if other developers...have to do too much downgrading to get their games to run on the Revolution that system is likely to suffer the same fate as the Gamecube.
Far from developers dumbing down for it, Gamecube was powerful enough that Nintendo wanted ports from the PS2 to be graphically enhanced. This generation, cross-platform games have to be playable at HD resolution on the 360 and PS3. Hollywood (Revolution's GPU) won't have trouble running such games at standard TV resolution.
On the CPU side, each of the consoles is a bottleneck for cross-platform games.
PS3 will have the worst general-purpose multi-threading performance, since it has a single PPE core compared to 360's three and Revolution's two.
XBox 360 will be the weakest in physics operations, assuming the PS3's SPE units can be used for physics. Revolution, with its dedicated physics processor, will be in the top spot.
Revolution will have the worst single-threaded performance, assuming it uses the same PPE units as 360 and PS3. This is likely given Revolution's low price point and small form factor. However, if it's instead based on PowerPC 970, it would have the best single-threaded performance of the three.
If you have trouble figuring out a console controller perhaps you should go back to the field and settle in the soil with the other vegetables.
Standard controllers are natural now that I'm used to them, but I can still see them from the point of view of a non-gamer. It's not that they couldn't figure them out, it's just unnatural and takes more practice than it's worth. If you're not used to controllers, a good chunk of any game's difficulty is in the control scheme.
I can vaguely remember learning the original NES controller. It was hard, and it took a good bit of practice before I could consistantly get Mario past that first goomba. I'd actually try to get this move the controller to get him to jump. Both my younger brothers also hopped the controller around when they were learning.
Nowadays, it's not just all the buttons, it's the whole 3D platformer control issue. I haven't had much interest in console games since the SNES, simply because 3D platformers are much harder to interact with than 2D ones: figuring out where floating objects are in 3D space, suboptimal camera angles, mentally remapping the analog stick to the game world every time the camera angle changes, keeping 3D level layouts in your head so you don't get lost, etc. The Revolution controller is the first with actual 3D input, which can't help but make navigation in 3D space much easier and more intuitive.
What I fear will happen is that the games with the revolution will be considered to simple/dumbed down etc for the existing gamers to buy and therefore the necessary word of mouth will not be generated.
Nintendo releases games targeted to different audiences, so the core gamer need not fear. While the "non-game" Nintendogs is their heavy hitter with the casual crowd, they also have Mario and Luigi: Partners in Time a game which requires you to control four characters at once. Likewise, the Mario Party series and Twilight Princess are aimed at totally different audiences.
Ignoring core gamers, the non-gamers that learn about it and enjoy it will start spreading their own word-of-mouth. And non-gamers will be much more likely to listen to fellow non-gamers than to core gamers.
Bits as a performance measurement sort of went away with the PS2 generation.
For good reason. Dreamcast, GameCube, and XBox were all 32-bit consoles, and PS2 was 64-bit*. None surpassed the N64 bit-wise, so the bits were kept quiet.
*Sony claimed Emotion Engine was 128-bit, because it processed two 64-bit words per cycle, thus 128 bits per cycle. A true 128-bit processor has a 128-bit word size.
And, loading time is utterly irrelevant. It will take you longer to apply a decent filter to an image - even with macosx and the graphics core - than it will to start photoshop.
Not true. Since Aperture uses Core Image, filters are real-time. They're hardware accelerated, so Aperture requires a beefy graphics card.
I don't see force feedback, but there's a "home" button for recalibration. Want to sit somewhere else in a multiplayer game without telling your character to walk off a cliff? Just hold down home while you do it.
The expansion port gets lost amid the hubbub over the motion/tilt sensing (which is not gyroscopic, from my understanding); it lets developers add whole controller segments. Nintendo's going to make good use of it, but since many core functions are handled by the Revolution controller (e.g. power supply, console connection, motion sensing), it could also make custom controller segments cheap enough to be designed for and included with individual games: racing wheels, fishing rods, custom button configurations, etc. Some GB and GBA games do this already, with rumble, tilt sensors, and even solar cells built into the cartridge.
The end-user benifit is the use of Claria-supported programs for free, instead of paying for the ad-free version.
Claria lets developers release an ad-supported version of their software without developing their own ad framework, thus allowing them to release free software and still make a profit. Popular programs Go!Zilla and Kazaa both used Claria at one point.
