This.  It fits OnlIve's microconsole far more so than it does the idea of powering Nintendo's next home console.

I also believe they'll continue to use PPC architecture.  IBM could do the same thing for Nintendo's next home console that they did for Wii by taking a current IBM product and re-engineering it a little to better fit Nintendo's request for efficiency.   A reduced clock Power6 or even Power7 makes far more sense to me than this chip does.

Something else to consider. The PS3, Wii and X360 are all based on PowerPC architecture.  That means devs have had a lot of years working with them.  Sticking with a similar architecture for Nintendo's next home console would enable easy B/C and ensure developers can get to work quickly on quality code. out of familiarity.

 I don't know what Nintendo should or is doing, I'm just playing devil's advocate ...

There are always going to be advantages and disadvantages of taking different approaches with hardware designs; and depending on what other design decisions you make, you could design a system using (just about) any CPU and have it be reasonable.

With a company like Nintendo (or any console manufacturer) you're interested in more than just the raw performance; or how that raw performance relates to energy consumption and cost. They have to determine an approach that will lead to the greatest overall performance in game, with the lowest development effort, at the lowest manufacturing cost, and under a set energy consumption limit. While I can't know (at this point in time) whether Nintendo decided that a lower performance CPU paired with a PPU would achieve these gains better than going with a higher performance CPU, I can always argue that that is a decision that Nintendo could have made.

There aren't many chips out there which could compete with a 2nd generation Bobcat/Fusion for performance/watt and performance/mm^2. There aren't even that many which can compete with the first generation bobcat/fusion.

But...one could easily argue that timing certainly suggest OnLive more so than Nintendo.  All those other factors aside, a “major customer preparing to launch a new gaming platform” alludes to the upcoming launch of the OnLive microconsole more than Nintendo given the unliklyhood that Nintendo will launch a new home console anytime soon (at least a year a way if not more) and the multimedia functionality of the Marvell chip mirror the reported functionalities of OnLive's microconsole.

Yes, I hope for something like AMD Fusion too, it would be interesting.

The original GB and SNES were only 14 months apart in Japan fwiw.  Still, I think the idea that Nintendo can't release a handheld in the same year as a console gets a overly negative light from GC's relative failure... the DS didn't really take off until 2006, until the DS Lite had launched, that that didn't brunt Wii one iota sales wise...

Let's hope it''s true, it's about time for a new generation cause the current hardware is starting to show it's age. If Nintendo launches a new system Microsoft and Sony  will probably forced to so in te near future aswell. Although it think Nintendo will stick with their strategy and create a less powerfull and afordable system, a quad-core cpu(although depending on it's speed) would suggest that it's not going to be much powerfull than the current 360 and PS3, just like Wii isn't much more powerfull than PS2 and Xbox.

The number of cores isn't a valid metric determining comparable system capabilities.

This 4 core chip from Marvell would get ripped to shreds by a high end 4 core chip from Intel.  Another example are that both the Wii and PS3 are technically single core processors.  The PS3's 7 SPE's are not "cores" as they are not independant.   

I even doubt the next consoles from Sony and MS use more than 4 cores.  For closed environment, specialist hardware, rampinging up the cores has a drastic rate of diminishing returns.

I just wanted to add that real world performance is determined more by how well suited a processor is to a problem than by theoritical processing power. To explain what I mean, consider HD video play-back ... Many mobile processors today are far better at playing back HD video than desktop processors that are a couple of years old (and an order of magnitude more powerful) because they have instruction sets that make them "better suited" to decoding HD video.

While videogames are more complicated than video playback they are a class of applications where a very small portion of the code base takes up the vast majority of the run-time; and (regardless of the game) this small portion of code is very similar for all games. What this means is that games respond very well to increasing the instruction set on a processor to get increased real world performance.

Now, I'm not anticipating anyone doing this but it is theoritically possible for a console manufacturer to licence several major game engines, port them to their (potential) hardware, profile their performance, and to add instructions to their CPU/GPU until they got the performance they desired.

Yes, GBA and Gamecube were released in 2001

Also, they launched both DS Lite and Wii in 2006.