RFR: 8332455: Improve G1/ParallelGC tasks to not override array lengths

Kim Barrett kbarrett at openjdk.org
Tue Jul 9 10:13:34 UTC 2024


On Tue, 9 Jul 2024 09:39:26 GMT, Kim Barrett <kbarrett at openjdk.org> wrote:

> > > I have a couple of ideas for alternative mechanisms that I might explore.
> > 
> > 
> > Would you mind sharing the ideas?
> 
> I'll respond to other points separately.
> 
> Here's a (very drafty) suggestion. https://github.com/openjdk/jdk/compare/master...kimbarrett:openjdk-jdk:array-splitting?expand=1

A different idea is to have segregated queues for oops and partial array
tasks, as is already done in some cases by some collectors. That complicates
processing, stealing, and termination testing somewhat, but avoids the more
complicated allocation scheme for the state. One could even consider
segregated queues for oop* and narrowOop*, avoiding the whole task type
dispatching. I've long wondered about the tradeoffs there.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/19282#issuecomment-2217233677


More information about the hotspot-gc-dev mailing list