Has anyone tried the Edison? Is it good? It seems to me that it's a Raspberry Pi with better specs, wifi, etc, plus Arduino compatibility, which sounds amazing, but I haven't figured out how it connects to stuff. How do I power the little die? Do I need breakout boards for everything?
The Raspberry Pi is intended for a different use - it's comparing apples to oranges.
The Edison is intended for embedded applications - it has no display support built in (like DVI/HDMI), less USB support, and less "plug-and-play" ports.
Sure, but I (and most of the hackers I know) use the Raspberry Pi like an Edison (i.e. only Wifi connected to USB, headless/SSH only, GPIO pins), so the Edison would be a better fit for my projects, if I could figure out how it works.
As the SparkFun article mentions, with the cost of the Wifi, powered hub, cables, SD card, and Bluetooth adapter add up. The Raspberry Pi is meant as a low-cost teaching computer, whereas what I want is a SoC to hack on.
Maybe it's as simple as programming it with the board and then detaching it and powering it.
See, that's where I don't get what Edison is about.
If I want a SoC capable of doing low cost user-interface type embedded tasks I need parallel LCD support (or at least LVDS). I'd get an NXP LPC18/40 or an STM32 at the low end or an i.MX5/6 at the high end. Way cheaper with way better application support.
If I wanted to do IoT sensing/reporting, this whole thing is just overkill. Again, a low cost LPC or STM coupled with a GainSpan or TI radio module (that I can bake directly to my PCB). To me, and I could be wrong here, IoT isn't about putting 500 MHz computing resources at the nodes and certainly not for a $75 BOM adder.
At least the LPC and STM are chips I can put directly down on my board and run out of the box. An SOM with a single-sourced connector (and Hirose is notorious for designing unique stuff nobody else wants to duplicate) is just commercial suicide.
I want a low-power, Wifi-enabled computer for various projects. Arduino shield compatibility means that I can create things like my license-plate-reading garage door opener, or irrigation system, while combining a fast processor and lots of RAM with all the nice hardware Arduino has.
And that's fine for that kind of application. You could also do just as much with a cheaper system (but you'd need a lot more software work so maybe that's Intel key point: you can shovel in all the X86 code you've been slogging around with you for decades).
But, in my opinion, the examples you mentioned are not IoT. Unless you're planning to put one of these at every sprinkler head.
Not sure how power is applied, it's not as obvious as for the Arduino (since the Edison is more highly integrated and more designed to be built into products than to be directly fiddled with, I guess).
pcduino3 has been available for a while now with wifi built in - I've found these boards quite useful for mobile applications. I've been using them to run a django app quite nicely where the raspi wasn't quite powerful enough!
Oh, that sounds great! Is it Arduino compatible? How do you use all those libraries? I don't understand how something that runs Linux can be Arduino compatible, since Linux isn't a RTOS.
It's neat, but hard to come up with a use case that's not a part of a commercial product just yet. As a unit the 0.4mm pitch connector is absolutely impossible to solder in a DIY environment, you must buy a breakout board, and suddenly it's not a sub $50 device anymore. All of the use cases I could think of were better served by a device I could actually tinker with, so I hedged on a Minnowboard Max instead.
its a gimmick, there is no clear market for it, no clear use scenario, no real documentation, a lot of design fails (230Hz GPIOs IF/when Intel lets you use buildin Quark, its TURNED OFF now)
In terms of the specs, the Edison is great. After tinkering around with it for a bit I found I didn't really like the bundled software Intel released with it.
At the end of the day I want something just like this, but with a net-installish base linux distro, and something with better temperature range than 0-40C! It's too complicated for tinkering, and the temperature range is too narrow for commercial products. It falls into the middle, where it's probably not great for anyone. I'm sure Intel can fix that.
I'm guessing the reason is cost (I think the quark chip alone is $9, + DDR3 memory + cost for BGA placements + 4 layer impedance controlled boards etc...)
