Interesting thanks. How much more powerful would that set up be than 3DS and Vita ?
If the Vita's GPU is close to iPad 3's clocks, then the Mali T760MP2 GPU in the MT6732 would be almost twice as fast, as seen in both GFXBench T-Rex and 3dmark's Ice Storm Unlimited:
http://www.futuremark.com/hardware/mobile/Elephone+P6000/review
http://www.futuremark.com/hardware/mobile/Apple+iPad+3/review
There's no way to compare the 3DS' GPU because that PICA 200 doesn't exist in an Android device. Plus, it doesn't seem like the GPU is OpenGL ES 2.0 compliant (no pixel shaders AFAIR).
Though from looking at the games, it's safe to assume that it's a lot slower than the Vita's SGX543MP4 (>5x slower, at least).
One of the major reasons I think they might fall short of Vita's specs is because Nintendo's franchises don't tend to need powerful hardware because of their artstyle and the fact that third parties won't put PS360 gen like budgets into pushing a handhelds visuals, so would there be much point in creating a handheld much more powerful than Vita ?.
Maybe we should be asking the question the other way around.
What would Nintendo gain in developing a SoC that in 2016 manages to perform worse than a Vita? As I mentioned above, there are already very cheap off-the-shelf chips that would run faster and probably consume less than a Vita already.
So what would be the point of spending R&D resources to develop something with the sole purpose of performing like crap?
If Nintendo could save $50 per console by matching or falling a little short of Vita's specs then I think it would be a no brainer for them after how Vita has done in terms of the mass market.
They wouldn't save $50. They probably wouldn't even save a single dollar. The Vita's design is now 5 years old. It's not powerful by any modern standards anymore. Why the determination of making something slower?
Do you know how powerful they could make a console based on a $299 price point and ARM based architecture ? Would PS4 levels of performance be possible for instance ?
At this point, it depends on how much profit Nintendo would be willing to make on the console's release, and the date of release.
For a late 2016/ early 2017 release, I believe they could order AMD to build a ~4B transistor SoC with 8GB HBM2 using 14/16nm and put it in a tiny $300 console, if they were willing to sell the consoles at little to no profit. That would definitely get them leveled with - or even surpassing - the PS4, no matter which architecture they chose. They would, however, need to step up on the power budget compared to their last 3 console generations, which is another thing where Nintendo is crazily conservative.