I like where we're headed with these sorts of devices coming out. There appears to be some serious demand for powerful and tiny computing devices that are cheap enough to almost be considered disposable. This makes them perfect for weekend projects that require a little more umph than a typical Arduino can handle (and in some cases, like the Raspberry Pi's, they can even be cheaper).
I'm excited to see what the future brings in this nook of the technology/hobbyist industry.
I think a way to bridge Arduino shields up to a normal general purpose computing environment at a reasonable cost (e.g. http://omer.me/2012/05/introducing-ponte/) will be the true killer app for hardware hacking. Being able to write lua/python/ruby/etc to talk directly to hardware & sensors will be incredibly powerful
There's no way long-term Arduino is going to stay at it's current specs & environment, any more than TI-89 calculators could
Don't underestimate the usefulness of hard real time io that's easy to implement. It's nice being able to be able to guarantee that some function will run exactly 0.2 microseconds after a voltage hits a pin, for example. Simple micro controllers where you program the bare metal will always have a place for this kind of thing. Doing that kind of thing in python on Linux is a can of worms. But, abstracting away comms between a python script on a pc to an arduino talking to sensors is definitely someone worth perusing, though one is basically there with pyserial and some glue code.
For embedded computing, the thing that makes a board nice to work with is good documentation (board schematics and full datasheets/reference manuals), access to the hardware, and consistent behavior. In the $20-$200 range, there's lots of choices with more power than an Arduino. I understand there's excitement about Raspberry Pi because it's particular performance is closer to what ~$200 range boards were delivering for Raspberry's @25/$35. That's really a very nice achievement, but I'm cautious to see what degree of access to hardware will be possible over time. It might be well worth an extra cost, or giving up some performance if I'm getting additional time to do cool stuff vs. spending time reverse engineering capability that could have been accessed by reading datasheets.
It is nice that there are more development boards cropping up. But in this case, it is not such as big deal as the Raspberry is.
What is a big deal is the price of their LCD modules. They are very cheap compared to what most development kits offer. It is strange that in this 'Retina' world, there are still crappy LCDs being sold by more than $200 (assembled monitors, no matter how small, costing many times that amount).
Granted, these lCDs only seem to work with ODROID-X. Anyone knows of a decent LCD module that will work with a Raspberry Pi and doesn't require sacrificing your firstborn? That's for mobile applications (like a carputer), otherwise I'd just hookup an old monitor.
LCDs are generally cheap...if you buy them in large quantities. Small quantities are a bit harder, especially when they go in and out of production on short notice like they do.
The other cost factor is building the interface between the LCD and the CPU. Usually you're designing a custom flex ribbon and interface board along with a power supply for the backlight. Not many of these cheap LCDs use LVDS, they're all still on old-school parallel TTL.
This is a long shot, but in one of the videos on the hardkernel.org site they show the ODROID-X working with four miniature web cams. Does anyone know what brand of webcam they are? They look like they would work very well for a project I'm working on to record pollinators visiting flowers.
I think one of the great things about the Raspi is that it is not competing on being powerful, it's got an community aiming at an accessible and fun experience.
$130 vs $35 is not a signifigant difference? The difference is the Pi is an impulse buy (when you can get it). For me, anything over $100 gets asked 3 questions: what am I using it for? How much time can honestly spend with it (the most important question)? Is it worth the price? Negative on 1 and 2 and yes for 3.
The thing about the Pi is I don't feel guilty when I don't use it. Nor do I when I leave it on for weeks.
Nice, looks like a fun little box. Although its another data point in the 'how-the-heck-is-ouya-going-to-cost-99-dollars' question. I wonder if Samsung is more forthcoming on how to program the GPU than either Broadcom (RPi) or TI (Panda/Beagle) are.
There's some intent in Ouya to whore themselves out as some kind of developer platform, when all that is wanted, desired, or necessary is a solid stable reference Android platform for developing games on. Ouya seems to want to inject themselves as "something more," without making any statements about what is short about Android at large, aside from a lack of a passable GPU in the Google TV world.
I suggest you wait until the new GPU architectures come out at the end of the year or early next year. They all have OpenCL 1.1 support, including Mali T604, Adreno 320, Tegra 4 and PowerVR 6200.
Mali T658 coming out late 2013 will have 4x the performance of Mali T604, I believe thanks to the shared cache between the CPU and GPU, which should increase the compute performance.
Not to be a sour-puss but there are a large number of these boards available now. It's exciting, but there really are quite a few of them. I think XBMC may be entering an even newer phase of awesome with the ubiquity of these.
edit: How does one find out what video codecs are supported accelerated by the Mali 400?
Yea, I am not sure if "quad-core Raspberry Pi" is that much of a selling point. CuBox is much more interesting with:
- gigabit ethernet
- e-sata port (with port multiplier)
- infrared receiver
- out of the box, already comes with a box
- comprehensive wiki for a development platform
By the way, a 150 Mbps wifi dongle is sold for about 5 USD here (I too was surprised by how cheap the wifi module is, and went to check the price list).
I own a openrd platform, which isn't perhaps quite as up to date as the CuBox chipset, and it too has many of the features you mention.
But it is not a fast system. Gigabit ethernet? Sure, but good luck getting anywhere near that speed. Similar for the e-Sata port. Oh, sure, I can get a 60 mbps samba share up on the thing, but that caps out it's cpu.
The specifications for what it supports does not indicate what kind of performance it will have. The quad core Samsung processor will almost certainly dominate the Marvell unit's performance. Or maybe not: I don't have one. I've just been underwhelmed by what throughput Marvell has delivered me.
Any idea what bus the eSATA port is attached to? the DreamPlug is similar to the CuBox but the eSATA is hooked up internally via USB2, limiting its speed dramatically
They do finally have stock? There was a thread just earlier this week where I mentioned this but there was a linked forum post where they've not been fulfilling preorders, etc.
Plus there's some straight up weird stuff on their Word doc exported "Spec Sheet".
I was just thinking that yesterday: lots of small android/allwinner A10 based RPI competitors coming out, eg this one which i really want, after failing miserably at getting my rpi to do anything useful: http://www.dealextreme.com/p/ak802-mini-android-4-0-network-...
Its priced about the same as a rpi when you consider that it includes wifi and a power supply.
>failing miserably at getting my rpi to do anything useful
Really? What are you trying to do with it? My RPI won't play 1080p video but I've got a tiny little app using omxplayer to play back HD content from my server very, very reliably.
I'm excited to see what the future brings in this nook of the technology/hobbyist industry.