*Stable* platforms for RIOT-OS

Hello everyone,

Could you tell me what are the best hardware platforms to work with RIOT OS?

To be honest, I need to get on with my PhD-related work, and MSP430-based platforms are still to unstable to progress fast enough. I'll still try to use and debug it as much as I can, but I now need to obtain some results as soon as I can.

So I'd like to know what boards/motes you're using when you need to have things done quickly, things that must "just work".

Thanks in advance,

Hi Kévin!

Could you tell me what are the best hardware platforms to work with RIOT OS?

Good question. The origins of RIOT were mostly developed for ARM7 platforms, i.e. the MSB-A2 [1]. Unfortunately, this platform is not available on the market and probably the only feasible way to access it from outside FU Berlin is to register on the DES testbed [2]. However, this is probably not a viable solution for low-level programming stuff like MAC protocols.

Recently, a lot of work was going on for ARM Cortex-M platforms, particular STM based boards. In the near future, several people from the RIOT community will most probably focus on the nodes used in IoT-Lab [3], the STM discovery boards and Arduino. So, even though these platforms still have some open issues, I guess that they will mature soon. For the IoT-Lab node we're planning to merge the board/cpu dependent code from thirdparty_* to RIOT's main repository this week.

To be honest, I need to get on with my PhD-related work, and MSP430-based platforms are still to unstable to progress fast enough. I'll still try to use and debug it as much as I can, but I now need to obtain some results as soon as I can.

I can understand your frustration - especially, I know the feeling when trying to get some results from experiments and having to struggle with stupid, f*cking hardware/driver/low-level problems instead. However, personally, I like the MSP-430 platforms a lot and still hope that we can find the problems there, so that these boards can be more useful with RIOT in the future.

Best, Oleg

[1] Create new page · RIOT-OS/RIOT Wiki · GitHub [2] http://www.des-testbed.net [3] Create new page · RIOT-OS/RIOT Wiki · GitHub

Hello Oleg,

To be more precise on my case, it's very hard for me to work on MAC protocols because of problems with hwtimers on MSP430 motes (TelosB and Z1) : the timers randomly fail to fire after a variable (random) time (from 5 seconds to 1 minute), and this prevents me from getting "good enough" results. And my boss at the lab is becoming quite concerned since he can't see any practical result coming out from my work.

As you said, it's very frustrating since I strongly think that RIOT's code is clean: we are facing some obscure issue with timers, and TI's documentation and "quality assurance" are lacking, to say the least (seeing issues like the one I mentioned in issue #426 on GitHub is quite appalling)... MSP430 CPU design is clean, but the peripherals that come around it are *much* less clean...

I'm looking forward to using RIOT on Arduinos: we already have a port for the Due (Cortex-M3), and I see that porting on AVR8 (Mega 2560) is going on... This may be a huge step forward.

Best regards,

Le 24/06/2014 19:03, Oleg Hahm a �crit :

Hello again,

Le 24/06/2014 19:03, Oleg Hahm a �crit :

[...] I can understand your frustration - especially, I know the feeling when trying to get some results from experiments and having to struggle with stupid, f*cking hardware/driver/low-level problems instead. However, personally, I like the MSP-430 platforms a lot and still hope that we can find the problems there, so that these boards can be more useful with RIOT in the future.

Speaking about MSP430, maybe I just found a clue! Look at #426.

Best, Oleg

[1] Create new page · RIOT-OS/RIOT Wiki · GitHub [2] http://www.des-testbed.net [3] Create new page · RIOT-OS/RIOT Wiki · GitHub

_______________________________________________ devel mailing list devel@riot-os.org http://lists.riot-os.org/mailman/listinfo/devel

Regards,