Re: Multi-threaded compilers

Eric Gouriou wrote:
>
> I am looking for papers/references about parallelization of compilers
> (and NOT compilers for parallel machines), or distributed compilation.
>
> From my unsuccessful searches on the net, I could deduct that:
> a) Those who looked at it thought the problem was not worth
> of attack (too easy).
> b) Those who worked at it found distributed compilers useless.
> c) The only people interested are the compiler companies and
> are not interested in sharing their experience.


I worked on this topic sometime ago - thoretically.
My opinion tends to b). Although there are some results concerning
parallel attribute evaluation. But because of Amdahl's law the total
speed up is usally bound by three or four, since scanning and parsing
take a serious amount of time (20-40% on average) and both are done
sequentially before.
Thus, I searched for some really local parsing strategies and figured
out some essential properties. Short presentations are available in
    ftp://ftp.math.tu-dresden.de/pub/reports/wir/Lampe
A full description is there only hardcopy in German (please ask for it).
A very good (maybe a little outdated) overview about papers on this
topic is
  Akker, Alblas, Nijholt, Oude Luttighuis: An Annotated Bibliograhpy
    on Parallel Parsing. Univ. Twente, 7500 AE Enschede (NL), PO Box 217,
    Memoranda Informatica 89-67, Dec. 1989
Hope that helps you.


Jurgen
--

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