• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

OG speculates on Revolution, WHOA

First lets start with some comments by Iwata.

"Of course, we are applying advances in technology. But when you use those advances just to boost the processing power, the trade-off is that you increase power consumption, make the machine more expensive and make developing games more expensive. When I look at the balance of that trade-off -- what you gain and what you lose -- I don't think it's good. Nintendo is applying the benefits of advanced technology, but we're using it to make our machines more power-efficient, quieter and faster to start.

We all know that multi core processors require alot of power, the PowerPC G5 is in the 100watt range. The move to the 90nm processor help decrease its power consumption to about 65watts. Going multi core only increases the amount of power needed, but the heat that comes from a multi core processor is the real problem. Its going to be a while before any chip is moved to the 65nm process. Which leads me to this.

Liquid-Metal Cooling Technology for CPU Cooling
As current cooling solutions reach their limits, the industry is in need of a unique advanced cooling solution. nanoCoolers' liquid-metal cooling loops have all the attributes to fill that need. Part of the uniqueness of our thermal solutions is in its simplicity. The cooling loop consists of liquid metal as the working fluid, a heat source exchanger, an ambient heat exchanger, an electromagnetic pump, and interconnecting tubing. There are also attaching mechanisms, a fan, housing structure, etc. but the simplicity of the solution is quite unique.
http://www.nanocoolers.com/technology_liquid.php

The boiling point of the material is in excess of 2000°C, the pics illustrates a small form factor. Which would be beneficial to the size of Revolution. Its also tech that doesn't produce alot noise.

liquidmetal_setup_diagram.gif


Combined this with IBM Strained Silicone directly on Insulator(SSDOI) process which reduces power consumption, without affecting performance.

The PS3 will require alot of power with so many connections, so Nintendo chose to provide only two USB ports and a proprietery AV component. You reduce power consumption because there is no harddrive to power. So Nintendo they go for onboard flash memory. There's only two moving part in the console(fan and DVD drive).

How do Nintendo plan on making it possible for indie developers to produce titles that look as good as the big boys. Well check out this patent, http://appft1.uspto.gov/netacgi/nph...s1=Nintendo.AS.&OS=AN/Nintendo&RS=AN/Nintendo

The patent basically says that a developer can pre-render a area of the game and convert it into a exact replica in realtime using this feature.

In one embodiment, we use a known technique called cube mapping to pre-render images defining a 3D scene. Cube mapping is a form of environment mapping that has been used in the past to provide realistic reflection mapping independent of viewpoint. For example, one common usage of environment mapping is to add realistic reflections to a 3D-rendered scene. Imagine a mirror hanging on the wall. The mirror reflects the scene in the room. As the viewer moves about the room, his or her viewpoint changes so that different objects in the room become visible in the mirror. Cube mapping has been used in the past or provide these and other reflection effects.

We use cube mapping for a somewhat different purpose--to pre-render a three-dimensional scene or universe such as for example a landscape, the interior of a great cathedral, a castle, or any other desired realistic or fantastic scene. We then add depth to the pre-rendered scene by creating and supplying a depth buffer for each cube-mapped image. The depth buffer defines depths of different objects depicted in the cube map. Using the depth buffer in combination with the cube map allows moving objects to interact with the cube-mapped image in complex, three-dimensional ways. For example, depending upon the effect desired, moving objects can obstruct or be obstructed by some but not other elements depicted in the cube map and/or collide with such elements. The resulting depth information supplied to a panoramically-composited cube map provides a complex interactive visual scene with a degree of 3D realism and interactivity not previously available in conventional strictly 2D texture mapped games.

This could shorten the development process. This combined with a development enviroment similar to GC(API), should help reduce development cost.

Now for the comments from Reggie about the Revolution controller, someone suggested that the GC controllers could be used to play those NES,SNES,N64 games. Only problem with this is that Reggie could of easily explained this as a fact. But its still to difficult to come to some type of conclusion. So I'll leave that portion alone.

I think Nintendo goal is to provide a powerful console, without having to include HDMI,DVI, (4)USB, Harddrive, Digital output, Analog AV all that requires money and power. You won't be limited in your ability to connect to these features except a harddrive , there just won't be 10 ports to do it.

Developers will have a recognizeable development environment, with additions that make it even easier to create visually outstanding titles.
 
I think it's about time to make an official speculation thread on the Revolution so that the front page isn't flooded with these threads.

