home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mitech!gjc
- From: gjc@mitech.com (George J. Carrette)
- Newsgroups: sci.math.symbolic
- Subject: Re: long symbolic runs
- Message-ID: <2529@mitech.com>
- Date: 31 Aug 92 10:12:02 GMT
- References: <JAFFER.92Aug29231912@camelot.ai.mit.edu>
- Organization: Mitech Corporation, Concord MA
- Lines: 28
-
- In article <JAFFER.92Aug29231912@camelot.ai.mit.edu>, jaffer@zurich.ai.mit.edu (Aubrey Jaffer) writes:
- > Has anyone run SYMBOLIC mathematical compatuations which, using the
- > best commands for the job, took days to complete and yet produced
- > answers small enough to be useful?
-
- Yes. I know someone who runs computations for days and days on lispmachines.
-
- > If this is a common occurence then it would seem to be a good
- > idea for symbolic math systems to dump their core images every few
- > hours so that large computations will not be lost. Do any systems
- > offer this feature?
-
- If I recall correctly, this was possible to do in Multics Macsyma.
- Where data segments were forced to be organized into specific "file-backed"
- areas of virtual memory.
-
- It was also possible to do this on a lispmachine, although you had to
- go to some extra effort to make sure that your control stack was preserved
- across a "disk-save" operation.
-
- The "control stack" management is always your biggest problem, especially
- when you mix things up in the way that SCM/SIOD or most compiled-lisp
- implementations would tend to do.
-
- -gjc
-
-
-
-