Are there any resources available that describe memory leaks and Igor specifically? Relatively green programmer here, and most everything I find on the net is about C++ or Java, just not sure how applicable it is. I have written some code that does some fairly heavy iterations, and am starting to wonder if I need to somehow free up memory by redeleting variables, or just how Igor handles that..
whilst we are on the topic of memory/fragmentation, I've always found it interesting that virtual machines running in programs such as Parallels, Virtual Box, etc, can claim their big slices of memory without too much difficulty. I'm not sure how fragmented this is though.
Given that larger and larger arrays/waves are becoming commonplace I am betting that a large amount of those datasets are sparse, i.e. filled with zeros. That is certainly the case for a lot of my datasets. I am wondering if IGOR is contemplating a move to handle sparsity in the future.
"Leaks" are (essentially) bugs. We try to minimize these.
You should be able to reclaim most memory by deleting waves from your experiment using KillWaves (or the Kill Waves dialog in the Data menu).
If the waves are saved to disk then you don't lose anything, but temporary waves and waves that can be re-created from the original data are no loss.
The more difficult problem is "fragmentation".
Take a look at the "Memory Management" help topic.
Type the following into Igor's command line:
--Jim Prouty
Software Engineer, WaveMetrics, Inc.
August 8, 2012 at 03:01 pm - Permalink
Given that larger and larger arrays/waves are becoming commonplace I am betting that a large amount of those datasets are sparse, i.e. filled with zeros. That is certainly the case for a lot of my datasets. I am wondering if IGOR is contemplating a move to handle sparsity in the future.
August 8, 2012 at 04:23 pm - Permalink