Don't be quick to discount Nintendo's influence; they have quite the history of changing controllers forever. Their previous innovations include the d-pad (Game & Watch), shoulder buttons (SNES), the analog stick (N64), expansion ports (N64), and rumble (N64). The analog stick is especially notable, as it was also thought to be just a gimmick when it was introduced. Also, with the traditional-style shell Revolution has the best of both worlds; a standard-style controller with full motion and tilt sensing.
The RC is the first controller with three-dimensional input; it supports six degrees of motion (three displacement and three rotational), while an analog stick only supports two. Two analog sticks together only cover four. Factor in the Revolution's d-pad and analog stick, and you have 10 degrees of motion while making three-dimensional navigation more intuitive.
I bet Nintendo will opt for a subscription based service
Possibly, but I doubt it because wi-fi connection could easily have been subscription-based, and Nintendo opted to give it away.
For size, I bet the Revolution will be the smallest again, but traditionally Nintendo isn't big on "slick" designs
True, but the Game Boy Micro and the DS redesign (as well as the Revolution design itself) indicate that they're learning. Also, Nintendo's said that the Revolution will be its smallest console yet.
your really talking about a game console with the same specs as each of the new game systems, albiet a little anemic compared to the PS3 or Xbox360, as was the Gamecube in its generation.
The PS3 and 360 will be amazing in streaming media operations (which translates to better graphics), but both use deep-pipelined CPU cores with unimpressive cache and small or nonexistant branching predictors, so branching performance suffers. Poor branching performance won't hurt graphics; but will limit processes like AI, game control code, and physics. With Nintendo's games-over-specs mentality, Broadway (the Revolution's CPU) will most likely have either a beefy cache and/or be PowerPC 970-based and thus have nice branching predictors, either of which would make it the most powerful next-gen for branching-intensive code. Leaked specs indicate a dedicated physics processing unit, which would make Revolution games feel the most realistic by a good margin if true. Microsoft and Sony aimed for super-powerful graphic machines, and they succeeded; Nintendo just wants a game machine, and they'll make a good one.
Nintendo's game-centric mentality is at fault for the GameCube's perceived weakness; their target audience doesn't care about specs, so they didn't trumpet the GameCube's specs much. Spec-wise, GameCube is close to the XBox, and they're both far ahead of the PS2. In my experience, GameCube games are much smoother-looking than XBox or PS2.
one thing Nintendo NEEDS TO PAY ATTENTION TO, the gamer market is now longer composed of people 15 and under. Nintendo really needs to start focusing on adult gamers
In general, I don't think Nintendo makes games for kids; they make games for everyone that are appropriate for kids. However, some Nintendo games have parts that would be disturbing to some children. Ocarina of Time has bloodstained floors, walls made out of bones, and blood being coughed up. In the Wind Waker finale, Link embeds his sword in Ganondorf's skull.
That said, they do have an image problem among gamers in a certain age range in the U.S. I suspect cultural differences make it hard for them to see how there games are perceived over here. However, I don't see them changing focus to the adult market as long as their family-friendly games keep selling as well as they do. Unlike Microsoft and Sony, Nintendo has to be profitable in the gaming market to survive; and
By skimping on cache and using chips with no branching predictor, PS3 and 360 are "getting rid of the whole hardware capability. It won't be harder to implement a feature, it will be impossible."
From current rumored specs and Nintendo's gameplay-over-graphics mentality, Revolution will likely have some combination of a larger CPU cache, a 970-based CPU (instead of the PPE used in the PS3 and 360), and an Ageia physics processor.
None of these would improve streaming media performance (the main focus of the competitors' hardware), but any of these would make the Revolution the most powerful console on the block for branching-intensive code; code for things like AI, physics, and game-control code.
That said, 360 and PS3 will dominate in the streaming media applications they're designed for; likely enough to match or surpass the Revolution's graphics at the much higher resolutions required for HD support.
Nintendo's already said there will be a "standard" style shell attachment, and the Revolution works with Gamecube controllers; a straight control port would be trivial.
...you can add and extract items without any boxing or unboxing. That's arguably...the main advantage of C#'s generics over Java's
Having used Java's generics since JSE 1.5b1, I can attest that the minor performance and consistency loss that comes from Java's inability to parameterize primitive types is trivial next to the problems of type erasure.
Java's erasure mechanism limits its generics to compile-time checks and cast avoidance. Working around erasure requires client code to pass in a type token, which your code then has to keep track of. C# just handles everything for you.
My suspicion is they were planning to release 8.5 (the free version) for their 10th birthday, and it got pushed back. As a stopgap, they gave away activation codes, effectively releasing a free version of Opera.
Open source means - someone else may change it - so serious bug can be fixed next hour, no next month.
This is assuming that the closed-source development team is slow to fix serious bugs, and that the open-source team will accept patches and release an update when they do.
Opera has a fast dev team, so security problems will often get fixed faster in Opera than in Firefox. I've seen quite a few posters here that have had better experience getting problems fixed in Opera than in Firefox, and none the opposite way. See my post here.
The biggest Opera security problem recently was the news-making cross-platform popup origin spoofing vulnerability. Secuna released a security advisory to the general public on 6/21/2005. Opera fixed it with Opera 8.01 on 6/16/2005, five days before the report was released. Firefox, on the other hand, took until 7/12/2005 to fix it in Firefox 1.0.5, three weeks after the advisory's public release.
Also, this vulnerability was the biggest security fix in Opera 8.01; Firefox 1.0.5 fixed 9 vulnerabilities that were more severe.
In my experience, people get fed up with IE and just switch...nobody was holding out for free Opera.
Sounds familiar. I switched to Opera before Firefox was an option, then to Firefox when it came out.
After a few months, though, I got fed up with Firefox and switched back to Opera.
And, Opera's ads were indeed a deterrent. I like a compact, minimal interface, and the banner ads unnecessarily took up two toolbar-heights of space. When Opera started using Google text ads instead my biggest reason for staying with Firefox was gone; the text ads were slim and unobtrusive enough that I hardly noticed when they went away.
Don't get me wrong; they're both great. Firefox is massively customizable, improving fast, and has functionality I miss when I'm using Opera. If I ever want to add my own functionality, I'll be right there in Firefox learning XUL. But Opera is faster, lighter, more stable (IMX), works with more sites (again, IMX), has much better state persistance, and comes with just about everything I want out of the box. It also has features I use a lot that Firefox can't duplicate yet: spatial navigation, smooth image zooming, quick preferences, etc.
Firefox's compatibility and stability will improve, and extensions/upgrades will probably come out to add the Opera features I like. I'll check it out again when 1.1 comes out, and in future versions if 1.1's not good enough; but for now I'm sticking with Opera.
Firefox can indeed do a regular Google search, just like Opera does; type "google firefox". It's also easier to add new search engines in Firefox than in Opera.
FYI, I was a dedicated Opera user until Firefox came out, switched to Firefox for a few months, then moved back to Opera when I got tired of Firefox's hassle, incompatibility, and bloat.
Having to wildly gesture for every game would just tire people out.
If you have to wildly gesture, it's the game designer's fault, not the controller's. If used right, the Revolution controller can make complex interfaces more intuitive without too much motion.
For example, 3D platforming navigation is traditionally complex, since you have to deal with 3 movement axes and 2 rotation axes. Here's a 3D platformer navigation scheme using only the Revolution's motion sensor and without too much motion:
You start out holding your controller like you would for any other system; this becomes your home position. To start walking, move the controller slightly in the direction you want to walk, moving it back to home to stop. This simplifies non-forward motion, since you can strafe and move backwards without holding a special rotation-locking button. To look around, point the front of the controller gently in the direction you want to look, again moving back to home to stop. Want to duck? Move it down. Want to jump? Just bob the controller slightly.
You get the idea. The revolution can be used on the same principle as an analog stick, using only slight motions away from the center; only in 6 axes instead of 2.
I actually have personal experience with how intuitive this controller should be. When I was little and first started playing NES games, I would bob the controller whenever I wanted the character to jump. If I wanted to jump over, I'd bob up and to the right. I would have been right at home with one of these.
It also sounds like Nintendo's presence in the console market may continue to slide further into a niche with the Revolution
Quite possibly, but Nintendo's goal with the Revolution is to appeal to the casual gamer. So, while their target audience may indeed be a niche from the perspective of the current market, they see the current market itself as a niche in a much broader potential market.
What'll make cross-platform work hard is they're difficult in different ways.
The 360 has three PPU cores, so without multithreading you can only use a fraction of the available power. With the PS3 you instead have to do low-level SPE unit programming, and any code that can't be adapted has to run on the single PPU.
