wellnhofer at aevum.de
Mon Sep 12 20:37:06 UTC 2011
On 12/09/11 22:04, Andrew Whitworth wrote:
> Okay, here's what I'm going to do. I'm going to rip out the ms2 GC
> core, rip out the command-line --gc option, and flatten the GC API
> calls down in a branch. That way we can see if there really is any
> kind of performance gains to be had and if so, how much. If we like
> what we see, we'll set up the necessary configure-time architecture to
> make the inf GC a selectable option at configure time.
Thinking more about it, inf GC will only work as a runtime option for
programs that don't consume too much memory. Compiling parrot or Rakudo
with inf GC only will certainly fail. But it should be possible to get
similar behavior to inf GC by tuning the GC memory thresholds. So ignore
my concerns about inf GC.
More information about the parrot-dev