I can't find a better forum, so I will ask my questions here:
(1) The UT memory fragmentation studies.
(2) Concurrent, real-time garbage collection.
(3) Resources.
(1) I've been reading the discussion on memory fragmentation in Mark
Johnstone's dissertation and the Wilson and Johnstone paper based on
the dissertation research
(http://www.cs.utexas.edu/users/oops/papers.html). In this work, after
removing implementation overheads, existing allocators experience as
low as 1% fragmentation. However, these same allocators with their
implementation overheads (headers, footers, etc) show more like 30%
fragmentation.
I understand that the intent is to show that the existing polices
already "solve" the fragmentation problem, and that the problem for the
future is "recognizing necessary overheads in existing implementations"
(the final sentence of Johnstone's chapter 2).
So in theory our existing polices have 1% fragmentation, but their
implementations show 30%. What is the strength of this result?
Perhaps different polices, while theoretically no better, would lead to
implementations with less overhead. And as a working programmer, I
need to be concerned with the implemented overhead of the allocator.
(2) What is the state of the art of concurrent, real-time garbage
collection? Paul Wilson's survey of a while back didn't extend itself
to a discussion of concurrent techniques. I've read the Huelsbergen
and Winterbottom paper... has there been anything since? This leads me
to my last query:
(3) What is the central resource for storage allocation and garbage
collection on the web? Does anyone track conferences, papers,
workshops, etc?
Thanks for your time.
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
<·············@my-deja.com> wrote in message
·················@nnrp1.deja.com...
> I can't find a better forum, so I will ask my questions here:
>
> (3) What is the central resource for storage allocation and garbage
> collection on the web? Does anyone track conferences, papers,
> workshops, etc?
There is a bibliography on the subject on the Harlequin web site.
I don't know if it's up to date.
Try : http://www.harlequin.com/mm/reference/bib/full.html
Marc Battyani
"Marc Battyani" <·············@csi.com> writes:
> <·············@my-deja.com> wrote in message
> ·················@nnrp1.deja.com...
> > I can't find a better forum, so I will ask my questions here:
> >
> > (3) What is the central resource for storage allocation and garbage
> > collection on the web? Does anyone track conferences, papers,
> > workshops, etc?
>
> There is a bibliography on the subject on the Harlequin web site.
> I don't know if it's up to date.
> Try : http://www.harlequin.com/mm/reference/bib/full.html
It's lagging a little, as we haven't lately had time to spare on
editing the bibliography (other parts of the Memory Management
Reference have had a little time). It's got lots of useful references
though, many with abstracts. I'm thinking of doing a minor update in
a month or so.
Richard Jones' bibliography <URL:http://www.cs.ukc.ac.uk/people/staff/
rej/gcbib/gcbib.html> is more up-to-date.
--
Pekka P. Pirinen
Adaptive Memory Management Group, Harlequin Limited
"Don't hate the media. Become the media." - Jello Biafra
On Thu, 16 Sep 1999 01:50:55 GMT, ·············@my-deja.com wrote:
> (3) What is the central resource for storage allocation and garbage
> collection on the web? Does anyone track conferences, papers,
> workshops, etc?
Henry Baker's Web site might provide useful resources:
ftp://ftp.netcom.com/pub/hb/hbaker/home.html
Paolo
--
EncyCMUCLopedia * Extensive collection of CMU Common Lisp documentation
http://cvs2.cons.org:8000/cmucl/doc/EncyCMUCLopedia/
From: Fernando Mato Mira
Subject: Re: storage allocation and concurrent gc
Date:
Message-ID: <37E2568B.24F4ECAC@iname.com>
·············@my-deja.com wrote:
> (1) I've been reading the discussion on memory fragmentation in Mark
> Johnstone's dissertation and the Wilson and Johnstone paper based on
> the dissertation research
> (http://www.cs.utexas.edu/users/oops/papers.html).
> (3) What is the central resource for storage allocation and garbage
> collection on the web? Does anyone track conferences, papers,
> workshops, etc?
Haven't you been in ftp://ftp.cs.utexas.edu/pub/garbage/ already??
Regards,
Fer
·············@my-deja.com writes:
> I can't find a better forum, so I will ask my questions here:
>
> (1) The UT memory fragmentation studies.
> (2) Concurrent, real-time garbage collection.
> (3) Resources.
The best forum for this kind thing is The Garbage Collection List
<·················@iecc.com> -- but it's perhaps not very well known.
> (1) I've been reading the discussion on memory fragmentation in Mark
> Johnstone's dissertation and the Wilson and Johnstone paper based on
> [...]
> So in theory our existing polices have 1% fragmentation, but their
> implementations show 30%. What is the strength of this result?
> Perhaps different polices, while theoretically no better, would lead to
> implementations with less overhead. And as a working programmer, I
> need to be concerned with the implemented overhead of the allocator.
I think the point of their approach is that reducing overheads is a
very different kind of problem, and probably much easier than reducing
fragmentation, so it makes sense to separate those concerns.
Your point about different overheads for different policies is valid,
but in practice implementations of the same policy vary more than
between policies. I.e., quality of the implementation counts.
My view on overheads is that different patterns of allocations need to
be separated, so that small objects or predictably changing sets can
be managed with lower overheads. If you ask on gclist, you'll no
doubt get more views, possibly including Wilson's and Johnstone's.
> (2) What is the state of the art of concurrent, real-time garbage
> collection? Paul Wilson's survey of a while back didn't extend itself
> to a discussion of concurrent techniques. I've read the Huelsbergen
> and Winterbottom paper... has there been anything since?
Not that I know of.
> (3) What is the central resource for storage allocation and garbage
> collection on the web? Does anyone track conferences, papers,
> workshops, etc?
As usual on the web, there's several sites:
The Memory Management Reference
http://www.harlequin.com/mm/reference/
the Garbage Collection Page
http://www.cs.ukc.ac.uk/people/staff/rej/gc.html
Dynamic Storage Allocation and Memory Management Information Repository
http://www.cs.colorado.edu/%7Ezorn/DSA.html
GC FAQ
http://www.iecc.com/gclist/GC-faq.html
You've found the Texas repository and somebody mentioned Baker's
interesting papers. For more, see <URL:http://www.harlequin.com/
mm/reference/links.html>.
--
Pekka P. Pirinen
editor, The Memory Management Reference
"If you don't look after knowledge, it goes away."
- Terry Pratchett, The Carpet People