[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openjdk-hotspot-gc-dev
Subject:    Question about MemoryService class.
From:       Y.S.Ramakrishna () Sun ! COM (Y Srinivas Ramakrishna)
Date:       2009-02-08 0:11:17
Message-ID: 7240ee4e6734.498db2a5 () sun ! com
[Download RAW message or body]


[I am cross-posting on to the hotspot-gc-dev list, since
that's likely where this discussion belongs; i'll bcc
the hotspot-dev list on this response, so interested folk can
follow the discussion on the former list.]

Sorry about the cryptic note.
G1 is short-hand (within HotSpot circles) of
the "Garbage-First Garbage Collector" which was
checked into the open jdk repos a few months ago.
For related background, see:-

http://portal.acm.org/citation.cfm?id=1029879

@inproceedings{1029879,
 author = {Detlefs,, David and Flood,, Christine and Heller,, Steve and Printezis,, \
Tony},  title = {Garbage-first garbage collection},
 booktitle = {ISMM '04: Proceedings of the 4th international symposium on Memory \
management},  year = {2004},
 isbn = {1-58113-945-4},
 pages = {37--48},
 location = {Vancouver, BC, Canada},
 doi = {http://doi.acm.org/10.1145/1029873.1029879},
 publisher = {ACM},
 address = {New York, NY, USA},
 }


I am thinking that the latitude allowed by the basic
collection set mechanism and the ability to collect an
arbitrary collection set at specific points,
should, in principle, allow a range of policies
for the choice of specific collection sets at specific
times, allowing one to simulate, in effect, an arbitrary
n-generation system by means of segregating objects into
collection sets by age. Much handwaving here; and,
as usual, the devil's in the details... so consider this
loud thinking about possible approaches, rather than
concrete direction.

over and out.
-- ramki

----- Original Message -----
From: "Colin(Du Li)" <dawn2004 at gmail.com>
Date: Saturday, February 7, 2009 1:25 pm
Subject: Re: Question about MemoryService class.
To: hotspot-dev at openjdk.java.net


> Hi, ramki,
> 
> Thanks you for your reply.
> Does G1 mean the "generation 1" in generational GC?
> Why do you think it might be moot? I wonder if you can give more
> explanations.
> Thanks.
> 
> Colin.
> 
> y.s.ramakrishna wrote:
> > 
> > 
> > of course, G1 and the possibilities/flexibility it allows
> > probably makes something like this somewhat moot (at least as
> > far as my intuition goes here).
> > 
> > -- ramki
> > 
> > ----- Original Message -----
> > From: Y Srinivas Ramakrishna <Y.S.Ramakrishna at Sun.COM>
> > Date: Saturday, February 7, 2009 1:09 pm
> > Subject: Re: Question about MemoryService class.
> > To: "Colin(Du Li)" <dawn2004 at gmail.com>
> > Cc: hotspot-dev at openjdk.java.net
> > 
> > 
> > > Hi Colin --
> > > 
> > > If you have a use case where 3 or more generations will
> > > provide distinct advantages, please feel free to
> > > propose an OpenJDK project for this and we'll be
> > > happy to help you navigate around in the relevant parts of
> > > the source code so you can get this done. The basic
> > > functionality will likely be quite straightforward,
> > > if tedious in some of the details, and in limiting
> > > the performance impact upon the basic 2 generation system.
> > > 
> > > -- ramki
> > > 
> > > ----- Original Message -----
> > > From: "Colin(Du Li)" <dawn2004 at gmail.com>
> > > Date: Saturday, February 7, 2009 11:08 am
> > > Subject: Re: Question about MemoryService class.
> > > To: hotspot-dev at openjdk.java.net
> > > 
> > > 
> > > > Thanks a lot?
> > > > I have fixed it.
> > > > To constrain the two-generation collector makes the system less
> > > extensible.
> > > > I wonder this part will be improve the the future.
> > > > 
> > > > 
> > > > mandy.chung wrote:
> > > > > 
> > > > > Colin(Du Li) wrote:
> > > > > 
> > > > > > Hi, guys.
> > > > > > 
> > > > > > Could you give me some information about MemoryService in 
> hotspot. 
> > > It
> > > > seems
> > > > > > like a tool to monitor the memory usage, right? But it only 
> > > support 
> > > > 2
> > > > > > generations. I wanna divide the heap into three generations, so 
> it 
> > > gives
> > > > me
> > > > > > error information. Can I turn it off? Will it cause problems if 
> I 
> > > 
> > > > turn it
> > > > > > off. How can I turn it off without side effect?
> > > > > > 
> > > > > > 
> > > > > The memoryService and memory pool implementation is the VM 
> support 
> > > 
> > > > for 
> > > > > the java.lang.management API.  
> > > > > 
> > > > > You need to change the current implementation to support 3 
> > > > generations.  
> > > > > The memory pool/service implementation is not limited to 2
> > > generations.
> > > > > 
> > > > > Mandy
> > > > > 
> > > > > > Thanks a lot!
> > > > > > 
> > > > > > Colin(Du Li)
> > > > > > 
> > > > > > 
> > > > > 
> > > > > 
> > > > > 
> > > > 
> > > > -- 
> > > > View this message in context:
> > > http://www.nabble.com/Question-about-MemoryService-class.-tp21880830p21891770.html
> > > 
> > > > Sent from the OpenJDK Hotspot Virtual Machine mailing list 
> archive 
> > > at 
> > > > Nabble.com.
> > > > 
> > 
> > 
> 
> -- 
> View this message in context: \
> http://www.nabble.com/Question-about-MemoryService-class.-tp21880830p21892968.html \
> Sent from the OpenJDK Hotspot Virtual Machine mailing list archive at  Nabble.com.
> 


[prev in list] [next in list] [prev in thread] [next in thread]