This, however, is some great speculation, though. I like how you cover the size/cooling issues and the possible graphical issues.
 
Here we go again..

I'm going to assume the absolute worst for Nintendo's next console, so tha dissapointment is not a possibility.
 
without having to include HDMI, DVI

Iwata already said the sytem will have DVI.

Interesting piece, but some of that stuff is very speculative... the only thing that now makes sense is the fact that Revolution's CPU may not be multicore like the other two.
 
OG_Original Gamer said:
The patent basically says that a developer can pre-render a area of the game and convert it into a exact replica in realtime using this feature.

In one embodiment, we use a known technique called cube mapping to pre-render images defining a 3D scene. Cube mapping is a form of environment mapping that has been used in the past to provide realistic reflection mapping independent of viewpoint. For example, one common usage of environment mapping is to add realistic reflections to a 3D-rendered scene. Imagine a mirror hanging on the wall. The mirror reflects the scene in the room. As the viewer moves about the room, his or her viewpoint changes so that different objects in the room become visible in the mirror. Cube mapping has been used in the past or provide these and other reflection effects.

This sounds really interesting and Ive never heard of it before. Any techie input to confirm or deny would be greatly appreciated. :D
 
Most of my hardware know-how is on the processor detail side of things, so I can't really say definitely whether this is a wet-dream or valid speculation.

But, if any of it is steeped in fact, me likey.

I'm curious, though. Are you presuming that because the cooling system is so advanced, that Nintendo/IBM are going to cram a really fast 65nm chip in there, making up for the technology gap between Rev and the other two? Or does this design basically concede that the hardware is probably going to be less powerful than the other two across the board?

Also, is that cube-mapping effect you spoke of done strictly in software? If it is, couldn't MS and Sony copy that really easily by adding it to their own APIs and showing it off? (Could make sense that Nintendo wouldn't want to reveal that trick until after the competition finalized their first run of dev kits, perhaps ...)

Anyway, exciting stuff. Now I'm going to intentionally forget about all of it so Nintendo doesn't let me down when they announce that Revolution is going to be powered by a little Pikachu with electric clamps on its little red cheeks.
 
Juice said:
Most of my hardware know-how is on the processor detail side of things, so I can't really say definitely whether this is a wet-dream or valid speculation.

But, if any of it is steeped in fact, me likey.

I'm curious, though. Are you presuming that because the cooling system is so advanced, that Nintendo/IBM are going to cram a really fast 65nm chip in there, making up for the technology gap between Rev and the other two? Or does this design basically concede that the hardware is probably going to be less powerful than the other two across the board?

Also, is that cube-mapping effect you spoke of done strictly in software? If it is, couldn't MS and Sony copy that really easily by adding it to their own APIs and showing it off? (Could make sense that Nintendo wouldn't want to reveal that trick until after the competition finalized their first run of dev kits, perhaps ...)

Anyway, exciting stuff. Now I'm going to intentionally forget about all of it so Nintendo doesn't let me down when they announce that Revolution is going to be powered by a little Pikachu with electric clamps on its little red cheeks.

Actually, I believe because of this advanced CPU cooling technology Nintendo CPU will be just as powerful as the other consoles. One of the points of arguements about what performance the Revolution CPU is, is that the console small form factor prohibits it from having a multi core processor because they produce so much heat. With liquid metal cooling, having a small form factor console shouldn't be an issue.

As for as patent and its implementation, I would assume its in hardware. Cube mapping is a hardware feature, I'm not sure about depth buffering.

When I mentioned the 65nm process was pointing to the benefit lowered power consumption, you don't lose transistor count, IIRC.
 
OG_Original Gamer said:
Actually, I believe because of this advanced CPU cooling technology Nintendo CPU will be just as powerful as the other consoles. One of the points of arguements about what performance the Revolution CPU is, is that the console small form factor prohibits it from having a multi core processor because they produce so much heat. With liquid metal cooling, having a small form factor console shouldn't be an issue.

As for as patent and its implementation, I would assume its in hardware. Cube mapping is a hardware feature, I'm not sure about depth buffering.

When I mentioned the 65nm process was pointing to the benefit lowered power consumption, you don't lose transistor count, IIRC.
This whole speculation, in my opinion, is one of the most logically-written ones out of all the revolution speculation pieces. It also sounds a lot like something Nintendo would do. Of course, that doesn't mean they'll do it, but I wouldn't be surprised to see something similar. The explanation for the image patent is pretty good, too.

