Re: Commercial compiler constructed via generator?

> > The problem is: Are there examples showing that a well-designed
> > compiler generator doing better than hand-coded program in
> > average during some compiler construction stages?
>
> I'm not sure if I understand your question 100% but in the report on the
> COCO/R project Hanspeter M/ssenb/ck stated that a COCO-generated
> front-end for the ETH Oberon-2 compiler had a better performance than the
> earlier, handwritten front-end.
>
> You should find the COCO/R report somewhere on ftp.inf.ethz.ch.


It's no longer on the ETH servers so far as I know. You can find a copy on


          cs.ru.ac.za /pub/coco Coco.Report.ps.Z
          ftp.ssw.uni-linz.ac.at /pub/Reports Coco.Report.ps.Z


...


"the Coco/R generated compiler is 23% faster than the original compiler.
This is due to the fact that the generated compiler reads the source text
into main memory at once, while the original compiler reads it character by
character. Without this improvement the generated compiler is about 10%
slower than the original one" ...






Pat Terry, Computer Science, Rhodes University, GRAHAMSTOWN 6140, RSA
cspt@cs.ru.ac.za or cspt@giraffe.ru.ac.za or pdterry@psg.com
Voice +27-461-318291 or +27-461-318292 FAX +27-461-25049
--

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