🚀 go-pugleaf

From article <3903@ssc-bee.ssc-vax.UUCP>, by carroll@ssc-vax (Jeff Carroll):
> In article <8930@dnlunx.pttrnl.nl> venemans@dnlunx.UUCP (Venemans P.H.A.) writes:
>>What C compilers are available for the Motorola DSP56000 chip? I have
>>heard of the Motorola compiler, but some people told me it is a rather
>>buggy one.  Is that true, and -if so- are there any alternatives? (Did
>>someone already adapt gcc?). 
> 
> 	The local Moto guy told me not to buy the 56K C compiler - that I
> should use assembler instead.
> 
> 	I have some literature from a third party vendor that suggests that
> this compiler puts most of its data in Y memory. This could adversely affect
> efficiency of your code on some applications.
> 
> 	A C compiler for this chip would present some interesting challenges.
> 
> 
> -- 
> Jeff Carroll
> carroll@ssc-vax.boeing.com
> 
> "Do you think I care? ... I have an infinite amount of money."	-Bill Gates

I have worked with the 56000, TI320x0, and AD2100 series of DSP chips.  I
haven't had the chance to play with the C compiler for the AD2100 yet.
However, the C-compiler for the 56000 was terrible.  I was quite impressed
with the one for TI's chips.  The C-compiler for the Moto part could use
some optimization techniques built in.  And yes, the biggest complaint I had
was that I couldn't force variables into both X/Y data spaces.  I did end up
writing most of my stuff in assembly (which I liked better than TI's).
What's interesting is the AD2100 chip.  The bus structure is almost identical
to the 56000.  The sales rep I talked to claimed that their C-compiler would
allow me to declare variables into either data space.  (I have my doubts
about this one, but I'll wait and see.)

Dave Scholten