bruce at hoult.org
Mon Jun 13 05:30:29 PDT 2011
On Mon, Jun 13, 2011 at 9:13 PM, Florian Weimer <fw at deneb.enyo.de> wrote:
> * Glauco Masotti:
>> What about suggesting those guys in the R&D to use the GC?
> There once was a development branch which used garbage collection. It
> did not offer adequate performance. I believe it is mentioned in this
> video: <http://vimeo.com/12614626>
I didn't have time to watch the whole thing but I wasn't aware of
proper GC having been tested in Mozilla.
What I know *did* happen (because he briefed me on what he'd done) is
that Graydon Hoare tried using bdwgc as a malloc/free implementation
instead of the standard malloc/free. jemalloc was also tested and they
ended up going with that because it caused less fragmentation than
However they did extensive turning with jemalloc to achieve that,
while bdwgc was used out of the box, with no tuning at all. From
memory, jemalloc had special case strategies for extremely small
objects, and also for very large ones.
More information about the Gc