We have the same issue. First lot of the product is built using Edison. We are exploring Samsung ARTIK IoT Platform - ARTIK 520 IoT Module to avoid changing mechanicals of our product.
It fits in the same form factor as Edison and our code is entirely portable to it.
Once we can change the mechanicals, ARTIK 530 is sweet too with $30 for 1K qty.
I would like to toot into the same horn regarding the Edison: Luckily I only started evaluating in in the beginning of the year only, so I did not make too many commitments yet, but it is just a pity:
- The Edison is a unique product, which makes it a bit less approachable than a Raspberry, for example because of the connector and 1.8V data lines. But that is also its strength! The form-factor is just awesome for the power it offers.
- I probably misjudged it, but I saw the Edison just getting traction with, for example, Google's IOT-Platforms adding support to it (this was the moment I started paying attention to it)
- Having bought one including its breakout board, I find it actually incredible approachable for what it is...
I really hope that there will be _some_ sort of replacement in the future. The other boards people posted here are also awesome, but it makes me a bit sad that the Edison is pushed over board this quickly... [just to make the statement]
Hi, is the Artic 520 something like pin-to-pin compatible with Edison?
Otherwise I was considering PICO boards from Technexion... take this as example:
Yes, it's more expensive, though performance-wise it may be comparable.
Any of you have experience with that?
Also longevity is an issue, wouldn't like to make the same mistake again.
Don't know what you think of this one::
It also uses an Intel Atom processor, has Yocto and Window support, which should work with math kernel that I might want to use.
Don't see much about wireless communication, ..... The good thing is that it doesn't get hit by the RED directive. Intel Edison CE certification according to RED directive
That PICO board is a find. It's the only one I've seen where you have a prayer of moving off Edison to something else without a major hardware redesign. That it has an (allegedly) Edison-compatible Hirose connector is a huge win. Though I wonder whether you could physically put this board on, say, an Edison Arduino breakout without the other two connectors getting in the way. Or the Sparkfun blocks.
The price point is comparable as well. The only thing I see missing is a realtime co-processor to stand in for the Quark, but I wonder how many designs need it or made use of it. It was one of the more intriguing parts of the Edison, for me anyway, and I was looking forward to seeing what it could do. Alas, I guess not.
But as you mention, some gauranteed EOL timeframes would be reassuring. If you can't trust a company as big as Intel to honor their commitments, then who?
No, ARTIK 520 is not pin compatible. We'll have to redesign our landing custom but everything will fit our current product.
PICO board looks good but I have two concerns. It's bigger and it doesn't have 5GHz WiFi. If these two things are not deal breaker for you, I think PICO board is good and worth a shot.
Edison never had a fair chance!
The device is brilliant, with dual core + MCU. And what ever else is hidden in there (GPU?).
But the documentation is largely closed, people needed to rely on drivers / kernel provided (last commit to 3.10.98 on 8 Sep 2016, and to the future WIP-3.19.5 on 24 May 2016). The MCU support never developed further (Windriver Helix promised but never happend), Rocket OS upstreamed to Zephyr, but Edison locked from using Zephyr. And then the aged Yocto build system based on aged Yocto Dizzy. The Yocto meta-intel-edison layer (last commit to master 11 Mar 2016).. Not to mention the much touted cloud service closed down also somewhere last year.
I agree, not much people jumped up to join in. yoneken got preempt_rt working, that must have been a lot of work.
But it's hard to figure out what can be done with this device (everything) just by experimentation.
For commercial products without set EOL date and crumbling development from Intel, it was a huge bet for most to gamble on Edison.
Such a waste. The development done prior to launch must have been enormous. And the dates above show Intel gave up already somewhere Mar 2016.
And now, with proper upstream kernel support around the corner, U-Boot near providing ACPI, I can only hope Intel will be concluding it gave up too early. And with all this work in place and the lessons learnt the hurdle to launch a new even more advanced x64 based device will be smaller.
Say it ain't so Intel?! Really disappointed in this announcement. I was so excited to finally see a small Intel architecture processor with a great little embedded Linux distro. Spent many hours learning Yocto and BitBake build system, only to find out the future of this platform is disappearing!
What the IoT / Maker & Embedded space was always missing was an Intel architecture product until Edison & Joule. Sure, I've got a couple Raspberry Pi's too. However, building for ARM can be such a pain! I was really happy to see i386 architecture so as to avoid having to recompile everything for ARM & running into the constant dependency headaches there. I was even going to build out a small Kubernetes cluster on these devices to really test out clustering them.
Now after only a short period of time, before even giving enough time to marketing this and growing the community around it, Intel decides to discontinue the entire product & presumably the line of business?! Really a huge mistake if you ask me too! Probably all came down to some earnings numbers in some exec's spreadsheet. This over obsessive capitalism and quarterly currency bookkeeping ideology has spelled the death of innovation yet again! Part of the reason FOSS exists if you ask me... We wouldn't be as far as we are with Open Source technologies these days if business execs and accountants all had their say! A very upsetting & tragic story. Especially so for all those people who have startups or other projects with thousands invested in developing! Wow... the lack of consideration or caring in the cold announcements are inexplicable.
Bah... at least Raspberry Pi still has ResinOS for a more minimal container-centric Linux distro. I suppose all of the choices are ARM based now... better get recompiling... crap!
For me it was also a bummer to hear this and the alternatives just do not seem to be there. Any news about alternatives that have WIFI, the same small form factor (I used the mini breakout) and the ability to act as an USB host next to reasonably extensive IO? I was almost there in my project but it seems unwise to invest in a technology that is now disgarded. Let alone that boards do not seem to be available in Europe anymore.