Re: sgi kernels
- To: mathgroup at smc.vnet.net
- Subject: [mg15056] Re: sgi kernels
- From: gwinn at ma.ultranet.com (Joe Gwinn)
- Date: Wed, 9 Dec 1998 04:12:36 -0500
- Organization: Gwinn Instruments
- References: <74ak3o$4f1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
If you look once per hour, does the available memory slowly vanish? There appear to be some memory leaks in Mathematica, or it may be caching things, I don't really know what, and I have noticed that it's sometimes useful to store intermediate results to disk, nuke the kernel, and resume. There is an option on the kernel and on the front end to display a memory use "thermometer", so it's pretty easy to track it as the hours pass. Joe Gwinn In article <74ak3o$4f1 at smc.vnet.net>, Andrew Watson <abwatson at mail.arc.nasa.gov> wrote: > I am trying to run a very long duration, but not particularly complex or > memory-intensive program on a kernel running on a Silicon Graphics > machine with lots of memory. It always dies without explanation, > usually after several hours. > > Has anyone else experienced this behavior, and is anyone aware of any > fixes/workarounds? > > Regards, > > > Andrew B. Watson > MS 262-2 > NASA Ames Research Center > Moffett Field, CA 94035-1000 > (650) 604-5419 (650) 604-0255 fax > abwatson at mail.arc.nasa.gov http://vision.arc.nasa.gov/