[ECOS] Are eCos gnutools planned to be released with a more recent(>3.2.1) version of libgcc

[ECOS] Are eCos gnutools planned to be released with a more recent(>3.2.1) version of libgcc

[etienneg]

Hi

I'm currently working on porting eCos on Motorola PowerPC 860
(400Mhz). I experimented problems using floats while treating an IT of
3ms period.

float's handling took me too much time, for exemple:
 conversion float -> ulonglong = 500 µs
 float x float                 =  45 µs
 float / float                 = 100 µs

in comparison with DIAB's library, these measures are incredibly long
(especially conversions to or from floats). I tried to replace most of
the float's libGCCv3.2.1 functions by float's libGCCv3.3.3 functions. 

 conversion float -> ulonglong = 17 µs
 float x float                 = 10 µs
 float / float                 = 42 µs

Thx to this modification, I do not waste time anymore :
libGCC 3.2.1 -> IT is treated in 2,5ms
libGCC 3.2.1 -> IT is treated in 1,2ms


Obviously there's a great improvement in recents versions of the GCC
library. Are eCos gnutools planned to be released with a more
recent(>3.2.1) version of libgcc?

--
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] Are eCos gnutools planned to be released with a more recent(>3.2.1) version of libgcc

[Gregory HEINRICH]

euh... tu connais l'existance des DSR ? ca te permet de differer le
traitement de la routine d'interruption. je sais pas si c'est ce que
tu cherchais...
tu as reçu des réponses à ta question ?

-------------------
> Hi
> 
> I'm currently working on porting eCos on Motorola PowerPC 860
> (400Mhz). I experimented problems using floats while treating an IT
of
> 3ms period.
> 
> float's handling took me too much time, for exemple:
>  conversion float -> ulonglong = 500 µs
>  float x float                 =  45 µs
>  float / float                 = 100 µs
> 
> in comparison with DIAB's library, these measures are incredibly
long
> (especially conversions to or from floats). I tried to replace most
of
> the float's libGCCv3.2.1 functions by float's libGCCv3.3.3
functions. 
> 
>  conversion float -> ulonglong = 17 µs
>  float x float                 = 10 µs
>  float / float                 = 42 µs
> 
> Thx to this modification, I do not waste time anymore :
> libGCC 3.2.1 -> IT is treated in 2,5ms
> libGCC 3.2.1 -> IT is treated in 1,2ms
> 
> 
> Obviously there's a great improvement in recents versions of the GCC
> library. Are eCos gnutools planned to be released with a more
> recent(>3.2.1) version of libgcc?
> 
> -- 
> Before posting, please read the FAQ:
http://ecos.sourceware.org/fom/ecos
> and search the list archive:
http://ecos.sourceware.org/ml/ecos-discuss
> 
> 

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