Those are the big CPU difficulties, but thethe PPU's used in both systems have weak/nonexistant branching predictors, which lowers production costs. Poor branching performance doesn't hurt streaming media (they're excellent for graphics), but 360 and PS3 won't be much better at AI than current systems.
Hardly any information out about Revolution's CPU, but it's the last chance for easy development or truly next-gen AI. Keep in mind though, that Nintendo's the most budget-concious of the three.
Slashdot Mirror
Far from developers dumbing down for it, Gamecube was powerful enough that Nintendo wanted ports from the PS2 to be graphically enhanced. This generation, cross-platform games have to be playable at HD resolution on the 360 and PS3. Hollywood (Revolution's GPU) won't have trouble running such games at standard TV resolution.
On the CPU side, each of the consoles is a bottleneck for cross-platform games.
PS3 will have the worst general-purpose multi-threading performance, since it has a single PPE core compared to 360's three and Revolution's two.
XBox 360 will be the weakest in physics operations, assuming the PS3's SPE units can be used for physics. Revolution, with its dedicated physics processor, will be in the top spot.
Revolution will have the worst single-threaded performance, assuming it uses the same PPE units as 360 and PS3. This is likely given Revolution's low price point and small form factor. However, if it's instead based on PowerPC 970, it would have the best single-threaded performance of the three.
I can vaguely remember learning the original NES controller. It was hard, and it took a good bit of practice before I could consistantly get Mario past that first goomba. I'd actually try to get this move the controller to get him to jump. Both my younger brothers also hopped the controller around when they were learning.
Nowadays, it's not just all the buttons, it's the whole 3D platformer control issue. I haven't had much interest in console games since the SNES, simply because 3D platformers are much harder to interact with than 2D ones: figuring out where floating objects are in 3D space, suboptimal camera angles, mentally remapping the analog stick to the game world every time the camera angle changes, keeping 3D level layouts in your head so you don't get lost, etc. The Revolution controller is the first with actual 3D input, which can't help but make navigation in 3D space much easier and more intuitive. Nintendo releases games targeted to different audiences, so the core gamer need not fear. While the "non-game" Nintendogs is their heavy hitter with the casual crowd, they also have Mario and Luigi: Partners in Time a game which requires you to control four characters at once. Likewise, the Mario Party series and Twilight Princess are aimed at totally different audiences.
Ignoring core gamers, the non-gamers that learn about it and enjoy it will start spreading their own word-of-mouth. And non-gamers will be much more likely to listen to fellow non-gamers than to core gamers.
*Sony claimed Emotion Engine was 128-bit, because it processed two 64-bit words per cycle, thus 128 bits per cycle. A true 128-bit processor has a 128-bit word size.
My guess is that the "home" button is for calibration. With the Revolution's online capability, pause isn't always an option.
I don't see force feedback, but there's a "home" button for recalibration. Want to sit somewhere else in a multiplayer game without telling your character to walk off a cliff? Just hold down home while you do it.
The expansion port gets lost amid the hubbub over the motion/tilt sensing (which is not gyroscopic, from my understanding); it lets developers add whole controller segments. Nintendo's going to make good use of it, but since many core functions are handled by the Revolution controller (e.g. power supply, console connection, motion sensing), it could also make custom controller segments cheap enough to be designed for and included with individual games: racing wheels, fishing rods, custom button configurations, etc. Some GB and GBA games do this already, with rumble, tilt sensors, and even solar cells built into the cartridge.
Wavebird was RF, so there's no reason not to use it for Revolution also.
The end-user benifit is the use of Claria-supported programs for free, instead of paying for the ad-free version.
Claria lets developers release an ad-supported version of their software without developing their own ad framework, thus allowing them to release free software and still make a profit. Popular programs Go!Zilla and Kazaa both used Claria at one point.
Don't be quick to discount Nintendo's influence; they have quite the history of changing controllers forever. Their previous innovations include the d-pad (Game & Watch), shoulder buttons (SNES), the analog stick (N64), expansion ports (N64), and rumble (N64). The analog stick is especially notable, as it was also thought to be just a gimmick when it was introduced. Also, with the traditional-style shell Revolution has the best of both worlds; a standard-style controller with full motion and tilt sensing.
