home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ogicse!uwm.edu!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!ds2.uh.cwru.edu!muzic
- From: muzic@ds2.uh.cwru.edu (Ray Muzic)
- Newsgroups: sci.math.num-analysis
- Subject: MATLAB memory problems
- Message-ID: <muzic.714762125@ds2.uh.cwru.edu>
- Date: 25 Aug 92 17:02:05 GMT
- Article-I.D.: ds2.muzic.714762125
- Sender: news@usenet.ins.cwru.edu
- Organization: Case Western Reserve University, Cleveland, Ohio, (USA)
- Lines: 34
- Nntp-Posting-Host: ds2.uh.cwru.edu
-
-
- We have noticed a very undesirable behavior of MATLAB with respect to
- memory managment. This occurs on both Ultrix MATLAB (Ultrix 4.2, MATLAB
- 3.5f) and VAX/VMS MATLAB (VMS 5.4-3, MATLAB 3.5f).
-
- AS we understand it, for each MATLAB assignment and for every expression or
- subexpression, MATLAB allocates memory in which to store the results.
- MATLAB, however, does not return the memory to the system until one exits
- MATLAB. This in itself is not all that bad as it could conceiveably be done
- in the interest of efficiency. What is unforgivable, however, is that MATLAB
- does not re-use this memory!
-
- To see this for yourself, define a 512 by 512 matrix and save it to disk.
- Then load the matrix into memory several times. Between successive loads,
- show memory (Ultrix: 'ps -au' or 'vmstat', VMS: 'show memory') allocation
- and you will probabaly see it increase with each successive load.
-
- Does anyone else observe this behavior on other platforms?
-
-
- Ray Muzic
- Ed Ellert
- Don Johann
- Gary Muswick
-
-
-
-
- --
- * Ray Muzic, Jr., Ph.D. | "I try not to think, it slows the progress." - rfm2 *
- * muzic@ds2.uh.cwru.edu | Biomedical Engr, CWRU & Image Analysis Ctr, UH __o *
- * Voice: (216)-844-3543 | Div Nuclear Radiology, Univ Hosp of Cleveland -\<, *
- * Fax: (216)-844-3106 | 2074 Abington Rd, Cleveland, OH, 44106, USA....O/ O *
-