Jeffrey,

Versions 8.0 and 8.1 both have some problems with memory management that should be addressed somewhat in version 8.2. You might take a look at the following TS note that addresses issues with out-of- memory when fitting GEE's using GENMOD. This note seems to imply that all procedures which use CLASS statements can be adversely affected in version 8.0 and 8.1.

http://www.sas.com/service/techsup/unotes/SN/003/003057.html

The online help files state the same memory requirements for mixed under version 8.0 as my paper copy of version 6.12 documentation states. Given the TS note above, the memory requirements given in the online documentation for version 8.0/8.1 are wrong. Moreover, we cannot figure out how large a model can be fit.

If you are getting 8.2 soon, then some of these issues should be resolved. Exactly what memory model SAS will be using in 8.2 is not clarified, to my knowledge. So, I don't know if your problems will be solved by going to 8.2, but the problems should be fewer. Of course, the "buy more RAM" option can also apply. On NT, you can fit larger models (although at probably unacceptably slow pace) by setting up a pagefile space which is large enough to allow the model to run. You are then using disk space as memory. This can be painfully slow. But in the off chance that you are right on the cusp of being able to fit the entire model into memory, then increasing your pagefile space might be a viable option. Of course, shutting down applications that aren't needed is something that can get you some extra memory. Also, system reboots, if there has been memory leakage, will sometimes give you additional memory to work with.

Hope there are some thoughts here that might assist you.

Dale

>Date: Mon, 14 May 2001 12:08:10 -0400 >Reply-To: Jeffrey Fagan <jfagan@LAW.COLUMBIA.EDU> >From: Jeffrey Fagan <jfagan@LAW.COLUMBIA.EDU> >SUBJECTTo: SAS-L@LISTSERV.UGA.EDU > >Windows 95/NT , 256K RAM, SAS V8. >I'm runing Proc Mixed, Glmm800 w/ N=47,000 and I keep getting Out of >memory error. I tried on a subset ( n=1,300) of data and it ran. Is >there a way to fix this, so I can run the whole data file. I used the >default SAS settings for memory. Are there any known problems with V8 >related to memory? > >You help will be greatly appreciated. > >-Jeff

--------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@fhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------

------------------------------------------------------------ --== Sent via Deja.com ==-- http://www.deja.com/