On the chance that the WORK library is increasing, if you don't need to save files from previous iterations, delete its members at the beginning of each loop with

proc datasets kill; /* WORK is the default library */

The log will have information on deletions. RTM on proc datasets if you want more control.

On Wed, 31 Mar 98 EST Dave Simon said: >I'm using SAS 6.12 on a 486/66 under Win95, 16mb RAM to run a simulation > >program. The program loops through a few thousand simulations. Each >loop generates a data set of 1000 observations, and then does a >regression. Each individual loop only takes a few seconds. All that >gets saved from each simulation >are a few regression coeficients, but after a few hours of running, the >simulation dies. I assume the machine runs out of resources, but the >machine freezes up so I can't tell for sure. All variables are >over-written each time, so it seems that the thousandth simulation >should have as much resources as the first. >If anyone has a general suggestion for getting something like this to >run, I would >greatly appreciate the help. >Thanks, > >David Simon >dsimon@jhsph.edu