Re: [rvm-research] Question about CopyMS MMTk plan

Copy MS uses a very simple algorithm.   You can think of it as half way to being a generational collector.   Young objects are allocated separately (in a copy space) and are copied if they survive their first collection, but they are not collected independently.   The whole heap is collected at every GC, so there is no need for a write barrier.

This GC is one of the first steps in the MMTk tutorial; I highly recommend you take a look at it (although the tutorial itself is in need of updating, the general ideas still hold).

http://jikesrvm.org/MMTk+Tutorial

--Steve


________________________________
From: Baptiste Saleil [[email protected]]
Sent: Tuesday, June 11, 2013 11:24 PM
To: [email protected]
Subject: [rvm-research] Question about CopyMS MMTk plan

Hello there,

I'm currently working with JikesRVM's GC MMTk.
I see in documentation differents StopTheWorld plans for the GC but have an question about one.

Can you please give me more details about the algorithm used by the CopyMS plan[1] ?
Is it using the "Best of Both Worlds" algorithm from Sylvain Stanchina and Matthias Meyer (ISMM'07) ?
What exactly is what you call "nursery" and how do you use it ?

Thanks.
Baptiste Saleil

[1] http://jikesrvm.sourceforge.net/apidocs/latest/org/mmtk/plan/copyms/CopyMS.html

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev

_______________________________________________
Jikesrvm-researchers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers