This is the mail archive of the guile@cygnus.com mailing list for the guile project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: garbage collectors


   Date: Tue, 18 Nov 1997 15:41:02 -0500
   From: Jim Blandy <jimb@red-bean.com>
   Mime-Version: 1.0
   Content-Type: text/plain; charset=us-ascii
   Content-Transfer-Encoding: 7bit
   Cc: Guile List <guile@cygnus.com>
   References: <lbiutsuo8r.fsf@boron.kurims.kyoto-u.ac.jp>
   X-Mailer: VM 6.34 under Emacs 19.34.1
   Sender: owner-guile@cygnus.com
   Precedence: bulk


   Aubrey Jaffer says he has used SCM for very long-running computations
   without any problems.  He does (I think) circuit simulations with it.
   Guile's GC is identical in principle to SCM's.

   In theory, yes, conservative GC can cause an unbounded amount of
   garbage to be retained.  In practice, we haven't seen this as a
   problem.  We've found storage leaks in Guile, but they have never been
   the GC's fault --- just the humans, as usual.

My co-worker and I have logged thousands of hours of circuit
simulations using SCM on Linux, MS-DOS, and Windows machines with RAM
varying from 4 to 32 Megabytes.  Currently under Linux 2.0.30 the sizes are:

  PID USER     PRI  NI  SIZE  RSS SHARE STAT  LIB %CPU %MEM   TIME COMMAND
  632 jaffer    15   0  2904 2904   376 R       0 92.1  9.4   6:31 scheme
  314 jaffer     0   0  5072 5072  2324 S       0  0.3 16.4   0:16 emacs

I rewrote the simulator this year; previously, simulations excercised
saved continuations severely.  The memory footprint does not grow.

During this three year span, we have fixed 4 GC bugs.  Two of these
were introduced with enhancements after Guile split from SCM.  Of the
other two, I think only I and one other person experienced one and the
other was never reported (a theoretical bug).

For several years my livelihood, and those of my co-workers have
depended on the results of these simulations.  SCM has contributed
much towards our success.

	    http://www-swiss.ai.mit.edu/~jaffer/Work.html

-- 
			     -=-=-=-=-=-
I am a guest and *not* a member of the MIT Artificial Intelligence Lab.
      My actions and comments do not reflect in any way on MIT.