From: LARRY J EVANS (jcampbell3_at_[hidden])
Date: 2001-08-30 10:34:50
----- Original Message -----
From: Nick Barnes <Nick.Barnes_at_[hidden]>
To: David F. Bacon <dfb_at_[hidden]>
Cc: <David.Chung_at_[hidden]>; Garbage Collection Mailing List
<gclist_at_[hidden]>
Sent: Thursday, August 30, 2001 5:39 AM
Subject: Re: [gclist] zillion objects and gc
> At 2001-08-29 16:42:16+0000, "David F. Bacon" writes:
>
> > you might be interested in some work that we did at IBM watson in the
> > jalapeno JVM on concurrent cycle collection that was recently published
> > in PLDI and ECOOP; see
> > http://www.research.ibm.com/people/d/dfb/papers.html for copies.
>
http://groups.yahoo.com/group/boost/files/GarbageCollectByTmplSpecializa/
contains
a c++ implementation of the uniprocessor algorithm referenced in
Bacon01Concurrent.pdf
file found on the above papers.html page. The version currently on boost
(in the gc_sel_ssp.zip
file in file lib/gc_sel_ssp/special/subjtop_lazy_crc.cpp )
does not implement to improvement that's mentioned below fig 1 in the pdf
file; however,
I THINK further development after that upload does have the improvements.
I'm away from
my computer
for about 12 days, but then I can check to make sure and upload whatever
I've got to boost. Since boost is
currently working on a threads library, maybe soon some boost threads
experts could
help implement the concurrent version mentioned in the Bacon pdf file.