Memory issue with Plugins, but I guess that's not the whole thing

[Rainer M. Engel]

I thought I encountered problems with some distinct plugins, running
them continuously in a loop via macro.

(Winx64, FIJI most recent/up to date), plugins:
- bUnwarpJ
- Linear Stack Alignment with SIFT
- Apply saved SIOX segmentator

I encounter them reproducible, when a bunch of files is processed via
macro loop. When I disable the plugins (comment them out) and let the
macro run itself (opening and closing files), everything is fine.

I'm not sure if this is a system issue only on Windows and I tried this one.

- open the Systems's RAM viewer (i.e. Task Manager>Performance)
- start FIJI
- open "Memory Monitor"
- load an Image Sequence as Stack (1,5GB) -> Ram is occupied
- duplicate that Stack, everything fine, ~3GB occupied
- closing both stacks doesn't release the RAM

I mean I double clicked inside of the Memory Monitor, but that does not
free the memory. Is this wanted? I remember running into a dead ends
with macros and certain plugins caused by full RAM. I don't know if this
is still a plugin issue or another..

Best regards,
Rainer


--
Rainer M. Engel, Dipl. Digital Artist
scientific|Media GbR
Pichelsdorfer Str. 143
13595 Berlin

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[dscho]

Hi Rainer,

On Fri, 2 Aug 2013, Rainer M. Engel wrote:

> - load an Image Sequence as Stack (1,5GB) -> Ram is occupied

I *could* imagine that it is an issue related to Virtual Stacks that I
completely forgot about... did you check the "as virtual stack" box?

Ciao,
Johannes

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Rainer M. Engel]

Am 02.08.2013 16:52, schrieb Johannes Schindelin:

> Hi Rainer,
>
> On Fri, 2 Aug 2013, Rainer M. Engel wrote:
>
>> - load an Image Sequence as Stack (1,5GB) -> Ram is occupied
>
> I *could* imagine that it is an issue related to Virtual Stacks that I
> completely forgot about... did you check the "as virtual stack" box?
>
> Ciao,
> Johannes
>

No, I imported all images directly. Wayne informed me about that Java
does not release the RAM as soon the images are closed but it marks it
available again to ImageJ.

Perhaps I need some more time to offer more precise information and
provide macros which than may reproduce malicious behaviour.

Maybe I simply made some tests and got nervous about the fact that
nearly all RAM is gone and I only made a test-run with 1% of the data.
As long as it runs trough, it's ok..

Thank you very much for your effort.

Best regards,
Rainer

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[dscho]

Hi Rainer,

On Fri, 2 Aug 2013, Rainer M. Engel wrote:

> Am 02.08.2013 16:52, schrieb Johannes Schindelin:
> > Hi Rainer,
> >
> > On Fri, 2 Aug 2013, Rainer M. Engel wrote:
> >
> >> - load an Image Sequence as Stack (1,5GB) -> Ram is occupied
> >
> > I *could* imagine that it is an issue related to Virtual Stacks that I
> > completely forgot about... did you check the "as virtual stack" box?
>
> No, I imported all images directly. Wayne informed me about that Java
> does not release the RAM as soon the images are closed but it marks it
> available again to ImageJ.

I usually disclose such information on the list where Google can find it,
and I'll do so now: you can force Java to release the memory by clicking
on the status bar in the main window (you might have to hit it twice).

If you are more technically inclined and want to find out what causes
memory leaks (i.e. stale references that prevent the memory from being
made available again), I recently put a very short description on the Fiji
Wiki how to analyze that efficiently with VisualVM:

        http://fiji.sc/Debugging#Debugging_memory_leaks

> Perhaps I need some more time to offer more precise information and
> provide macros which than may reproduce malicious behaviour.
>
> Maybe I simply made some tests and got nervous about the fact that
> nearly all RAM is gone and I only made a test-run with 1% of the data.
> As long as it runs trough, it's ok..
>
> Thank you very much for your effort.

Hopefully my information was useful. If not, I hope to be able to help you
resolve the problem in the future.

Ciao,
Johannes

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html