While theoretically true and, given current technology perhaps practically true, technology changes. There are phone CPUs today that run at 1GHz but it was only a few months ago that the N900, at 600MHz was the top of the heap. I wouldn't be surprised to see a CES announcement of a 1.5GHz processor and I wouldn't be surprised to see 2.0GHz shipping within the next year.

With more processor power comes more capability to handle overhead. Few would argue that the Maemo should eliminate a graphical interface and run everything from a command line but it would be faster (and that's how I used to dial my modem). For that matter, Maemo, even if it only ran a command line, is, like Linux itself, an unnecessary overhead. If you want speed, program in the CPU's assembler language.

The Android VM, Python, Java, etc. do take resources but they give back portability and if you want to support multiple devices with multiple versions of an OS or even multiple OSs, it's worth the overhead. Nokia, with one OS on one device doesn't have to worry about this. . . . Yet.

Exactly. Nokia deals with all this themselves. Just like Apple. Why waste resources?

It doesn't matter if they came out with a 16-core 900ghz super-charged killerwhale processor... the fundamentals are still exactly the same.

You can run more, faster, with a native program than with a virtualized or "interpreted" one on the same hardware.

Android is entirely about portability since Google does not do hardware (until the Nexus One). Maemo, and iPhone are not. You are also correct about the GUI.. which is why I *always* have a terminal screen or 10 open on my desktop at home.. I do most work directly in the terminal.

A GUI however, has become an end-user requirement. As long as the software and OS work.. the end-user doesn't particularly care if it's interpreted-driven or native or what.

Here's an argument to your portability however - In Maemo, I *can* recompile software for ARM and run it if no Maemo solution exists. In Android, I do not have this option at all - the entire application would have to be rewritten in java to work with the VM. Thus, in this example, Maemo is more adaptable to a users needs.

I enjoy programming in perl and ruby, and will probably learn python once I get the N900, because it is simpler and faster for me to *code* the apps in those than in C/C++ and compiling it. But - the drawbacks must be acknowledged, and a decision made, whether that is the best solution for the software you are writing or not. Forcing one solution over the other I dislike.. which is why Maemo is the poster child for total awesomeness in all the world, it can do both .

Well have you seen same kind hype on N900 as iPhone? I think not. N900 is nerdy device and as a such there will be very little market share.

I'm not saying that we need 60 different in a year models from nokia like we used to. But it's really difficult to get good market share with only one device unless you have brainwashed cult behind you. Nokia is just too far away from that

You are correct to some extent. The way I see it is that because the N900 is built to a "nerdy" audience it also has chances of becoming successful since the same audience are the ones that develop applications and help with other issues, filling bugs, visiting this forum etc...

M6, then, should be something out of this world. Nokia promised that M6 will be the next step of its platform and tried to convince everyone that the N900 is a device for a specific audience.

Nokia, with M6, needs then to build a better Ovi store(much better) and package the device in a more attractive way... Since looks still sell a LOT.

At least this is what I think

Unlikely. The over-gigahertz domain is not as attractive as such clock speeds have severe consumption implications on current (and next year's) fabrication processes. Multi-core processors make a lot more sense, as they can provide a good balance between raw power and consumption. The Osprey and OMAP4 are good examples of exploring that domain.

i guess that depends on the math involved.

that is, will it be more power efficient to go at Ghz, but for a shorter time, or at Mhz, but longer time?

in the end, it ends up a question of watts/second, and how many seconds are needed to complete a task.

but i guess thats under a idealized, single task scenario. Ones multitasking and such gets involved, one also have to ponder how long it takes to get in and out of a given task, and if there ever will be a point where the system can sleep as all tasks are waiting for some kind of event.

i guess thats where a dsp and similar comes in, as its basically a kind of specialized multicore (hell, one can even claim that with a GPU). As such, it can perform a encoding or decoding task at much lower watt/second then the cpu.