[Oberon] Re. Other ARM platform

Chris Glur easlab at absamail.co.za
Mon Apr 30 05:35:21 MEST 2007


Maksimiuk Darek wrote:
>   There is very powerful alternative to the  "bare" ARM processor.  
>   Please have a look at : 
> http://www.atmel-grenoble.com/dyn/resources/prod_documents/7001s.pdf
> The D740 processor is a dual core system that consists of ARM processor 
> and VLIW DSP core 
> called mAgic.  It does a complex multiplication within one signal clock. 
> It gives 1 GFLOP peak performance @ 1.2 W 
>   (1024 points complex FFT in 60 us) !!!!
> 
>  I think, this could be a good candidate for the AOS system !!!
> 
Texas Instruments also has a version with DSP facilities.
Many portable devices will want dsp routines: audio..modems..etc.
I guess the dsp would be programmed, separately from [or using]
the AOS facilties ?  Already N-O has a terrible/unintuitive
'set-based' syntax for register, memory, port programming.

I speculate that AOS could make some better tool[s] to
program the dsp.  But I don't see that the dsp instructions
would be used in AOS applications ?

juergkoch wrote:
> I had Bluebottle up and running in 2002. Fine thing! Five years ago
> (this is no joke), it gave me nevertheless the impression that it
> demanded too much on hardware compared to the benefit it would 
> give over native oberon. 

I've also been waiting for AOS to mature.
Apparently it's pre-emptive [? multi-threaded]; does this mean
it should use multiple CPUs for full benefit ?

BTW, although I haven't yet studied the ARM instruction set, 
its popularity is also due to the [6502 effect] simplicity; and
the lower power consumption too.
IMO combining complexity [eg. in the software] with simplicity
of hardware is not a good idea.

== Chris Glur.



More information about the Oberon mailing list