The RC is the first controller with three-dimensional input; it supports six degrees of motion (three displacement and three rotational), while an analog stick only supports two. Two analog sticks together only cover four. Factor in the Revolution's d-pad and analog stick, and you have 10 degrees of motion while making three-dimensional navigation more intuitive.
Possibly, but I doubt it because wi-fi connection could easily have been subscription-based, and Nintendo opted to give it away.
True, but the Game Boy Micro and the DS redesign (as well as the Revolution design itself) indicate that they're learning. Also, Nintendo's said that the Revolution will be its smallest console yet.
The PS3 and 360 will be amazing in streaming media operations (which translates to better graphics), but both use deep-pipelined CPU cores with unimpressive cache and small or nonexistant branching predictors, so branching performance suffers. Poor branching performance won't hurt graphics; but will limit processes like AI, game control code, and physics. With Nintendo's games-over-specs mentality, Broadway (the Revolution's CPU) will most likely have either a beefy cache and/or be PowerPC 970-based and thus have nice branching predictors, either of which would make it the most powerful next-gen for branching-intensive code. Leaked specs indicate a dedicated physics processing unit, which would make Revolution games feel the most realistic by a good margin if true. Microsoft and Sony aimed for super-powerful graphic machines, and they succeeded; Nintendo just wants a game machine, and they'll make a good one.
Nintendo's game-centric mentality is at fault for the GameCube's perceived weakness; their target audience doesn't care about specs, so they didn't trumpet the GameCube's specs much. Spec-wise, GameCube is close to the XBox, and they're both far ahead of the PS2. In my experience, GameCube games are much smoother-looking than XBox or PS2.
In general, I don't think Nintendo makes games for kids; they make games for everyone that are appropriate for kids. However, some Nintendo games have parts that would be disturbing to some children. Ocarina of Time has bloodstained floors, walls made out of bones, and blood being coughed up. In the Wind Waker finale, Link embeds his sword in Ganondorf's skull.
That said, they do have an image problem among gamers in a certain age range in the U.S. I suspect cultural differences make it hard for them to see how there games are perceived over here. However, I don't see them changing focus to the adult market as long as their family-friendly games keep selling as well as they do. Unlike Microsoft and Sony, Nintendo has to be profitable in the gaming market to survive; and
By skimping on cache and using chips with no branching predictor, PS3 and 360 are "getting rid of the whole hardware capability. It won't be harder to implement a feature, it will be impossible."
From current rumored specs and Nintendo's gameplay-over-graphics mentality, Revolution will likely have some combination of a larger CPU cache, a 970-based CPU (instead of the PPE used in the PS3 and 360), and an Ageia physics processor.
None of these would improve streaming media performance (the main focus of the competitors' hardware), but any of these would make the Revolution the most powerful console on the block for branching-intensive code; code for things like AI, physics, and game-control code.
That said, 360 and PS3 will dominate in the streaming media applications they're designed for; likely enough to match or surpass the Revolution's graphics at the much higher resolutions required for HD support.
Sources:
PPE's poor branching performance
Advantages of large cache or 970-based CPU
Revolution Ageia rumor
Nintendo's already said there will be a "standard" style shell attachment, and the Revolution works with Gamecube controllers; a straight control port would be trivial.
...you can add and extract items without any boxing or unboxing. That's arguably...the main advantage of C#'s generics over Java's
Having used Java's generics since JSE 1.5b1, I can attest that the minor performance and consistency loss that comes from Java's inability to parameterize primitive types is trivial next to the problems of type erasure.
Java's erasure mechanism limits its generics to compile-time checks and cast avoidance. Working around erasure requires client code to pass in a type token, which your code then has to keep track of. C# just handles everything for you.
My suspicion is they were planning to release 8.5 (the free version) for their 10th birthday, and it got pushed back. As a stopgap, they gave away activation codes, effectively releasing a free version of Opera.
Open source means - someone else may change it - so serious bug can be fixed next hour, no next month.
This is assuming that the closed-source development team is slow to fix serious bugs, and that the open-source team will accept patches and release an update when they do.
Opera has a fast dev team, so security problems will often get fixed faster in Opera than in Firefox. I've seen quite a few posters here that have had better experience getting problems fixed in Opera than in Firefox, and none the opposite way. See my post here.
The biggest Opera security problem recently was the news-making cross-platform popup origin spoofing vulnerability. Secuna released a security advisory to the general public on 6/21/2005. Opera fixed it with Opera 8.01 on 6/16/2005, five days before the report was released. Firefox, on the other hand, took until 7/12/2005 to fix it in Firefox 1.0.5, three weeks after the advisory's public release.
