Re: "out of memory or dimension too large for Octave's index type"

[Francesco Potortì]

Yan Zhang:
>>I got this "out of memory or dimension too large for Octave's index 
>>type" issue when I was trying to create a matrix of size 5184 by 3904
>...
>>The only thing is that the octave program has been running for a while. 
>>But I've cleared all the variables by running 'clear all'
>>command and results of 'whos' command show no variable exists in local 
>>workspace.
>>
>>The version of octave is 4.2.1 32 bit, OS is Windows 7 64bit. I 
>>understand that switching to octave 64bit might solve this issue, but 
>>it is not a viable solution for my current situation.

Francesco Potortì:
>The size of the matrix is 160MB, nothing to do with 64 bit indexing.
>
>As far as I can remember, this may have to do with memory fragmentation
>under Windows.  I am sure Windows users can be more specific, and that
>if you look for that keyword you should find earlier explanations, but
>I suspect that rebooting the system will help at least for a while.

Yan Zhang:
>Restarting octave is not an option as the "out of memory" issue happens
>in the middle of a program. Although I didn't use any java interface
>functions in the program, I tried the "javamem" function and found no
>issues. I also checked the windows services and processes but found no
>abnormal usage of memory.

As also Philip was suggesting, have you considered rebooting your system
just before launching Octave, and avoid launching any other programs
while Octave runs?  Fragmentation is a system issue, non Octave's.

>It also occurred to me that the issue might be caused by fragmentation
>before I posted my question, but I couldn't find a solution to it other
>than switching to 64 bit octave. I'm wondering if there is a way to
>defragment memory used by octave (but not by restarting), or any other
>suggestions.

As far as I understand, memory fragmentation is a system problem on
Windows for 32-bit programs, and you can do nothing apart from the
suggestion the Philips and me already gave.  Again, if you look in this
list archives for those keywords you may be able to find some more info.

-- 
Francesco Potortì (ricercatore)        Voice:  +39.050.621.3058
ISTI - Area della ricerca CNR          Mobile: +39.348.8283.107
via G. Moruzzi 1, I-56124 Pisa         Skype:  wnlabisti
(entrance 20, 1st floor, room C71)     Web:    http://fly.isti.cnr.it