Via introduces APC Paper: A $99 Android PC
apc.io
apc.io
Ubuntu especially are now focusing on blingy desktops that somehow require really new cards and hardware only, and yet are also trying to throw away the traditional desktop to cram themselves onto a phone. And yet with high end requirements they are now not the desktop OS of the light weight / poor hardware people, they are for the high end only paradoxically.
Android is coming the other way: a good core set of features now slowly building out into a potential desktop replacement. Because the desktop is still important, especially for anyone looking to do any productivity work.
And after watching a fast demo on the site, I see another thing Android is doing right. Somehow even though it's incredibly low resource it has animated desktops. I was first exposed to the idea in the mid 2000s with a video demoing Enlightenment 17 showing animated desktops. I even made a bug with Gnome asking for the feature. No dice. Even with all their modern "bling" and high and hardware requirements, they can't do live wallpapers/desktops. So with a lame static desktop, try running them on this hardware? EXPLDOE. Won't work.
Meanwhile light little android also has this cute feature.
And this really just underscores to me that in all ways Ubuntu/Unity/Gnome shell are just going in all the wrong directions while Android seems to be increasingly going in many of the right ones.
Bling is so much less work than games that even a low-end GPU can drive Unity's bling. The only problem is when you have no working 3D driver at all.
Phones actually have better 3D drivers than desktop PCs, because Linux is the dominant kernel for phones. Every phone SoC has working GLES 2.0 Linux drivers. (The drivers are proprietary but that doesn't bother Ubuntu.)
If you can't find support for animated desktops in Unity, then it's probably because none of the people who actually write the code thought it was a valuable addition.
Retro-fitting Android onto desktop devices isn't the solution, there are many lightweight linux distributions if you want an actual desktop OS.
Animated desktops atleast to me are a gimmick, and the reason other linux distros haven't been able to optimize their graphics is because of the desktop parts and unreliable drivers (Nvidia & AMD), now most of these micro-PCs use mobile parts (ARM Cortex processors, PowerVR graphics cores.) which have readily available open-source drivers or OS-optimized proprietary blobs.
Obviously a desktop linux ("lightweight" or otherwise) makes more sense on a desktop than a retrofitted android, since android will always be, first and foremost, a mobile OS.
That said, android's app stores have produced, in a fairly short timespan, an entire ecosystem of polished, useful apps that users love. Without android, many of these applications wouldn't exist at all.
As a developer, a regular old linux distribution on my desktop makes more sense[1]. As a normal everyday user who just wants the additional amenities offered by a big monitor and a real keyboard, "android on the desktop" makes more sense.
[1] - I'm actually partial to FreeBSD.
Also, while ChromeOS may make sense for schools and businesses that have adopted the cloud fully, I'm not sure ChromeOS is that good of a option for regular consumers, other than having a more optimized interface for the desktop (a lot more than even Windows 8).
But for a consumer, does ChromeOS offer that much more than Chrome for Android?
Though to some extent possibly putting a debian chroot onto an SD card might almost be enough. There are a few different attempts to start putting dev environments onto android (and then there's also the ipad as a devtool "buy a linode and ssh to it" method...)
but yeah, sadly even on some devices that means cracking/rooting them just to be able to properly chroot, but with that you'd at least have all the compilers you want and vim and/or emacs :P
Then we could start looking at bootstrapping more tools I suppose
Use ConnectBot (ssh) and hacker's keyboard, and you're all set.
I mean even Microsoft is pushing for full-blown touch interface on the desktop. Android right now is certainly not worse from an interface point of view. If anything it might be more familiar to Windows users because of its more desktop-like UI and icons, than Windows 8's Metro interface is.
For many billions of people Android is going to be their first OS, an OS that is open source, and comes in very affordable devices. Give it 3-5 years and I think it's going to become a real competitor not just in the "mobile OS" space, but in the whole OS space.
I think it's a bigger jump than just a little optimization. Consider:
* no proper windowing
* no OS support for copy and paste of anything except plain text - you can't
even copy some bold text from one part of an email to another
part of the same email!
* the mouse support is really mouse-simulating-a-finger, with all its clumsiness
(for example, you have to "long click" to get functions you'd normally "right
click" for on a desktop), there's no "mouseover"
* apps are killed randomly in the background and often don't restore their state
upon returning eg: switch from your word processor to your browser and back
again and there's a *chance* your word processor got killed. Depending on the
quality of the code, it'll start up blank, perhaps with the same document, but
almost never in exactly the state you left it.
