Re: End of optimization...

Let me add back a bit of the context for this reply:


> Fergus Henderson <fjh@cs.mu.oz.au> relied:
>
> mayan@sandbridgetech.com writes:
>
> >Compilers only make sense for platforms - i.e. for some kind of
> >generally programmable hardware for which a moderate number of
> >applications will be developed.
>
> I think this ignores technology improvements which mean that it is
> possible to cheaply generate a somewhat optimizing compiler for
> a new chip, by just giving a machine description for the new chip.
>


and I replied with some skepticism about retargetable compilers.


> Alan Lehotsky <qsmgmt@earthlink.net> writes:
>
> >
> > Well, I missed much of this thread, but I did a gcc port to a
> > specialized CPU . . . The compiler, assembler and linker were
> > functional in 2 man-months (4 calendar months). The compiler was used
> > to write EXACTLY one program . . .
> >
> > So, I believe that retargetable compilers are not a myth....


Indeed having spent a fair amount of time (15-20 years ago) working on
retargetable compilers based on general machinery and using machine
descriptions, I came to feel that there were only a few situations
where it made sense to do things that way, and the one you describe is
not one of them. Retargeting gcc, lcc and other such specific
compilers does indeed seem to be an effective way to go.


Fergus Henderson has posted another note in this thread pointing to
retarget.com indicating that the other approach is still alive.


-- Bob


--
Robert L. Knighten
Robert@Knighten.org

Return to the comp.compilers page.
Search the comp.compilers archives again.