I realize there’s lots of good reasons to go with ARM (and my favorite Linux device right now is ARM-ish (Nokia NIT, which is TI OMAP based)) ... but with the current market focus on the coming MID platform, I wonder if BugLabs has considered an Intel Atom based unit, at least as a prototype/proof-of-concept.
I can see lots of things to think about there… like, I’m sure the Atom will consume more power (yielding less battery life), at a given MHz rating, than an ARM … but I don’t know how much real estate the Atom based chipsets take up compared to the chipset the Bugbase uses …. if the Atom is smaller, maybe it would make for a smaller main board, and thus make room for more battery.
The win would be: more standard linux software to leverage, as well as being able to just use Ubuntu Mobile and/or Moblin for the OS, instead of having support a specific version of Linux.
I’m also unsure about the cost. The Atom is supposed to be a lot cheaper than Intel’s other offerings, but I have no idea how it compares to the chips that are in the Bugbase…
I’d just be interested in seeing an objective comparison (as opposed to a religious one) between the two approaches. I’m not suggesting that Atom will automatically be better. But I doubt it’s a foregone conclusion that it wont be.
Have you seen this article?
It is a pretty fair comparison. My 2 cent summary, Atom will be more powerful than ARM, ARM will be cheaper and consume less power. I see Atom more along the lines of a embedded PowerPC competitor than anything else. Right now I think you generally go up the cost/power/heat/speed curve from PIC or AVR, ARM, PowerPC, and eventually it will be the same but PowerPC or Atom. It sounds interesting but I think this quote kind of is the whole thing in a nutshell:
They're barely shipping now, the BUG has been in the works for quite a while, down the road it could be a factor but today it basically isn't an option.
(not speaking for BUG Labs)
It is not true that you will be able to leverage more open-source software with the Atom. At the moment everything open source and mobile is ARM, from Debian on ARM, to OpenMoko, Android, Opie, and Maemo.
There aren't many GUI apps from the x86 world you want on a screen this small, and most of the tools from the GNU/Linux world can just be re-compiled for ARM. Ubuntu recompiles most of the Debian packages it uses in Ubuntu and that is all x86 hardware, it is not unusual or difficult to re-compile all the stuff we need specific for our architecture.
So the real question is where can you get mobile apps from that you might like to use? Opie, Maemo, Android, GTK+ Mobile, OpenMoko are all for ARM and screens around the BUG's size. Right now Maemo and Android are much larger projects than Ubuntu Mobile, and consider that Maemo is in the wild, the BUG is almost there, and Android is coming soon, I wonder if Ubuntu Mobile is going to be late to this party? They say they are aiming bigger than a phone but smaller than a regular PC. That almost sounds like a distro for a Asus EEEPC or tablet, not really a mobile device, which makes me wonder if the apps developed for it will be all that relevant for the BUG. I think with the double LCD the BUG will be inline with Android specs, while Ubuntu Mobile says they are aiming larger than that.
So maybe when it starts getting a bit more main stream it would be worth considering, but at the moment it sounds too unproven and too new to be an option for a product shipping this quarter. I think it will be a considerable amount of time before Intel pushes deep into the market, keep in mind ARM shipped something like 2 BILLION chips last year. I'm glad Intel reached out to Canonical and if you read between the lines is basically paying them to make Ubuntu Mobile for their platform, because I am a Ubuntu user I would like Canonical to stay in business, but it will take a while for Ubuntu Mobile to catch up when there is such incredible momentum already.
Like I said, I wasn't looking for a product to ship now, or even in the near-ish announce-able future. I'm just wondering if they've started considering it, or gotten it to a prototype stage (I definitely wouldn't expect anything more than a one-off prototype at this stage).
(I'm familiar with all of those except Opie...)
I'm a Maemo user, and I am assuming that the other ones listed have similar issues to what I'm about to describe: You can't just use random open source, much less random linux, software on Maemo. _SOME_ random Debian packages can be used, but mostly that's only command-line and daemon type packages. If it directly uses the GUI, then you need to do hildon based customizations, and possibly more. So first you whittle it down from "random" to "Debian", from "Debian" to "Debian ARM", from "Debian ARM" to "Hildon-ized", from "Hildon-ized" to "ported to work with the exact current version of Maemo" (the last iteration required a further set of customizations). So, yes, I stand by my assertion that you might get a better application ecosystem if you're using a more generic Linux platform.
My expectation is that Moblin and Ubuntu-Mobile will be better about this, exactly because you're starting with an x86 based CPU.
For most cases, that's probably true. But there may be cases where the need for an app that hasn't been fully ported will outweigh how difficult it is to use on that screen. And if there's _some_ way to display the screen externally, then that affects things too.. sure, maybe you wont use it on the tiny screen, but maybe you'll use it on an external display when you're not mobile. Or display it remotely (VNC or xhost) to your desktop (again, when you're not mobile).
And what about the users who aren't into recompiling and porting? They're going have an affect on the perception of how easy/enjoyable/useful/usable your platform is... which will probably also affect how successful your platform is.
I disagree. The real question is: how can you be as easily compatible with as many applications as possible, so that you can have the best application ecosystem available making it easiest for users to run whatever they want to (whether they want to be limited to a "mobile device" or whether they're also looking at it as a "light weight thin client laptop replacement" device, for which Maemo is _almost_ there ... one of the limitations being: inability to use random linux software ... but I already no longer feel the need to have/own a laptop). I think the "must be mobile and comparable screen size" is a high priority preference, but only a preference and not a requirement.