[gambit-list] gambit 4.6.4 taking long time to compile blackhole

Álvaro Castro-Castilla alvaro.castro.castilla at gmail.com
Wed Feb 15 18:06:27 EST 2012


Maybe just activating O2 if gcc >= 4.7 (if the patch Alex was telling about
works); then O3 if more than 8gb ram available and gcc >= 4.7.




On Thu, Feb 16, 2012 at 12:00 AM, Marc Feeley <feeley at iro.umontreal.ca>wrote:

>
> On 2012-02-15, at 4:41 PM, Alex Queiroz wrote:
>
> > Hallo,
> >
> > 2012/2/15 Marc Feeley <feeley at iro.umontreal.ca>:
> >>
> >>> I've read in the git commits that now the default flag is -O2, could
> this be the reason?
> >>
> >> It could.  Please modify the configure script so that it uses -O1 and
> check if that works better.  I would be interested in the time both ways,
> and the version of your gcc.
> >>
> >
> > FWIW, I opened a bug[1] for GCC that, after fixed for 4.7.0, did let
> > me compile Gambit-C with -O3 and using `make -j 3` on machine with 8GB
> > of RAM, very quickly.
> >
> > [1] - http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51389
>
> Neat!  I'm quite surprised that gcc -O3 when using --enable-single-host
> completes at all!  I have always used -O1 and only recently (this past
> week) noticed that the gcc I was using (v4.2.1) could compile at -O2
> without taking too much RAM.  I think -O3 is pushing it.  I may have to
> revert to -O1 as a default if people are having problems with -O2.
>
> If only I could find a good heuristic for the configure script to choose
> the "right" optimization level for the user's development environment.
>
> Marc
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.iro.umontreal.ca/pipermail/gambit-list/attachments/20120216/339a4c74/attachment.htm>


More information about the Gambit-list mailing list