Unfortunately, we have to wait and see. But until then, I'll be playing the few of the last good Cube games and my DS.
 
Um, can someone please explain to me the difference between a regular heatpipe and this supposed amazing new cooling technology? Aside from the pump (wheras the heatpipe just runs on convection) I don't see it. This sort of technology has been used for a while with laptops and SFF PC's.

edit: re: cube mapping, for them to generate a scene using a depth map wouldn't the texture resolution have to be massive to avoid it looking like a blurry interpolated mess?
 
Wow I can't believe I just read that whole patent. Sounds very interesting, and definitly something "Nintendo". Seems like a great solution if it can be done.
 
Now this was some nice speculation, it actually had some thought put into it rather than the simple stuff that has been discussed before. Very nice work.
 
Wow. Great work. Some comments / questions:

- I recall talk about a removable harddrive, small form factor, maybe I mistake with some of the other two consoles?
- Connections, Engadget claimed to have had a chance to toy around with the device, do we have any pictures of all the connections?
- Cube mapping, interesting, I'm not familiar with the technology, will it be easy to implement in cross-platform development?
 
Chittagong said:
Wow. Great work. Some comments / questions:

- I recall talk about a removable harddrive, small form factor, maybe I mistake with some of the other two consoles?
- Connections, Engadget claimed to have had a chance to toy around with the device, do we have any pictures of all the connections?
- Cube mapping, interesting, I'm not familiar with the technology, will it be easy to implement in cross-platform development?


Engadget also found out that the dvd drive wasn't in the console. I'd say output connections aren't finalized, as is everything else with the console. They'd be smart to at least give us the option of a harddrive, but I have my doubts.
 
Nintendo are trying to make it easier for developers, which makes sense. Content makes a platform, and the easier it is to produce, the more content you have. Hopefully Nintendo will - like Sony - adopt industry standards like OpenGL ES (maybe even SDL?) as that would simplify porting between the PS3, Revolution and non-DirectX PC games. With the huge budgets for next gen games, I assume it's going to be all about porting, so maybe we'll see a OpenGL vs. DirectX part deux?
 
The other reason I like this cube-mapping idea is that it sounds like the exact inverse of Stereoscopic system. Instead of creating 'holographic' images that project out of the screen, developers will be able to take 2D images and 3D-ize them into the screen.

The results will probably be just as stunning, with little hardware overhead from the sounds of it.
 
Power...
I'm gonna have to look this stuff up but I distinctly remember IBM (before Nintendo announced they were working on the CPU, and before it was even being called the Revolution) saying that the next system WOULD be very power (akin to a supercomputer) and that one of the big requirements of the next Nintendo system was Nintendo wanted it to be small. So, while I'm not too sure if it'll really be THAT powerful, there is "hope" that Nintendo aren't skimping in this category.

Cube-Mapping...
Interesting. I too think that Nintendo will be leaning to more pre-rendering technologies. Not to say as a shortcut or "taking the cheap way out", but really...not too many of the masses are too too concerned with how it's done, just so long as it looks good. Look at the success of CG movies and "real" movies that use alot of pre-rendered computer animation. Not saying all games should be done this way, but instead of real-time rendered environments that end up looking repetitive and end up costing alot of money/time to developement cycles, wouldn't it be nice to have an easier/cheaper solution that smaller developers as well as bigger developers could take advantage of?

Cheaper, more obtainable goals for smaller developers...
I've been thinking about this for a bit. Nintendo said "use the GCN's current API" when starting to work on their next generation system. To me, this means the system may not be very powerful, but more efficient. Maybe it won't be seen "as powerful as" competiting systems, but it'll probably be built with less bottlenecks and more easily obtainable goals for even the smallest developers. Together with Cube-Mapping and other "shortcuts", these games can look just as good as games that do those things in real time...all for a FRACTION of the time/effort/money it takes to get it done.