Also, this vulnerability was the biggest security fix in Opera 8.01; Firefox 1.0.5 fixed 9 vulnerabilities that were more severe.
Uh.. Firefox? I click once to highlight the address string, then Firefox lets me drag it to create a bookmark (something Opera doesn't do).
In Opera, you drag the page's tab or the page icon in the address bar instead of the address string.
In my experience, people get fed up with IE and just switch...nobody was holding out for free Opera.
Sounds familiar. I switched to Opera before Firefox was an option, then to Firefox when it came out.
After a few months, though, I got fed up with Firefox and switched back to Opera.
And, Opera's ads were indeed a deterrent. I like a compact, minimal interface, and the banner ads unnecessarily took up two toolbar-heights of space. When Opera started using Google text ads instead my biggest reason for staying with Firefox was gone; the text ads were slim and unobtrusive enough that I hardly noticed when they went away.
Don't get me wrong; they're both great. Firefox is massively customizable, improving fast, and has functionality I miss when I'm using Opera. If I ever want to add my own functionality, I'll be right there in Firefox learning XUL. But Opera is faster, lighter, more stable (IMX), works with more sites (again, IMX), has much better state persistance, and comes with just about everything I want out of the box. It also has features I use a lot that Firefox can't duplicate yet: spatial navigation, smooth image zooming, quick preferences, etc.
Firefox's compatibility and stability will improve, and extensions/upgrades will probably come out to add the Opera features I like. I'll check it out again when 1.1 comes out, and in future versions if 1.1's not good enough; but for now I'm sticking with Opera.
If only [Opera] could do flashblock.
6 03088
Here you go!
http://slashdot.org/comments.pl?sid=162788&cid=13
Firefox can indeed do a regular Google search, just like Opera does; type "google firefox". It's also easier to add new search engines in Firefox than in Opera.
FYI, I was a dedicated Opera user until Firefox came out, switched to Firefox for a few months, then moved back to Opera when I got tired of Firefox's hassle, incompatibility, and bloat.
Having to wildly gesture for every game would just tire people out.
If you have to wildly gesture, it's the game designer's fault, not the controller's. If used right, the Revolution controller can make complex interfaces more intuitive without too much motion.
For example, 3D platforming navigation is traditionally complex, since you have to deal with 3 movement axes and 2 rotation axes. Here's a 3D platformer navigation scheme using only the Revolution's motion sensor and without too much motion:
You start out holding your controller like you would for any other system; this becomes your home position. To start walking, move the controller slightly in the direction you want to walk, moving it back to home to stop. This simplifies non-forward motion, since you can strafe and move backwards without holding a special rotation-locking button. To look around, point the front of the controller gently in the direction you want to look, again moving back to home to stop. Want to duck? Move it down. Want to jump? Just bob the controller slightly.
You get the idea. The revolution can be used on the same principle as an analog stick, using only slight motions away from the center; only in 6 axes instead of 2.
I actually have personal experience with how intuitive this controller should be. When I was little and first started playing NES games, I would bob the controller whenever I wanted the character to jump. If I wanted to jump over, I'd bob up and to the right. I would have been right at home with one of these.
you are going to be swinging a 'bat'. throwing a 'ball'
Just don't slip!
It also sounds like Nintendo's presence in the console market may continue to slide further into a niche with the Revolution
Quite possibly, but Nintendo's goal with the Revolution is to appeal to the casual gamer. So, while their target audience may indeed be a niche from the perspective of the current market, they see the current market itself as a niche in a much broader potential market.
What'll make cross-platform work hard is they're difficult in different ways.
The 360 has three PPU cores, so without multithreading you can only use a fraction of the available power. With the PS3 you instead have to do low-level SPE unit programming, and any code that can't be adapted has to run on the single PPU.
Those are the big CPU difficulties, but thethe PPU's used in both systems have weak/nonexistant branching predictors, which lowers production costs. Poor branching performance doesn't hurt streaming media (they're excellent for graphics), but 360 and PS3 won't be much better at AI than current systems.
Hardly any information out about Revolution's CPU, but it's the last chance for easy development or truly next-gen AI. Keep in mind though, that Nintendo's the most budget-concious of the three.