore OSX on Firefox doesn't support everything that CoreImage does, unlike with WebKit, which just passes off to the system decoder. The upside is that the list of image formats Firefox supports doesn't vary by platform.
How many computers do you know that plug into 7200V 3-phase outlets? It's pretty standard when plugging computers into power-limited circuits to measure the allocated power by amperage. That's how power allocations are quoted at colocation facilities, for example: you rent a rack with a 20-amp circuit, or with a 30-amp circuit, or whatever. Unless stated otherwise, in North America it's implied that the circuit is at 110-V.
It's emphasizing mindsets that aren't actually a good idea in many modern contexts, though. In particular, it often is best to use automatic memory management, or you end up with a morass of C++-style FooManagers tracking all your resources, like Firefox, in the process leaking many of them and running more inefficiently than a global GC would. Working on machines where instruction/cycle count is a main issue in efficiency also teaches bad habits, since in modern machines cache locality and memory latency are much bigger issues, and some 1980-era optimizations to reduce instruction count are now anti-optimizations.
Unfortunately this does seem to be the case in the market the AAA companies are going for. It's getting a little less true overall, though. An "MMO" with hilariously ancient voxel graphics made by one guy has racked up about $1m in sales, because the super-simple, low-overhead, and low-programmer-hassle graphics free him up to do interesting things with the gameplay.
These do seem to be "alternative" games, though--- I can't imagine the mainstream game-review mags giving such a game a glowing review.
It's possible game budgets have overextended, and I personally would welcome a move towards lower-budget games: these really huge budgets are somewhat stifling for innovation, because there is very little risk you can afford to take with a $50m+ game. If you made ten $5m games out of that money, you could try out some more interesting things, and you'd also have smaller teams that can inherently move a little more nimbly (it's very hard to steer a ship the size of the current AAA dev teams, and changing anything requires heroics).
Nonetheless, I'm not sure one big-budget failure is enough evidence of a turnaround. The film industry has had a few large-budget films that failed so badly they bankrupted studios also, but pundits' predictions that those films marked a peak in film budgets all proved to be wrong.
Traditionally, nVidia had high-quality but closed drivers for Linux, while ATI had a low-quality but open ones (they also had a closed one, but it was pretty bad too). The main change seems to be that ATI's released a lot more specs lately, and has devoted more attention to producing non-crappy Linux drivers.
The thing that made choose-your-own-adventure books interesting was essentially hacking a limited notion of interactivity into a non-interactive medium, by asking users to manually enact GOTOs. But on a computer, we have interaction sort of built in, so the hack is uninteresting. Sure, you can still do it, and people might still like reading them, but it's not really its own category of thing, and we've had it forever. You can do it with a set of HTML pages linked to each other, or before that, with hypercard pages, and people actually did so, a long time ago, and did it more interestingly.
It's tricky, because professors often do have a reasonably good justification. I mean, of all the physics textbooks out there, presumably the one the prof wrote himself is the one that covers the material closest to the way he thinks it should be covered. It's also almost certainly the textbook whose contents he's most familiar with, whose exercises he can most reliably answer questions about, etc.
One problem is that "information" is largely supposed to make things easier by giving you access to something that was already done: someone else already went out there and collected meticulous information on frog populations, so it's easier to get access to that information than go out and count frogs yourself. But as information multiplies, sometimes it really is easier to just count the damn frogs instead of making sense of the voluminous and often inconsistent frog literature.
Diderot noticed this in 1755, in a famous passage:
"As long as the centuries continue to unfold, the number of books will grow continually, and one can predict that a time will come when it will be almost as difficult to learn anything from books as from the direct study of the whole universe. It will be almost as convenient to search for some bit of truth concealed in nature as it will be to find it hidden away in an immense multitude of bound volumes."
Slashdot Mirror
Wallclock time: they're claiming that this is, in absolute numbers, the fastest sort in keys-per-second yet demonstrated on commodity hardware.
Well, yeah, they're not claiming they invented radix sort. They're claiming that their GPU implementation of radix sort runs about 4x as fast as the CPU implementation you describe.
Well, by "it", I mean "any software they receive". The GNU position is basically that end users should always be able to modify any software they receive.
The GPL is focused primarily on freedoms of end users rather than developers. From that perspective, the freedom the BSD license gives to developers to put proprietary licenses on their code isn't very pro-freedom for their users, since their users are now prohibited from modifying it.
This is somewhat opposite from the usual argument I hear from BSD advocates. They argue that putting it in the license, like the GPL, is unnecessarily legalistic and problematic, and instead the license should be BSD, and encouragement to contribute back should be done via social pressure, PR, etc. Some BSD advocates, at least, argue that this approach overall results in at least as many return contributions, without being legalistic about it.
Somehow I rarely find them informative either, which is maybe a bigger problem with videogame journalism overall. Not that other kinds of reviews are always great, but I do sometimes learn things about a book, film, or album by reading a review---sometimes from both positive and negative reviews, and sometimes the review even adds an interesting perspective and background information for things I've actually already read/watched/listened-to. I very rarely get that feeling from game reviews.
Of course, this may be partly the games' fault as well. AAA titles focus so much on things like graphics and polish that the reviewers inevitably spend a lot of their time reviewing that, plus basic mechanical things like, "does the damn thing work", and "is multiplayer completely broken", leaving no time/interest for things that would be more interesting to me, like, "does this game break new gameplay ground?", and "in what ways are the core mechanics used in this game an interesting twist on games it draws influence from?"