Less Ports, driving down price & power consumption...
I agree. Less connections mean less for the chipset to think about. Less connections also means it'll be cheaper, run with less power requirements and can be built smaller. When I first was drawing up my own doodle for what I thought the Revolution would look like I instantly dropped pre-conceived idea's of a HD and I even dropped normal controller ports in favor of one EXT/Option port (to connect a seperate GCN port compatibility adaptor or HD, etc.) thinking that if it's ONLY 3 DVD cases thick that there would be no room. Then, when it was shown I was SHOCKED to find that not only was it smaller than my concept, but also the side of it opens up revealing all 4 GCN controller ports & 2 DigiCard slots! I mean DAMN...and it's gonna end up being even smaller? Anyways, the GCN ports (I don't imagine) will be used very much past full GCN BC, so that really means just an SD Card slot, 2 USB ports, 1 multi-out and the power jack. There's not very much power/connections/space required to accomedate for all of this so that leaves room for the machine to be small, cheap, less power consuming and still have room for a powerful chipset if what IBM said in the past is true (and if the efficiently designed GCN chipset is anything to go by).

A new thought...
Just thinking about this made me wonder...how is BC (processor wise), being handled? The GCN API is the foundation for developement for Revolution software developement afterall. The system, overall, is smaller than even the GCN, so would there be room for the GCN chipset or (as Sony has done) a "GCN on a chip" for BC? Or is the CPU & GPU so much like the GCN's "Gekko" & "Flipper" (right down to sharing the same kind of MoSys memory again) that they're simply the same only "beefed up"? Or maybe the "Hollywood" & "Broadway" processors in Revolution are powerful enough to emulate the GCN chipset in real time, elimnating the need for a "GCN on a chip"? Could this clue us in on how powerful Revolution might be? Maybe the "Broadway" is multiple "Gekko's" clocked higher, made smaller, made to run cooler, with more on chip memory and more tweaks to fine tune it's performance? And likewise, maybe the "Hollywood" is multiple "Flipper's" set-up in the same fasion. Maybe that could be why Nintendo has said that Revolution's power is 3x's that of GCN, 'cos Broadway is 3 optimized Gekko's & Hollywood is 3 optimized Flipper's??? This would be a cheap solution, allowing for more power and allowing for an easier "upscaling" in performance from a developing standpoint. The GCN was easy to develope for right...so why mess with what works? Why not just use the same technology, just up it's speed, efficiency and add more instructions?
 
UltraMagnanimous said:
Iwata already said the sytem will have DVI.
He didn't though. Quite a while back, he was quoted in an article, in response to someone asking if the Revolution would be a dual screen device like the DS, and he said it would connect to a "monitor". It has always been my opinon that this was misquoted as a literal computer moniter, when all he really meant was that it would connect to a display device like any other home console would. And regardless, read what OG said again. Not that there wouldnt be and DVI connection or anything, just that it would come out of a universal A/V out port and not directly from the back of the console.

after reading through many peoples speculations and ideas, and going from what Nintendo has said, I really think that, like with the DS, Nintendo is aiming to not only create a game playing revolution, but a game creation revolution. They have been saying some pretty bold things lately, but if they can back it all up with the end result, I will be more than impressed.
 
DrGAKMAN said:
Maybe that could be why Nintendo has said that Revolution's power is 3x's that of GCN, 'cos Broadway is 3 optimized Gekko's & Hollywood is 3 optimized Flipper's??? This would be a cheap solution, allowing for more power and allowing for an easier "upscaling" in performance from a developing standpoint. The GCN was easy to develope for right...so why mess with what works? Why not just use the same technology, just up it's speed, efficiency and add more instructions?
It would be nice if it worked that way, but unfortunately that's not the case. Multiple processors require access to the same resources, which gets tricky to coordinate. The Sega Saturn had this problem with it's two Hitachi chips.

The Rev is using a customer PowerPC processor, but it's not the same as Gekko. The same with the graphics - it's not the Flipper, it's a new chip entirely, but still based on ATI's tech. Backwards compatibility is going to be like running old pre-Win95 games on a Pentium 4 in WinXP; it will work, but it will most likely require some sort of wrapper (for GC stuff) or emulators (everything else).
 
Lhadatt said:
The Rev is using a customer PowerPC processor, but it's not the same as Gekko. The same with the graphics - it's not the Flipper, it's a new chip entirely, but still based on ATI's tech. Backwards compatibility is going to be like running old pre-Win95 games on a Pentium 4 in WinXP; it will work, but it will most likely require some sort of wrapper (for GC stuff) or emulators (everything else).

