[ECOS] eCos on MPC8313

[ECOS] eCos on MPC8313

[RONETIX - Ilko Iliev]

Hello,

Does somebody use MPC8313 with eCos?
If yes, could be possible to share the port?

We want to test the network performance of this CPU with eCos.

Did somebody make a comparison of the network performance of eCos and Linux?

-- 
With best regards,
Ilko Iliev
Ronetix Development Tools GmbH
CPU Modules, JTAG/BDM Emulators and Flash Programmers
Waidhausenstrasse 13/5, 1140 Vienna, Austria
E-Mail: iliev@ronetix.at; Web: www.ronetix.at


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] eCos on MPC8313

[Andrew Lunn]

> Did somebody make a comparison of the network performance of eCos and Linux?

In general, eCos is slower. eCos has different aims to Linux. eCos is
an RTOS, and so does not want the TCP/IP stack, which by definition is
none real time, from upsetting the real time behaviour. So eCos tries
to do most of the networking in threads, where as Linux has it in the
kernel. By placing networking in threads, they can be preempted by
higher priority real time work. However this introduces overheads,
which slows down performance.

If you want high speed networking, consider using Linux. If you want
real time performance, consider using eCos. 

     Andrew

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] eCos on MPC8313

[Edgar Grimberg]

Hi,
> Hello,
>
> Does somebody use MPC8313 with eCos?
I wrote a port for MPC8313E revision 2.1 (and a revision 1.0, because I 
had to :) .
> If yes, could be possible to share the port?
 The decision to make it public has not been made yet, also the code 
needs some cleaning up of the client's specifics before it can be posted 
and I'm not sure when I'll have the time to do it.
> We want to test the network performance of this CPU with eCos.
If you have specific code to test, I can help you with that.
> Did somebody make a comparison of the network performance of eCos and 
> Linux?
I have not made a comparison.

Regards,
Edgar

-- 
Edgar Grimberg
System Developer
Zylin AS
ZY1000 JTAG Debugger http://www.zylin.com/zy1000.html
Phone: (+47) 51 63 25 00


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] eCos on MPC8313

[RONETIX - Ilko Iliev]

Dear Edgar,

Edgar Grimberg wrote:
> Hi,
>> Hello,
>>
>> Does somebody use MPC8313 with eCos?
> I wrote a port for MPC8313E revision 2.1 (and a revision 1.0, because 
> I had to :) .
>> If yes, could be possible to share the port?
> The decision to make it public has not been made yet, also the code 
> needs some cleaning up of the client's specifics before it can be 
> posted and I'm not sure when I'll have the time to do it.
Maybe we can make the cleaning if you just remove the specifics and send 
us the rest - we will correct the makefiles and the scripts.

>> We want to test the network performance of this CPU with eCos.
> If you have specific code to test, I can help you with that.
We measure the network performance with the iperf tool, but it will be 
not easy to use it under eCos.
Could be possible just to make a simple FTP test on a Gigabit Ethernet 
connection?


Regards,
Ilko

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] eCos on MPC8313

[RONETIX - Ilko Iliev]

Hello Andrew,

Andrew Lunn wrote:
>> Did somebody make a comparison of the network performance of eCos and Linux?
>>     
>
> In general, eCos is slower. eCos has different aims to Linux. eCos is
> an RTOS, and so does not want the TCP/IP stack, which by definition is
> none real time, from upsetting the real time behaviour. So eCos tries
> to do most of the networking in threads, where as Linux has it in the
> kernel. By placing networking in threads, they can be preempted by
> higher priority real time work. However this introduces overheads,
> which slows down performance.
>   
I'm not sure if the real time behavior is a reason for slower networking 
if the CPU is not under heavy-duty.

regards,
Ilko

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] eCos on MPC8313

[Edgar Grimberg]

Hi,

>> Maybe we can make the cleaning if you just remove the specifics and send
>> us the rest - we will correct the makefiles and the scripts.
>
> It's a bit more than that, both technically and decision wise...
>

I've managed to commit the first attempt of eCos port for MPC8313E_RDB
(subversion at http://www.ecosforge.net/ecosforge/trunk/ecos_mpc8313).
There are some hacks ( /hal/powerpc/arch is duplicated) and some
hardcoded things in the ethernet driver.
This is work in progress: code is a cut/paste from a project, so
things are likely to be missing,  PHY is not working and there is a
horrible hack in reset words related to this, and so on.
Wrt tools, I suggest gcc 4.4.0 and binutils 2.19.1 because they
support e300c3 core "out of the box". I had worked with patched
versions of gcc 4.3.2 and binutils 2.18 to support e300c3, I'm in the
process of switching to the latest and greatest official tools.
Any help is appreciated!


 --
 Edgar Grimberg
 System Developer
 Zylin AS
 ZY1000 JTAG Debugger http://www.zylin.com/zy1000.html
 Phone: (+47) 51 63 25 00

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss