I've long thought this, but there may be a problem...
ARM processors probably can't get much faster without an unacceptable loss in battery life - for e.g., the iPhone/iPad are underclocked, and the Tegra 3 quad-core in the new Transformer Prime has three modes, each with lower clocking. Batteries might address this, but their rate of improvement is slow (no Moore's law there). The apparent solution is "multi-core", but we aren't able to parallelize in general, after decades of research. Note that desktops are stuck at quad-core - there's more cores on server chips, but they are usually doing many tasks already (e.g. webserving). The dramatic multi-core progress is in GPUs, which again is a naturally parallel domain.
You could switch to power-hungry performance when docked, like the Transformer Prime... but it seems a bit of an ugly chimera in general.
If performance is needed for development, then x86 desktops will win.
Eventually, ARM could catch up to what is needed (even if x86 keeps ahead) - but if this is to simultaneously satisfy mobile needs, it will be at a much slower pace than Moore's Law.
An ARM purely for desktop could catch up easily; but what benefit does it offer over x86, if its power-consumption advantage is irrelevant? And it lacks the massive ecosystem of x86 software. But that's not the idea discussed here, of a dockable mobile device.
Other factors: (1) ARM SoC is cheaper and smaller than x86; but Intel has made a SoC x86. (2) people like the convenience of the same data, apps and UI for mobile and desktop; but cloud provides the same data; webapps address the same apps and UI.
A prediction: the iPad 3 won't be quad-core. It's too hard to convert multi-core into user performance, so it's an efficient use of power and silicon. Instead it will have masses of GPU silicon (at least x4 the iPad 2, for the x4 pixels in the retina display - possibly x8, to give an sense of improvement in addition).
tl;dr battery life and multi-core prevent an ARM-based dockable desktop.
I never said ARM. ARM != mobile, a point you make yourself but apparently didn't follow through to its logical conclusion. I also did say the future. Only prototypes exist now (though they do exist), but I don't expect widespread adoption for at least another three years. In three years, a cell phone with at least as much power as a netbook of today will be perfectly feasible.
I don't much care what's in them. I'm much more interested in whether they are open computers with a cell phone attached, or closed cell phones with vaguely computery locked down capabilities.
These are just my thoughts, I wasn't attacking you. I was hoping for a refutation of my argument, if you can spare the time.
BTW: arguably, a cell phones are already comparable in performance to netbooks: my old eee PC has a 900MHz celeron (though less powerful than today's netbooks); the samsung galaxy 2 has a dual-core 1.2 GHz Cortex-A9, and the Transformer Prime has a quad-core 1.3 GHz Cortex-A9 (though that's a tablet, not a cell phone)
ARM processors probably can't get much faster without an unacceptable loss in battery life - for e.g., the iPhone/iPad are underclocked, and the Tegra 3 quad-core in the new Transformer Prime has three modes, each with lower clocking. Batteries might address this, but their rate of improvement is slow (no Moore's law there). The apparent solution is "multi-core", but we aren't able to parallelize in general, after decades of research. Note that desktops are stuck at quad-core - there's more cores on server chips, but they are usually doing many tasks already (e.g. webserving). The dramatic multi-core progress is in GPUs, which again is a naturally parallel domain.
You could switch to power-hungry performance when docked, like the Transformer Prime... but it seems a bit of an ugly chimera in general. If performance is needed for development, then x86 desktops will win. Eventually, ARM could catch up to what is needed (even if x86 keeps ahead) - but if this is to simultaneously satisfy mobile needs, it will be at a much slower pace than Moore's Law.
An ARM purely for desktop could catch up easily; but what benefit does it offer over x86, if its power-consumption advantage is irrelevant? And it lacks the massive ecosystem of x86 software. But that's not the idea discussed here, of a dockable mobile device.
Other factors: (1) ARM SoC is cheaper and smaller than x86; but Intel has made a SoC x86. (2) people like the convenience of the same data, apps and UI for mobile and desktop; but cloud provides the same data; webapps address the same apps and UI. A prediction: the iPad 3 won't be quad-core. It's too hard to convert multi-core into user performance, so it's an efficient use of power and silicon. Instead it will have masses of GPU silicon (at least x4 the iPad 2, for the x4 pixels in the retina display - possibly x8, to give an sense of improvement in addition).
tl;dr battery life and multi-core prevent an ARM-based dockable desktop.