[1]: https://plus.google.com/100275307499530023476/posts/ViCME1bb...
Hackborn starts with a stick ("I suspect we are going to need to start doing things to prevent you from impacting our app ecosystem") and then shows just a tiny glimmer of carrot ("Doing this correctly, without impacting our app developers... I also think it is something that needs to be done at the mainline platform level...")
So who knows what to make of that. My main reason for hope is that it's just so unusual for Google to speak out that way. All kinds of bizarre things get added to ROMs and they just ignore it. Then this comes and the Google engineers go bananas. It has to mean (or at least, I think it has to mean) that they have some plans around it, and they really don't want something incompatible coming out and becoming a defacto API ahead of that.
One PCs you have right click and double click, which are both replaced with long click. Using PC design interfaces on mobile will usually work (2 becomes 1). But the reverse fails because 1 cannot become 2.
For $50-$60 you can already buy an Android mini-PC with double capacity in each of these specs.
For instance:
http://www.aliexpress.com/item/MK809-II-Android-4-1-Mini-PC-...
But seeing that the Paper has an accessible JTAG port, they could really sell a JTAG adapter under the "Scissors" name.
Samsung has their Armdale board at $250:
http://www.arndaleboard.org/wiki/index.php/Main_Page
If you're going to drop ~$250 it might make more sense to just buy the Samsung ARM Chromebook or the Nexus 10 and slap your own Linux or Android build on it, unless you really need the easy JTAG connection or something.
There's supposed to be a cheaper ($135) option in the form of the 'ARMBrix Zero', but it is currently only up for preorder:
http://www.cnx-software.com/2012/12/26/135-armbrix-zero-exyn...
There's also the novena laptop bunnie and xobs are building (still in progress), if you want something very DIY (and presumably lots more expensive):
Looking at the Paper's new faster performance, the addition of GPIOs, and Android 4.0, and an optional case, it is clear that VIA has been listening to user feedback. They certainly delivered on my wishlist. In the end the 8750 was a $65 lesson on how far Android has come in such a short period of time.
Android applications are excellent, so while I see plenty of complaints about what it doesn't have, it is certainly getting the developer attention necessary for applications that can / will replace productivity applications on Windows machines. For the 80%, the desktop is nearly dead.
Further, Android is a great platform for embedded applications, and it looks like APC has delivered an inexpensive platform for the maker / hacker community.
Though a direct comparison to RPi is unfair, an APC comes with a power supply and 4GB of flash, so I consider a Raspberry Pi to actually cost $75: $45 (Actual street price of a Pi) $15 Memory Card $15 for a decent USB power supply (a low quality PS causes many of the RPi stability and Ext4 corruption issues I have seen on forums). (Also, Android is preloaded, so it is a much lower effort to get running.)
And the comment about fire made me laugh. The CPU of the 8750 might have reached 40 degrees under very heavy load and the power supply is external, so cardboard certainly won't be an issue. I thought it was funny because I actually use the cardboard box the 8750 comes in because it is virtually impossible to find a Neo-ITX case for the board.
Hope that provides some insight. I am glad to see VIA remains committed to this project.
That may be a reasonable thing to ponder until you try to run Ubuntu on a device which formerly ran Android. I thought the same as you and installed Ubuntu on my Asus Transformer TF101.
While the original Transformer sports reasonably dated specifications, it's provides a very neat, fast and responsive Android-experience.
I expected Ubuntu to offer the same, but with a wider-selection of (non touch-optimized) software. I was thoroughly disappointed. Just opening the Terminal Emulator took 5-10 seconds. Trying to open Firefox or any web-browser was a joke.
What most people fails to realize (I guess because Android has adopted the Java-is-slow-meme, either it wants to or not) is that Android is heavily optimized because it has to be responsive on ultra-light hardware. Given its feature-set, it's a very fast OS compared to most other out there.
Ubuntu on the other hand, is heavy enough to sometimes make my Core i7 wish it had more CPU-cycles. Putting Ubuntu on underpowered machines is something I wouldn't recommend to anyone.
Android obviously handles low memory situations better. Ubuntu can't match its ability to free up the ram used by inactive apps.
I'm not disagreeing with your general point, but I wonder if a lighter DE would make a big difference. (Unless you were already using Xfce, and got the results you describe... that would be bad news.)
Don't forget who this is aimed for. Alas, Ubuntu is still mostly for folks above some level of computer proficiency. Android, on the other hand, is probably one of the most "familiar" experiences for people using smartphones, and the (huge) amount of applications is easily accessible through the Play store with lots of reviews and hence high quality. You just don't get that for Ubuntu.