I know my Gekko/Flipper X 3 wasn't likely, but I was just wondering how BC would be handled and if that could give us some kind of hint as to what Nintendo is producing for Revolution's chipset. If it won't be "heightened Gekko/Flipper, then that either leaves "GCN on a chip" (which adds cost and takes up space in a space concious design) or new chips that are powerful enough to emulate & run GCN games. Any theories?
 
blackadde said:
Um, can someone please explain to me the difference between a regular heatpipe and this supposed amazing new cooling technology? Aside from the pump (wheras the heatpipe just runs on convection) I don't see it. This sort of technology has been used for a while with laptops and SFF PC's.

edit: re: cube mapping, for them to generate a scene using a depth map wouldn't the texture resolution have to be massive to avoid it looking like a blurry interpolated mess?

Reading the patent, I came across a portion mentioning the use of 1024x1024 textures. This also would require a large amount of memory and bandwidth.

This leads me to the question of what are the texture resolutions for XBox360 and PS3?

The technology is really more about the liquid metal coolant than anything else, but some of the pumps supposedly don't require any electrical power.
 
Slurpy said:
Here we go again..

I'm going to assume the absolute worst for Nintendo's next console, so tha dissapointment is not a possibility.
Such a strategy is not always successful. I remember wondering what Phantasy Star Online's story would be like, and remarking "Hey, at least it must be better than Phantasy Star 1!"
 
Square's probably gonna enjoy this cube mapping, they'll finally find something to do with their huge CG infrastructure next gen.
 
Pardon me, but I'm at a slight loss trying to conceptualise this 'cube-mapping'. It seems to me to be something like the pre-rendering of backgrounds seen in the playstation final fantasys and resident evil games or maybe what was tried on Donkey Kong Country. I know I'm way off, but thats how it sounds to me. Can anyone help me out here?
 
UltraMagnanimous said:
Pardon me, but I'm at a slight loss trying to conceptualise this 'cube-mapping'. It seems to me to be something like the pre-rendering of backgrounds seen in the playstation final fantasys and resident evil games or maybe what was tried on Donkey Kong Country. I know I'm way off, but thats how it sounds to me. Can anyone help me out here?


What my understanding is, is that the obviously drawback to having pre-rendered environments is the lack of interactivity with environment. What this feature does is remove the flaw of pre-rendered environments, making what was once non-interactive, interactive.

Imagine RE:Remake, with the same detail but the environment is interactive. Also you would have a 3D camera and none of what you experienced in the original(no unexpected shifting of the camera angle or position. So it would be a 3D environment.

The texture resolution would 1024x1024 so that textures wouldn't get blurry when you get close to them.
 
OG_Original Gamer said:
What my understanding is, is that the obviously drawback to having pre-rendered environments is the lack of interactivity with environment. What this feature does is remove the flaw of pre-rendered environments, making what was once non-interactive, interactive.

Imagine RE:Remake, with the same detail but the environment is interactive. Also you would have a 3D camera and none of what you experienced in the original(no unexpected shifting of the camera angle or position. So it would be a 3D environment.

The texture resolution would 1024x1024 so that textures wouldn't get blurry when you get close to them.

but this seems to be technology that anybody could use. Has nintendo patented this to the point where nobody else can take advantage of it, or have they patented a method that makes it easier to do or more practical? Is it something that would be built into the GPU or would the GPU be designed with this in mind... i'm just wondering why nobody takes advantage of it NOW. It sounds like it could be usefull even on current hardware.

edit// maybe it would be something built into the nintendo development tools?
 
OG_Original Gamer said:
Reading the patent, I came across a portion mentioning the use of 1024x1024 textures. This also would require a large amount of memory and bandwidth.

This leads me to the question of what are the texture resolutions for XBox360 and PS3?

The technology is really more about the liquid metal coolant than anything else, but some of the pumps supposedly don't require any electrical power.

I'm pretty sure you'd need larger textures than that to generate a believable environment. I *think* this sort of thing would only be used for skyboxes, where you don't need them to be very interactive and are seen from a more or less fixed point, maybe with a bit of parallax to give it some pop.

You might have something with using liquid metal, but as far as I know a phase change (liquid -> [CPU -> evaporation] -> [radiator -> condensation]) heatpipe setup would be more efficient. I haven't crunched any numbers so I could be wrong.

Pimpwerx said:
Heat exchangers are nothing new. And I still like a peltier unit better. But I digress. PEACE.

Peltier units are only useful when you need to pull the CPU down below room temp. The other side will radiate even more heat than the CPU itself, so you'd still need a big ass cooler to keep the whole setup from melting.
 
Top Bottom