Posted by
Thoralf Seewald on
Sep 23, 2007; 2:58pm
URL: http://imagej.273.s1.nabble.com/more-memory-tp3698348p3698350.html
Hi Gabriel,
I was sick at home, therefore I am answering so late - sorry.
To clean up the missunderstanding: Yes we use a selfwritten stack
opener - it is based on our experimental setup. This VIVA reader
opens a stack - the size differs between 1.7 and 2.5 GB - into the
memory: each image is loaded. This is absolutely necessary to get a
3d extraction out of this stack.
On a Linux-based workstation (32bit system, Fedora Core 5, Siemens,
4GB memory) I pushed the usable memory for ImageJ V1.36 inside the
command line up to 2670MB - enough for our application.
Now I am using a MacBook Pro (OsX 10.4, 4 GB memory) and ImageJ
1.38x, where I can change the usable memory inside of the program
(edit > options > memory & threads). But if I want to use more than
2100MB, ImageJ doesn´t start anymore.
This is my question: why can I use around 2.8GB on a Linux 32bit
distribution, but get on boarders with a Mac OsX system (also linux
based)
I used the compile & run option to start the virtual stack opener
(the same you mean), but it gaves me only wrong slides (only 48 of
315 images in wrong color).
I looked at the webpages you gave me, but it couldn´t solve the
problem. As I sayed, I guess, it is inside of the OsX Kernel, but I
have no idea how to get this problem (isn´t it science? ;) ) - I am
looking foreward.
Thank you
Thoralf.
Am 26.08.2007 um 14:47 schrieb Gabriel Landini:
> On Sunday 26 August 2007, Thoralf Seewald wrote:
>> At my application I am uploading with a selfwritten plug-in the whole
>> stack (all nearly 300 images). It is necessary to get the z-values
>> out of each image corresponding to the points of the other images.
>> Therefore I had to increase the 'memory & threads' option. But image
>> works only till 2100MB, if I want to use more, ImageJ doesn´t start
>> anymore. Inside of a 32bit-Linux distribution like Fedoracore 5 I
>> could use around 2750MB.
>
> Hi,
> I have the feeling that there is some misunderstanding of the previous
> suggestions given earlier, but I may be wrong.
>
> There is a plugin that is called "Virtual Stack Opener".
>
http://rsbweb.nih.gov/ij/plugins/virtual-opener.html> which now is included in IJ in the File>Import>Image Sequence
> (you check: Use Virtual Stack).
>
> This loads one slice at a time from disk, so you would not need to
> load the
> whole stack (and therefore would allow you to use stacks which are
> much
> larger than your memory).
>
> In a previous mail, you said that the virtual opener takes the same
> amount of
> memory than loading whole the stack. Can you verify this? This
> should not be
> the case because at any one time there is a single slice on memory.
>
> But now you say that you are using a self-written plugin that loads
> the whole
> stack, so this is not the virtual stack opener...
>
> Here is all the information on how to increase the memory available
> for IJ:
>
>
http://imagejdocu.tudor.lu/imagej-documentation-wiki/faq/how-do-i-
> increase-the-memory-in-imagej
>
http://imagejdocu.tudor.lu/imagej-documentation-wiki/faq/can-i-use-
> more-than-4-gb-of-memory
>
> I hope it helps.
> Cheers,
>
> Gabriel