SPSS "Not responding"
Locked
 
|
I’m building a cumulative file for British Social Attitudes 1983-2014, using syntax to recode positive missing values to negative, re-specify missing values and get rid of phantom labels. Each edition of the file is around 1gb. I try not to edit huge chunks of syntax, or have too many files open at once (one *.sav, perhaps two *.sps, one each for Word and Notepad). Each edition of the *.sav file is generated with File >> Save As, so I always have an earlier edition as backup. Likewise with the *.sps files, always with the same root as the *.sav file. Every now and again I get reports that SPSS is running out of memory, at other times that SPSS is “not responding”. This mostly happens when using Ctrl+C/Ctrl+V to copy/paste syntax from Notepad to SPSS or when using Ctrl+H to edit syntax within SPSS. Task Manager (TM) Applications then shows several SPSS items “not running”, and Processes shows memory between 60% and 99%, size around 2gb, leaving me no option but to close SPSS with TM: stopping a single item closes them all (and sometimes loses my SPSS settings. In the working folder several files of the following types appear: core.20160529.151929.4976.0001.dmp 1,183,354 .. heapdump.20160530.185435.8312.0004.phd 285,280 KB javacore.20160530.185435.8312.0005 1,086 KB Snap.20160530.185435.8312.0008 282 KB What exactly are these files? The full StatTransfer (syntax) text extracted from the initial cumulative *.sav file runs to 1502 pages and 273,603 words in Word, from which I extract chunks into another Word file to edit, then copy the text into Notepad and then from Notepad into SPSS. If I get a memory message from SPSS, and only then, will it help if I allocate more memory? Since SPSS is already “hung” I can only do this after closing it with TM. The FM says the default is 24576. How much should I allocate? SET WORKSPACE <number of bytes>. John F Hall (Mr) [Retired academic survey researcher] [SPSS24: Windows 7 Pro 32-bit] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop PS (for JonPeck) Thanks for the complex find/replace suggestions, but I need to keep the operations simple and clear for commentaries and undergraduate tutorials: even changing ^p^p to ^p may be too challenging! |
|
John, These are not reasonable sizes for syntax files. When you paste a block of syntax into the SE, it has to parse the entire set just as if it were going to be executed. And the result would be very difficult to check or maintain. Save logical sections in separate text files and run them with INSERT. That will give you a much more modular and maintainable structure and will not tie up large amounts of memory in the frontend. To understand what is going on, you need to understand that Statistics operates fundamentally as two processes. All the computation and data reside in the backend - the SPSS Processor. The frontend, the spss.exe process, holds the output, the SE content, and a little bit of the data for the Data Editor. Since the backend works through the syntax incrementally in sequence, it does not hold most of the syntax in memory, so it can handle an unlimited amount. SET WORKSPACE only applies to backend computational processing and only for a few procedures. Increasing that limit in general will just reduce the amount of memory available for the frontend or other processes. Don't increase this limit unless you get a message from a command that you should do so. Statistics saves updated preferences when the session terminates (necessary in order to handle multiple concurrent sessions), so if it crashes because it ran out of memory or disk or for some other reason, the preferences will not be updated, and temporary files won't be cleaned up. You can delete all those temporary files. On Mon, May 30, 2016 at 9:42 PM, John F Hall <[hidden email]> wrote:
|
Administrator
|
In reply to this post by John F Hall
John, given what you are reporting, I would try making the main syntax file a series of INSERT FILE commands that include a bunch of smaller syntax files. I don't know if it will solve all your problems, but it's what I would try.
http://www.spss-tutorials.com/spss-insert-command/ https://www.ibm.com/support/knowledgecenter/SSLVMB_21.0.0/com.ibm.spss.statistics.help/syn_insert_vs_include.htm
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
Jon, Bruce Thanks for your advice. In fact, that’s what I’ve started to do, but not with INSERT (which I’ll check out in the FM). Other colleagues have experienced similar problems when editing in Word and SPSS at the same time. I’m editing sections of syntax in Word to make sure there are no equivalent pairs of positive and negative missing values and that the negative values are the same as the positive. If not, I delete the negatives and insert a – sign before the positive value(s), export shorter sections to Notepad, copy the plain text from Notepad to SPSS and run separate jobs. I’m experimenting to see how many variables I can include each time, but am erring on the cautious side. This may sound cumbersome to you gurus, but it works, Anyway it’s raining for a couple of days and the gardening is all done, so it’s either SPSS or watching films. One or two other problems irritate me as a teacher and professional: Do I leave spelling errors in or correct them? Do I change obviously equivalent labels to a single label (eg where one has extra spaces)? Do I change labels starting with a lower case letter to UPPER case? John F Hall (Mr) [Retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com SPSS start page: www.surveyresearch.weebly.com/1-survey-analysis-workshop |
| Free forum by Nabble | Edit this page |