Porting NETMF to Cortext Ax processor?

What would be involved in porting NETMF to a Cortex Ax processor board?

The 1GHz+ speed and the abundant RAM would make for a very fast, low cost processor module. Itead Studio have an AW2041 board that is similar to the G400 idea and fits into an SODIMM socket. Something like this from GHI would be awesome processing power.

I realise there is Windows 10 for some boards but Windows 10 adds to much bulk and I like the simplicity of NETMF and no OS.

3 Likes

@ Dave McLaughlin - The AW2041 appears to be dual core. My guess is that .netmf wouldn’t know what to do with the 2nd core. But to your point a +1Ghz .netmf processor would be awesome.

If you’ve got the horsepower to run a real OS, you’re better off (for now) running real .NET and not hobbling yourself with the interpreter.

@ Dave McLaughlin - For such a MCU i would consider CMSIS, FreeRTOS, FatFS and C++ 11 programming. If you want to port netmf you lose a lot of time on reverse engineering netmf by the lack of documentation.

@ RobvanSchelven - Occasionally I checkout development environments for applications that don’t fit well in .Net MF. The tools you listed all look good but can you recommend an IDE for C++ 11 that you like or comes close to Visual Studio?

2 Likes

The only other option I have is Android and even the fact that this offers excellent debugging over USB etc, the issue I have is when you use a network connection, especially GPRS, then you run into the issue of excessive data because Android is doing a lot of stuff in the background that uses up data. Even a full blown .NET OS will do the same.

This is why I like NETMF, there is no background network traffic. If I dial up with GPRS and do no network stuff, the GPRS network will drop my connection after a certain time. With Android and a full OS there is enough background traffic to keep the connection alive and use up the precious data.

Like Gene, I like the VS setup for working so changing to something that most likely uses Eclipse will also require a steep learning curve. C# and NETMF is a jump straight into it option for me. :slight_smile:

Who cares, just port the thing. If I wanted a RTOS on a Cerb40 I’d write the code in C! The point of NETMF is ease of use and ability to debug. As for that second core, ignore it. You don’t have to implement everything that the CPU has. Can you imagine the RLP performance on a 1Ghz processor, on it’s own core? That fact alone justifies the endeavor.

What if NETMF is ported to the Quark processor that comes in the Intel Edison? Now that’s a thought.

2 Likes

@ Mr. John Smith - Have you ever tried to test en debug RLP code?

:wall:

Thats a good question. Until now i did not found a solution as nice as what VS with plugin offers. Sometimes i create a VS project holding the same files as in the ARM/KEIL uVision project. Just to refactor and format.

@ Dave McLaughlin - .NET Core on Ubuntu. Or, Mono, for a more mature platform. Code and debug in VS, and all the performance of native compilation. Plus, you’re not limited to what NETMF has.

1 Like

Nope, and I’m not looking forward to it either.

Given a suitable setup, debugging native code isn’t bad at all; indeed, it’s pretty much exactly like debugging NETMF code. Really the only difficulty is that the open-source tools tend to be a bit complex to string together. The Eclipse/gdb/STLinkV2 experience, once you’ve got it working, is quite comfortable.

I just love to see discouragement from some community members (godefroi in particular) every single time someone thinks of making faster than G400 hardware for NETMF… not! ::slight_smile:

1 Like

@ iamin - I make no secrets about where I think NETMF fits. Why in the world would you try to compete on the high end, when a Pi is $35, can run the full framework, and has peripherals that would cost you hundreds of dollars in Gadgeteer-land?

Yes, I think it’s a waste of time to port NETMF to Cortex-A. Last I checked, we were each entitled to our opinions.

[quote=“godefroi”]Last I checked, we were each entitled to our opinions.
[/quote]
I agree with you on this and this point only :wink:

@ Mr. John Smith -

What if NETMF is ported to the Quark processor that comes in the Intel Edison? Now that’s a thought.

Now that’s what I’m talking about.

The one thing that really gets me about this community is the groupthink. As long as bigger, faster, better is your goal, you’re ok. If Gadgeteer is the solution to every problem, then we all get along. If Azure is the greatest thing since sliced bread, then you belong here. If Microsoft can do no wrong, then you’re Duke Nukem. If Chris Walker is a crook and Agent was designed from the beginning as a scheme to defraud people, you’re a friend. As soon as you disagree with any of that you’re suddenly persona non grata.

A dissenting opinion is not a personal attack.

Of course, nay-saying isn’t the same as dissenting, and a consensus view is not the same as groupthink. Like in most things in life, it all depends on one’s perspective and biases, I suppose.

1 Like

@ mcalsyn - 100% agree on the consensus vs groupthink. Interesting read: http://www.au.af.mil/au/awc/awcgate/army/groupthink.htm

I’m loving .NET Core and frankly would do just about everything there instead of NETMF if it weren’t for one very big problem - I can’t just turn off my devices. This makes the Pi (and other OS devices) impractical for lots of my uses. So, I believe there is definitely a use case for high end micros running NETMF. However, there are many more uses cases on the low end. So, I don’t think we can just draw a line and say if you need more than xxMHz then you should eat Pi.