This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Process of computing retained sizes takes very long time and it is serial (uses only one core of my 4 core cpu), can that be performed in parallel and so faster?
Still analyzing crash dump. So far I can see lots of StreamTerm.OutputMonitor threads alive...
I am not sure if this is possible. It will be better to improve the algorithm. Note that GC can run in parallel with the computation.
please do not pay attention to the comment about crash dump - wrong IZ, sorry
I did not. I already thought that this comment was attached to wrong IZ.
From the related bugdb: "Open uploaded hprof Click Find button, to see the top retained objects Bug: Takes 15 mins" -- "I'm using my 64bit Linux box. I have not changed the memory IDE uses. I can open the same file on Windows in YourKit in 1m30s, Initializing Statics takes 16s. Click calculating exact retained size for the top 10 takes 10s. In addition to this it's 15 mins every time I want to look at the dump."
changeset: 213536:66c61c15f53c user: Tomas Hurka <thurka@netbeans.org> date: Thu Feb 23 13:42:00 2012 +0100 summary: issue #181194, improve efficiency of dirtySet cache changeset: 213535:61322416909d user: Tomas Hurka <thurka@netbeans.org> date: Thu Feb 23 13:39:51 2012 +0100 summary: issue #181194, memory map NumberList data, once they will not grow too much The above fixes improved retained size computation (including references and GC roots) on tested heap dump from 717 seconds to 232 seconds.
Integrated into 'main-golden', will be available in build *201202280400* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main-golden/rev/61322416909d User: Tomas Hurka <thurka@netbeans.org> Log: issue #181194, memory map NumberList data, once they will not grow too much
Done for 7.2.