[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: GC: cons sweeping
From: |
Dmitry Antipov |
Subject: |
Re: GC: cons sweeping |
Date: |
Tue, 03 Jul 2007 14:48:49 +0400 |
User-agent: |
Thunderbird 1.5.0.7 (X11/20061008) |
Stefan Monnier wrote:
So it seems to be able to speed up sweeping of cons cells by anywhere between
20% and 50% (up to a factor of 2 speed up). Now, what fraction of time is
spent sweeping cons-cells?
In general, typical GC spent 80% of time in marking and 20% of time in sweeping
(so even a tiny optimization of the first phase might be more important than a
much more solid enhancement of a second - but this doesn't mean that we should
not try to optimize both :-), and cons sweeping time is typically 2-5 % of the
whole GC time. So, since an average GC takes 20-30 ms on a 3GHz x86 CPU, I agree
that this optimization is likely not to be seen by an average user on an average
modern hardware.
Dmitry