It discourages the use of media that can't run on all the major platforms. Something that was "works everywhere except Linux" might, unfortunately, have a chance of catching on, due to not enough webdevs caring about Linux.
For formats in particular, though, I see some advantages to supporting the same formats everywhere. If "Firefox on Windows" and "Firefox on Linux" support different formats, that means that some webpages will be Windows-only or Linux-only, which defeats much of the cross-platform purpose of the web.
You can also achieve that, if your platform vendor is willing to put in the customization effort (Firefox is open-source, after all). Debian does this, modifying Firefox so the Debian version uses the already-installed system libraries for things like libpng.
The Gimp is just for porting reasons, I think. They wrote it in GTK+ initially, which is the native widget-set for their original target (GNOME), and porting it to different widget sets is work nobody's been particularly interested in doing.
It's a little broader: it appears to allow any "free-to-view" internet video to use the codecs royalty-free, even if that video is being used to make money through e.g. ads. It does exclude video where users are paying to watch it, like Hulu.
From the license summary [pdf], which hasn't yet been updated to indicate the indefinite moratorium:
Commercial usage, even without charging end users, off the internet, e.g. with free television broadcasts, still requires royalties.
That isn't a specifically H.264-related policy: Firefox doesn't use system codecs for anything, because they want the exact same experience on all platforms. For example, they use internal image decoders, rather than relying on OS services like OSX's CoreImage. The downside is that therefore OSX on Firefox doesn't support everything that CoreImage does, unlike with WebKit, which just passes off to the system decoder. The upside is that the list of image formats Firefox supports doesn't vary by platform.
I'm afraid your only choices are zero or infinite self-replicating replicators.
Apparently, the answer is yes.
In glorious future, we operate our computers as if they were theremins!
How many computers do you know that plug into 7200V 3-phase outlets? It's pretty standard when plugging computers into power-limited circuits to measure the allocated power by amperage. That's how power allocations are quoted at colocation facilities, for example: you rent a rack with a 20-amp circuit, or with a 30-amp circuit, or whatever. Unless stated otherwise, in North America it's implied that the circuit is at 110-V.
It's emphasizing mindsets that aren't actually a good idea in many modern contexts, though. In particular, it often is best to use automatic memory management, or you end up with a morass of C++-style FooManagers tracking all your resources, like Firefox, in the process leaking many of them and running more inefficiently than a global GC would. Working on machines where instruction/cycle count is a main issue in efficiency also teaches bad habits, since in modern machines cache locality and memory latency are much bigger issues, and some 1980-era optimizations to reduce instruction count are now anti-optimizations.
Unfortunately this does seem to be the case in the market the AAA companies are going for. It's getting a little less true overall, though. An "MMO" with hilariously ancient voxel graphics made by one guy has racked up about $1m in sales, because the super-simple, low-overhead, and low-programmer-hassle graphics free him up to do interesting things with the gameplay.
These do seem to be "alternative" games, though--- I can't imagine the mainstream game-review mags giving such a game a glowing review.
It's possible game budgets have overextended, and I personally would welcome a move towards lower-budget games: these really huge budgets are somewhat stifling for innovation, because there is very little risk you can afford to take with a $50m+ game. If you made ten $5m games out of that money, you could try out some more interesting things, and you'd also have smaller teams that can inherently move a little more nimbly (it's very hard to steer a ship the size of the current AAA dev teams, and changing anything requires heroics).
Nonetheless, I'm not sure one big-budget failure is enough evidence of a turnaround. The film industry has had a few large-budget films that failed so badly they bankrupted studios also, but pundits' predictions that those films marked a peak in film budgets all proved to be wrong.
That's Julius Caesar and Pope Gregory XIII to you.
Traditionally, nVidia had high-quality but closed drivers for Linux, while ATI had a low-quality but open ones (they also had a closed one, but it was pretty bad too). The main change seems to be that ATI's released a lot more specs lately, and has devoted more attention to producing non-crappy Linux drivers.
The thing that made choose-your-own-adventure books interesting was essentially hacking a limited notion of interactivity into a non-interactive medium, by asking users to manually enact GOTOs. But on a computer, we have interaction sort of built in, so the hack is uninteresting. Sure, you can still do it, and people might still like reading them, but it's not really its own category of thing, and we've had it forever. You can do it with a set of HTML pages linked to each other, or before that, with hypercard pages, and people actually did so, a long time ago, and did it more interestingly.
It's tricky, because professors often do have a reasonably good justification. I mean, of all the physics textbooks out there, presumably the one the prof wrote himself is the one that covers the material closest to the way he thinks it should be covered. It's also almost certainly the textbook whose contents he's most familiar with, whose exercises he can most reliably answer questions about, etc.
Screw Japan, free cephalopods!
One problem is that "information" is largely supposed to make things easier by giving you access to something that was already done: someone else already went out there and collected meticulous information on frog populations, so it's easier to get access to that information than go out and count frogs yourself. But as information multiplies, sometimes it really is easier to just count the damn frogs instead of making sense of the voluminous and often inconsistent frog literature.
Diderot noticed this in 1755, in a famous passage: