Call for help: memory issues with Rakudo and Parrot
will at coleda.com
Wed Mar 24 20:26:53 UTC 2010
On Wed, Mar 24, 2010 at 4:02 PM, Moritz Lenz <moritz at faui2k3.org> wrote:
> In short: Rakudo on parrot uses way too much memory. Please help us
> improve it.
> The long story:
> the Rakudo developers are currently observing huge memory usage when
> compiling Rakudo.
> In particular there's a step where the stage 1 compiler compiles all the
> setting files (concatenated into a single file). This step takes so much
> memory that it causes swapping and trashing on many machines, and
> setting a ulimit -v of 1.5G makes it fail because it can't allocate
> enough memory. So it uses at least 1.5G of virtual memory.
> I know that before the pcc_hackathon_6Mar10 branch merge a ulimit of 1G
> was sufficient (In the mean time the "core" has also grown, but only a
> small bit, not a factor of 1.5). So it seems that parrot's memory
> footprint actually has actually changed to the worse in the last weeks.
> Since Patrick is currently very distracted by real-life issues , and
> we want to get out Rakudo Star soonish, it would be very nice to get
> some support from Parrot folks.
> (In the long run we will try to split the compilation of the "core"
> files, but this is rather tricky; also I think that this is a
> manifestation of a parrot problem somewhere that rakudo currently exposes).
> Cheers, and many thanks in advance,
>  http://use.perl.org/~pmichaud/journal/40248
FYI, there is a ticket tracking this at:
I just marked it as a high priority ticket.
Will "Coke" Coleda
More information about the parrot-dev