Perplexing GC Time Growth

[Y Srinivas Ramakrishna]

Hi Paul --

[I am posting my response to the *-use list where the original discussion started.]

As the scavenge time plots which i posted earlier indicated, the scavenges increase in a
linear fashion and are not affected by remarks.

As I also indicated to Jason in a late follow-up yesterday, I had screwed
up in the specification for class unloading which, pre-6u4, need two flags
to be effective: XX:+CMSClassUnloadingEnabled -XX:+CMSPermGenSweepingEnabled.
Let us see what Jason says about the effect of those two flags.

--ramki

----- Original Message -----
From: "Ciciora, Paul" <Ciciora at cboe.com>
Date: Friday, January 4, 2008 7:30 am
Subject: Perplexing GC Time Growth
To: hotspot-gc-dev at openjdk.java.net


> If you don't mind me asking, what do the times look like after a remark
> cycle?  Also have you run with PrintTenuringDistribution? It's a nice
> way to make sure you have the optimal number of slots.
>