hardware architectures to run Ada on in 2008

hardware architectures to run Ada on in 2008

[Yochi Choresh]

How weird does it sound,
to intend to run GNAT compiled code on a PowerPC on top of some Linux
kernel
as the embedded system steering e.g. a drone?

Or is this something, that the industry has been doing already for a
couple of years?

Y.

Re: hardware architectures to run Ada on in 2008

[Jeffrey Creem]

Yochi Choresh wrote:
> How weird does it sound,
> to intend to run GNAT compiled code on a PowerPC on top of some Linux
> kernelor L
> as the embedded system steering e.g. a drone?
> 
> Or is this something, that the industry has been doing already for a
> couple of years?
> 
> Y.

It does not sound weird at all. It is also (as far as I know) not a 
configuration that is currently listed as having GNATPro support 
available for it which is not to say that you could not get support if 
you asked.

Various Linux kernels have gotten quite a bit better in the last few 
years with respect to real-time performance. I would expect a more 
typical configuration for the application you suggest would be vxWorks 
or LynxOS running on a powerPC running GNAT code. If you open beyond 
GNAT to other Ada compilers than things like Greenhills and a few others 
would also become possibilities.

Re: hardware architectures to run Ada on in 2008

[Niklas Holsti]

Yochi Choresh wrote:
> How weird does it sound, to intend to run GNAT compiled code on
> a PowerPC on top of some Linux kernel as the embedded system
> steering e.g. a drone?
> 
> Or is this something, that the industry has been doing already for a
> couple of years?

Quite realistic, not weird at all. About six years ago I was 
involved with an ESA project that built a prototype Mars rover. We 
used GNAT on a PC104/Intel x86 computer on-board the rover. The 
computer ran Linux from a DiskOnChip device. For development work 
we connected an IDE hard disk to the PC104 stack. We wrote our own 
Linux device drivers for the analog and digital I/O cards. We used 
plain Linux, no RT variant. The rover was controlled through a wire 
tether using UDP over PPP. The other end of the tether connected to 
a PC with a Java GUI.

For a production rover we would surely have used some other 
processor and a small kernel or Ada RTS. But for a ground-based 
system Linux+GNAT was OK.

-- 
Niklas Holsti
Tidorum Ltd
niklas holsti tidorum fi
       .      @       .

Re: hardware architectures to run Ada on in 2008

[Jeffrey R. Carter]

Jeffrey Creem wrote:
> 
> Various Linux kernels have gotten quite a bit better in the last few 
> years with respect to real-time performance. I would expect a more 
> typical configuration for the application you suggest would be vxWorks 
> or LynxOS running on a powerPC running GNAT code. If you open beyond 
> GNAT to other Ada compilers than things like Greenhills and a few others 
> would also become possibilities.

The OP might also want to consider RTEMS or MaRTE if Linux is not sufficiently 
real time.

http://www.rtems.com/
http://marte.unican.es/

-- 
Jeff Carter
"Help! Help! I'm being repressed!"
Monty Python & the Holy Grail
67

Re: hardware architectures to run Ada on in 2008

[Stephen Leake]

Yochi Choresh <Yochi.Choresh@gmail.com> writes:

> How weird does it sound,
> to intend to run GNAT compiled code on a PowerPC on top of some Linux
> kernel
> as the embedded system steering e.g. a drone?
>
> Or is this something, that the industry has been doing already for a
> couple of years?

Better would be GNAT on Lynx.

Lynx (http://lynuxworks.com) is a true real-time system, with a POSIX
interface. AdaCore offers full support for this combination.

I use it at work, but on x86 targets. Lynx is not as nice for running
development tools (such as Emacs) as Linux, but it is nicely
real-time. Writing device drivers is easier in Lynx than Linux. And
the AdaCore support is wonderful.

Hmm. I run Emacs on Lynx on x86; Lynuxworks offers a "self-targeted"
gcc for that CPU. I think they don't offer a self-targetd CPU for
PowerPC; you have to use cross target. I don't know why; you should
ask AdaCore and/or Lynx.

In any case, GNAT + Lynx on PowerPC is still more real-time, and more
supported, than GNAT + Linux on PowerPC.

-- 
-- Stephe

Re: hardware architectures to run Ada on in 2008

[Samuel Tardieu]

>>>>> "Yochi" == Yochi Choresh <Yochi.Choresh@gmail.com> writes:

Yochi> How weird does it sound, to intend to run GNAT compiled code on
Yochi> a PowerPC on top of some Linux kernel as the embedded system
Yochi> steering e.g. a drone?

As long as you do not have hard real-time constraint, it is perfectly
ok. We run GNAT on a sh4 board (with the Ravenscar profile) for a
robotics contest.

  Sam
-- 
Samuel Tardieu -- sam@rfc1149.net -- http://www.rfc1149.net/

Re: hardware architectures to run Ada on in 2008

[Yochi Choresh]

On Feb 25, 10:49 pm, Samuel Tardieu <s...@rfc1149.net> wrote:
>>>>> "Yochi" == Yochi Choresh <Yochi.Chor...@gmail.com> writes:

  Yochi> How weird does it sound, to intend to run GNAT compiled code
on
  Yochi> a PowerPC on top of some Linux kernel as the embedded system
  Yochi> steering e.g. a drone?

  Samuel> As long as you do not have hard real-time constraint,
  Samuel> it is perfectly ok.

Do you suspect, "my" configuration does not withstand real-time
constraints?

RT-Linux is still an option.

Do you doubt, that's good enough to fulfill such constraints?

  Samuel> We run GNAT on a sh4 board (with the Ravenscar profile)
  Samuel> for a robotics contest.

Re: hardware architectures to run Ada on in 2008

[Ludovic Brenta]

Yochi Choresh wrote:
> Do you suspect, "my" configuration does not withstand real-time
> constraints?

What are your real-time constraints?

--
Ludovic Brenta.

Re: hardware architectures to run Ada on in 2008

[Yochi Choresh]

On Feb 26, 8:06 pm, Ludovic Brenta wrote:

Yochi Choresh wrote:

YC> Do you suspect,
YC> "my" configuration does not withstand real-time constraints?

LB> What are your real-time constraints?

Steering a low-speed flying device and transmitting acquired-data.

I am sorry, I didn't mention that again.
Thanks for your patience.

Y.

Re: hardware architectures to run Ada on in 2008

[Ludovic Brenta]

Yochi Choresh <Yochi.Choresh@gmail.com> writes:

> On Feb 26, 8:06 pm, Ludovic Brenta wrote:
>
> Yochi Choresh wrote:
>
> YC> Do you suspect,
> YC> "my" configuration does not withstand real-time constraints?
>
> LB> What are your real-time constraints?
>
> Steering a low-speed flying device and transmitting acquired-data.
>
> I am sorry, I didn't mention that again.
> Thanks for your patience.

These are not real-time constraints; these are functional
requirements.  What are your real-time constraints, i.e. the rate with
which external events will arrive to your system and the maximum
acceptable latency for processing them?  Is the transmission of
acquired data subject to any real-time constraint at all?

-- 
Ludovic Brenta.