I was recently thinking about Intels potential role in future consoles and based on some interesting new information I felt that Intel was presenting one of the strongest cases for their participation in the next generation and that Nintendo is the leading candidate for using their technology in their console.

What Intel brings to the table:

Intel produces the best mobile CPUs out in the market today. Their performance per watt and performance within the laptop like power envelope expected for a next generation Nintendo console is second to none. They are the best potential supplier of a chip Nintendo could hope for as they are powerful, they sport excellent single thread performance and it is incredibly easy to extract that performance and the variability per chip produced is extremely low so almost every CPU in a wafer is a console candidate chip so their overall cost of production is very low. 

Intel also working on a next generation graphics chip based upon their earlier work Larrabee which they shelved the first implementation due to the difficulties of the software implementation. This chip would be useful to Nintendo because it implements Ray Tracing relatively efficiently.

http://www.drdobbs.com/high-performance-computing/218500694;jsessionid=0ICBW21HRB035QE1GHOSKHWATMY32JVN

Why would Nintendo want this? For one it visually differentiates their console from the other two if they still use traditional raster methods of graphical rendering. Secondly it lowers production cost and simplifies the production of their games whilst producing higher quality images and thirdly Nintendo has always been more concerned about producing visually pleasing art work but they have never cared for reproducing realistic images in their games so they need not be concerned about whether their games stand up to a side by side technical teardown. Ray tracing may not give them the same visual complexity but it will make their games more visually pleasing which suits Nintendo of all companies pretty well.

Why would Intel want to work with Nintendo of all companies?

Nintendo is the largest console manufacturer in the world. The mere mention of the idea that Nintendo would be using the ray-tracing architecture from Intel would give it instant developer attention and credibility. Because of who Nintendo are, hundreds of development houses around the world will be forced or given a great incentive to consider Intels new graphics architecture. If they have to throw in cheap access to their new CPU architectures to do so its really quite a comparitively small price to pay.

Nintendo is also the only console manufacturer who would be happy to stay on Intel's old process nodes. Microsoft or Sony would be bugging Intel every other year to gain access to the newest process node to make the chips smaller. On the other hand Nintendos chips would already be as small as Nintendo would want to make them as they are very conservative in terms of their hardware implementation. They would be happy to start on 22nm and finish on 22nm 6 years down the track. They wouldn't be competing with Intels high margin cutting edge CPUs for production at their cutting edge fabrication plants. This means the cost per chip for Intel would be very marginal and they can make profitable use out of their old fabs to produce chips for Nintendo.

 

Wouldn't that make it harder to develop for the N6 than the competition since it doesn't use that 'rasterization' devs are familiar with?

"relatively efficiently" is a telling statement.

Limiting the ray tracing implementation to interactive levels will make the graphics look worse than using rasterization techniques. I can't see this being viable untile the raytracing hardware is on the same processing time/image quality level as rasterization (which is insanely fast and even has programable hardware available)

Seems a bit far fetched.

SaviorX said: Wouldn't that make it harder to develop for the N6 than the competition since it doesn't use that 'rasterization' devs are familiar with?

Its a balancing thing. The X86 CPU is an order of magnitude simpler to develop for which balances the increases complexity of ray-tracing and it ought to take less time to produce the visual look of the game so overall production costs will be lower.

They don't need to make the game look realistic nor does it have to be as fast as rasterization so long as it meets the criteria of lowering overall development costs and differentiating the visual presentation of their console. It just has to have enough comparitive advantages and Intel can quite easily sway that balance with their CPU technology.

Honestly, I think Nintendo is more likely to use the AMD Fusion. Not just because their prior relationship with ATI, the Fusion seems to be more of reality than the Larrabee.

My main concern is what switching architectures might mean for backwards compatibility. I'm pretty sure that Nintendo will want their next console to run Wii software without any problems.

There are several different kinds of solution to that problem. The question is: Does the cost of solving it for an X86-based Wii 2 outweigh the benefits of switching to X86?

famousringo said: My main concern is what switching architectures might mean for backwards compatibility. I'm pretty sure that Nintendo will want their next console to run Wii software without any problems. There are several different kinds of solution to that problem. The question is: Does the cost of solving it for an X86-based Wii 2 outweigh the benefits of switching to X86?

They already have X86 emulators. They distributed them to developers before their official development kits arrived. This is in effect an already solved problem. In addition to this, the hardware behind the Wii will already be close to 12 years old by the time a Wii 2 arrives. The performance required for emulation would not be a problem.

I don't really see Nintendo/Intel making that much sense since Nintendo doesn't tend to use cutting edge hardware so they can keep costs low.

AMD is a much better match for Nintendo in a lot of ways. They produce reasonable performance on the cheap. Nintendo doesn't want a cutting edge chip to beat the competition on graphics, that proved to be a losing battle on the Gamecube (keeping up with the joneses graphics wise). Instead look for a cheap processor and graphics card that puts the Wii at PS3ish graphics by 2012 for under 300 dollars on release yet again.

Also the Wii isn't helped by X86 much since Nintendo like Sony, Apple and everyone who isn't Microsoft doesn't use direct X. Direct X is why X86 is different then other kinds of programing for the Xbox and the PC since direct X has a lot of X86 contingencies built in. If you use Open GL it matters far less what processor you use and so companies are free to go for the most cost effective option.

Open GL programming on a cheap AMD integrated platform for low total cost.

No, they tend to use hardware which suits their needs. I doubt you would call the 3DS obsolete before it released for instance.



Nintendo wants an effective chip which meets their needs which has a good cost/benefit ratio. Whether this is a new chip or an older design it doesn't matter. I did not mention beating the competition in graphics, however a unique visual presentation is desirable. Intel has every incentive to provide Nintendo with cutting edge technology for cheap, so the cost of said technology is moot.



How about excellent out of order execution, low latency, strong single threaded performance, easy programming environment and the best compilers in the industry. All this without even mentioning Direct X.

X86 is the most cost effective processor on the market today, except that Intel usually doesn't care to sell them on a cost+ basis. However if they had the right incentive they would.

Intel hype != reality.
As long as the deal is appealing, IBM can still offer the best price-performance compromise Nintendo is looking for with a PowerPC cpu, that BTW would make back compatibilitywith Wii and, if desired, GC, a lot easier than switching to x86.
Not to mention that Intel always sucked at GPUs.