I was stuck waiting in line during the opening ceremony this year. I'll keep an eye out though, I would love to test making my own Quark board, but I imagine even with an external assembly service / high quality boards doing the bring up would be extremely difficult.
amazing why? so you can have that sweet sweet compatibility with Windows 3.11? or access to cutting edge BCD opcodes?
maybe its the price point of 4x faster ARM SoCs? or 1Wat at idle when connected to a network?
what tools do you mean? other than binary only windows applications that you wouldnt run anyway because there is no GPU on this thing(unless you manage to install headless windows).
Just because the source is available doesn't mean the tool is portable, or even ported. You won't be able to apt-get install afl-fuzz (the tool used in this article) on Arm.
There's something wrong with the claim of "1Watt idle when connecting to a network", since the maximum power of that module is 1Watt at full performance.
Maybe the person who measured this hasn't used sleep modes properly?
I think the author is wrong about the CPU supporting HT. Merrifield is based on the Silvermont micro-architecture, which does away with HT and replaces it with OOE.
Maybe I could use the Edison as a foundation to build something I've been after for a while.
I'd like a 'thing' that would attach to the back of a pair of bookshelf speakers and allow me to stream music through the speakers via Bluetooth. The 'thing' would also need to act as an amp. I had seen 'The Vamp' but this doesn't offer stereo.
How did I completely miss the Edison? I love the idea of a dual-core x86 for running something like linux with a moderately powered MCU as a RTP. Does anyone know what it means by "<1W of power consumption" which is the power I've seen? I assume it will peak at over 1W with all 3 cores running all out? I'd like to see separate idle/loaded values for that.
37 comments
[ 3.0 ms ] story [ 62.3 ms ] threadThe Edison is intended for embedded applications - it has no display support built in (like DVI/HDMI), less USB support, and less "plug-and-play" ports.
SparkFun did a great write-up on unfair comparisons between the two platforms: https://www.sparkfun.com/news/1603
As the SparkFun article mentions, with the cost of the Wifi, powered hub, cables, SD card, and Bluetooth adapter add up. The Raspberry Pi is meant as a low-cost teaching computer, whereas what I want is a SoC to hack on.
Maybe it's as simple as programming it with the board and then detaching it and powering it.
If I want a SoC capable of doing low cost user-interface type embedded tasks I need parallel LCD support (or at least LVDS). I'd get an NXP LPC18/40 or an STM32 at the low end or an i.MX5/6 at the high end. Way cheaper with way better application support.
If I wanted to do IoT sensing/reporting, this whole thing is just overkill. Again, a low cost LPC or STM coupled with a GainSpan or TI radio module (that I can bake directly to my PCB). To me, and I could be wrong here, IoT isn't about putting 500 MHz computing resources at the nodes and certainly not for a $75 BOM adder.
At least the LPC and STM are chips I can put directly down on my board and run out of the box. An SOM with a single-sourced connector (and Hirose is notorious for designing unique stuff nobody else wants to duplicate) is just commercial suicide.
But, in my opinion, the examples you mentioned are not IoT. Unless you're planning to put one of these at every sprinkler head.
You typically need an "Edison Block" (like an Ardunino Shield) to connect to things, see e.g. SparkFun's portfolio at <https://www.sparkfun.com/categories/272/>.
Not sure how power is applied, it's not as obvious as for the Arduino (since the Edison is more highly integrated and more designed to be built into products than to be directly fiddled with, I guess).
its a gimmick, there is no clear market for it, no clear use scenario, no real documentation, a lot of design fails (230Hz GPIOs IF/when Intel lets you use buildin Quark, its TURNED OFF now)
At the end of the day I want something just like this, but with a net-installish base linux distro, and something with better temperature range than 0-40C! It's too complicated for tinkering, and the temperature range is too narrow for commercial products. It falls into the middle, where it's probably not great for anyone. I'm sure Intel can fix that.
I'm guessing the reason is cost (I think the quark chip alone is $9, + DDR3 memory + cost for BGA placements + 4 layer impedance controlled boards etc...)
Good to see you understand the purpose and design constraints oh so well...
apt-get whatever works just fine on arm
Maybe the person who measured this hasn't used sleep modes properly?
I'd like a 'thing' that would attach to the back of a pair of bookshelf speakers and allow me to stream music through the speakers via Bluetooth. The 'thing' would also need to act as an amp. I had seen 'The Vamp' but this doesn't offer stereo.