@ anthonys - I would not expect Windows 10 IoT Core to run on microcontrollers, at least not any time soon. It’s simply too large.
Here’s the basic stack, as I see it:
[ul]Full Windows: Desktops, Laptops, Tablets, Phones, etc.
Runs desktop apps (on x86/x64 procs) and UWP/Store apps (both x86 and ARM)
Windows 10 IoT Core
Designed for smaller devices like Pi, DragonBoard, etc.
Runs UWP apps only, and no desktop. Generally runs one app at a time, similar to “kiosk” mode
Includes code to handle I/O and sensors (I don’t think this code is in “big” Windows, but not 100% on that)
.NET Micro Framework
Optimized for the “tiny” world
Runs on mCus.
Does not run UWP or Win32 apps[/ul]
To add to the list above, the NETMF team is currently working on a project called Llilum (I think that’s the spelling) which promises to bring native execution speed to the NETMF world. There are others on the forum who are better versed in the status of this project, and hopefully they’ll chime in.
So I don’t think it’s a matter of IoT core “replacing” NETMF. They’re targeted at different devices and different (though often complementary) workloads.
Not sure its a wrong comparison,considering the price of a raspberry pi etc its worth investigating the jump to windows 10 core? Sometimes the extra investment in hardware could reduce effort by weeks.
The only main difference i see as the deciding factor is price and power usage at this point. Sure, windows 10 core may be an overkill for a sensor board but just don’t use the features you don’t need. If its solar then power usage wont be an issue either.
Another great benefit of mNET is that can be ported to windows core 10 easily…well i would assume so!
I think that for any given project you have to evaluate what’s available NOW that suits the needs of the project.
That might mean sticking with the smallest possible processor with the minimal additional components to serve your needs, powered by NETMF.
It might mean (if you’re willing to tolerate a rapidly moving target) using Windows IoT Core on a Raspberry Pi.
For one-offs, prototypes, and hobby projects, there’s probably no “right” answer. Computing power is fungible, and as long as the platform you choose supports the communications and/or sensors you want to use, any of them will do.
But if you’re going to production with a large number of devices, things probably get trickier. You’ll need to figure out how to turn that RPi prototype into a production board, and get it loaded with Windows IoT Core and your app. I’d argue that there are a lot more folks with experience taking NETMF projects to production than currently exists for Windows IoT Core.
In the end, the typical consultant answer of “it depends” applies here.
BTW - I agree with @ njbuch that eventually microcontroller hardware will get powerful enough to run IoT core, but still not sure whether that’s a desireable goal. Nice thing about NETMF is a smaller footprint, which means (among other things) less attack surface. For some workloads, a NETMF (or Llilum, or whatever comes next) that runs faster is still going to be a good choice.
I see this slightly more different… I think one day we won’t confuse everything with terms like “IoT Core”, “NETMF”, “Llilum”, “full .NET”, etc. One day everything will just support “.NET” and everything will be programmed through UWP (or whatever Microsoft renames it to by then…).
I’d like to see GHI add a SOM capable of running IoT Core to their catalog that’s suitable for commercial devices. Nothing coming from the Pi foundation meets that standard. Something in a G400 format would be great. Yes, similar to the old Pi Compute module…
Honestly, I haven’t pursued giving feedback on the Pi but I just assumed I wasn’t the only person with this complaint. Heck, just yesterday I was looking at a project on Hackaday where the guy had added an Arduino to the project solely for the purpose of telling the RPi to shut down…