There are lighter Window Managers, but most of these don't give quite such a modern , easy and complete user experience.
Even with that though, the kernel not getting updates and continued patches makes it harder to keep things built for an end user, but I'm not sure if that is so much of an issue past the Linux 2.x to 3.0 jump with Android devices.
Updates for anything Android outside of Nexus devices that Google maintains are generally far from automatic or dependable.
I've actually been mulling over starting a project like this for machines like the RPi and APC -- an OS where you basically boot into something like a REPL, and start coding away.
I'm nostalgic for the old 8-bit BASIC too, but primarily because you could switch the thing on and the interpreter was in ROM and ready to go a few seconds after powering on.
(http://geoffg.net/maximite.html)
> The Colour Maximite is a small and versatile single chip computer running a full featured BASIC interpreter with 128K of working memory and eight colours on a VGA monitor.
> It will work with a standard PC keyboard and because the Maximite has its own built in SD memory card and BASIC language you need nothing more to start writing and running BASIC programs.
> The Colour Maximite also has 40 input/output lines including an Arduino compatible connector. These I/Os can be independently configured as analog inputs, digital inputs or digital outputs. You can measure voltage, frequencies, detect switch closure, etc and respond by turning on lights, closing relays, etc - all under control of your BASIC program.
The immediacy is something lacking in current IDEs and ease of start. Turn on, type, and something happens.
I'm still hopeful I'll see an under $200 programmable computing device sold somewhere. Perhaps if OLPC is actually serious about selling at Walmart[1], it will happen.
1) http://www.engadget.com/2013/01/08/olpc-xo-tablet-walmart/
But there are a lot of good tools without creating anything new. IMO, the real places for improvement are learning methods.
My own idea has been to get my boy to start learning javascript via codecademy. We're seeing how that goes.
He's 11 and more interested in minecraft (which, IMO, has a similar set of activities such as wikis and researching via message boards as my own day-to-day why-doesn't-this-work programming issues), but we'll see how this goes.
The other thing that I did was to put together a desktop from parts with him over the holidays, which was a bit more fun than our usual yearly giant lego build, and I hope will give him an idea that these machines aren't magic-- they are put together, even if I don't have the skills to convey that at the circuit level.
How much did it cost to build? I would wager it was more than $200. To put together a PC is beyond a lot of parents skill.
It worth remarking on specifically because it is not where Android was designed to go and it fits rather awkwardly.
Imagine you sell taco kits. Then you discover some people are buying your taco kits, throwing away the shells, downloading instructions from the internet and using the other ingredients in different proportions to make bean soup. You also sell canned soup and it also sells decently. You might decide to bring out a bean soup range based on the taco kit. Maybe they think it tastes better then your intentional bean soup. Maybe they like making it themselves. Either way, it's worth noticing.
http://www.amazon.com/Best-Sellers-Computers-Accessories-Lap...
There's still a huge business market for browser apps such as Docs and it would take a while before there are equivalent Android apps for desktop form factor.
Interesting. With or without blobs?
For hardware acceleration only GK802 has it currently. You can purchase it from this thread - http://forum.xda-developers.com/showthread.php?t=2044045
If you don't need hardware acceleration there is an Ubuntu version for UG802 and MK808 - code.google.com/p/rk3066-linux/wiki/PicUntu.
EDIT: Would that ever get a UL seal? I wonder!
If any excess heat is generated by this thing, that aluminum backplate can probably sink it.
As for UL certification, this thing is low-voltage DC only. The power brick should be certified but that's probably all that's needed. And FCC class-B like usual.
Freescale got the hint and put a lot of the voltage regulation on-board so there's no need for the external PMIC anymore. That was always the dirty little secret of these SoC chips, you needed the companion chip to make it complete...
This could be a potential problem for Microsoft.
$20 cheaper, and a VGA port :D
I know if you tried to run Android on Netbooks using the 2.x codebase (which was possible) you'd have a horrible user-experience due to the platform's limited support for non-touch input-devices.
With Android 4.x the support for these has been improved to the point where you can call the thing a "PC" and almost use it as such.
They may just be referring to this. I don't know.
Would be neat for a Raspberry Pi and ironic for a DVR device.
I guess overheating is not a problem? I don't see any vents.
Also wondering if we connect a screen, a USB keyboard and power supply, will it still "blend into the office environment"?
There's much more than just a CPU inside the case.
if you don't like calling it a PC, just call it a computer.