This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc 3.3 garbage collector defaults
- From: Geoff Keating <geoffk at geoffk dot org>
- To: Mike Stump <mstump at apple dot com>
- Cc: Benjamin Kosnik <bkoz at redhat dot com>, Andi Kleen <ak at suse dot de>, zack at codesourcery dot com, gcc at gcc dot gnu dot org, libstdc++ at gcc dot gnu dot org
- Date: 27 Jan 2003 14:51:50 -0800
- Subject: Re: gcc 3.3 garbage collector defaults
- References: <200301271904.OAA22788@makai.watson.ibm.com><129688BA-323B-11D7-96B3-003065A77310@apple.com>
Mike Stump <mstump@apple.com> writes:
> On Monday, January 27, 2003, at 11:04 AM, David Edelsohn wrote:
> >>>>>> Benjamin Kosnik writes:
> >
> > Ben> Aaaaaaaaaaghh!!!!!! Why aren't these flags documented!!!! Why
> > aren't
> > Ben> the defaults sane???
> >
> > The defaults are very sane for a 128MB 200MHz system :-).
>
> Can you please support that with timing runs? Compare with a 16MB and
> a 32MB run, and show us the numbers, thanks. Personally, I think we
> should bump the numbers to 32MB. Apple has been shipping with 8MB as
> the default for a while now, even that number I think is too small.
> If no one can come up with data to support low numbers and we can come
> up with data to support 16MB, I think it would be bad to keep the
> number too small.
I have evidence that on one machine, a two-processor machine with
128Mb of memory, that GCC starts to thrash when it uses more than
about 40Mb of memory and two copies are running (this makes sense,
2*40Mb of GCC plus about 50Mb for the system = thrashing). This sort
of thing is why I want to try an adaptive approach.
--
- Geoffrey Keating <geoffk@geoffk.org>