Login  Register

Graphic card reqirement for 3D reconstructions???

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options Options
Embed post
Permalink
Reply | Threaded
Open this post in threaded view
| More
Print post
Permalink

Graphic card reqirement for 3D reconstructions???

Jirka Pergner
Hi.
I couldn´t find proper answer for my question, so I hope someone will be able to answer me.
We would like to update our computer for processing z-stacks from confocal microscope to 3D reconstructions. Each Z-stack file is about 4 GB or bigger. When I try to do 3D reconstruction with interpolation on our best computer (Intel Core i3-2130 (3M Cache, 3.40 GHz) LGA 1155; 8 GB RAM and integrated graphic core in mother board) FIJI is not able to process it. Could you please tell whether FIJI is able to use graphic card core for acceleration? We will definitely buy new RAM memory and now we need to decide whether the other limitation is graphic card or processor.
Thank you in advance for your answer.
Reply | Threaded
Open this post in threaded view
| More
Print post
Permalink

Re: Graphic card reqirement for 3D reconstructions???

Dubuis, Eric D
Hello

I got some problems with stacks too when doing rendering or analysis. The main problem isn't the graphic card (I'm using a nviudia quadro fx 3450 with 256MB memory, PCI) it's the memory.
Fiji is just a distribution of image J run under Java, java3D plus associated pluggins.
ImageJ uses a lot of memory and do not optimizes its use of it, so as imageJ Fiji is RAM greedy.

When I work on stacks, (1000x1000 pic grey scale 16 bits) for analysis, ImageJ get stuck (run out of memory) at around 320 images for a computer with 4GB RAM.

A few points to consider may help:
        1) you have stacks, how many images in each ? you may have to consider limiting the amount of images.
        2) what is the size in pixel of images ? 1000x1000 ? if you run out of memory rescaling the images (to 600x600 for example) may help
        3) How are they coded, 8 bits ? 12 bits ? 16 bits ? the more bits for coding the more memory they require. The amount of memory used grow exponentially when taking into account size, bits. It's even worse if you use color images as the size is tripled.
        4) you say interpolation, this process uses lot of memory too

Check your computer first before buying more RAM, if it's a 32bit PC processor the max is 4GB ; if it's a 64bit PC processor (which I think it is) the max is 16GB
For Mac I'm not sure, but I think the max is 128GB for a 12 core.

I would recommend first to rescale the images and convert them to 8bits before interpolation.


Hope this will help

E.D.

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Jirka Pergner
Sent: 24 January 2013 11:52
To: [hidden email]
Subject: Graphic card reqirement for 3D reconstructions???

Hi.
I couldn´t find proper answer for my question, so I hope someone will be able to answer me.
We would like to update our computer for processing z-stacks from confocal microscope to 3D reconstructions. Each Z-stack file is about 4 GB or bigger.
When I try to do 3D reconstruction with interpolation on our best computer (Intel Core i3-2130 (3M Cache, 3.40 GHz) LGA 1155; 8 GB RAM and integrated graphic core in mother board) FIJI is not able to process it. Could you please tell whether FIJI is able to use graphic card core for acceleration?
We will definitely buy new RAM memory and now we need to decide whether the other limitation is graphic card or processor.
Thank you in advance for your answer.



--
View this message in context: http://imagej.1557.n6.nabble.com/Graphic-card-reqirement-for-3D-reconstructions-tp5001512.html
Sent from the ImageJ mailing list archive at Nabble.com.

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

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
| More
Print post
Permalink

Antw: Graphic card reqirement for 3D reconstructions???

reinhard rachel
In reply to this post by Jirka Pergner
Hi Jirka,

> We would like to update our computer for processing z-stacks from confocal
> microscope to 3D reconstructions. Each Z-stack file is about 4 GB or bigger.
> When I try to do 3D reconstruction with interpolation on our best computer
> (Intel Core i3-2130 (3M Cache, 3.40 GHz) LGA 1155; 8 GB RAM and integrated
> graphic core in mother board) FIJI is not able to process it. Could you
> please tell whether FIJI is able to use graphic card core for acceleration?
> We will definitely buy new RAM memory and now we need to decide whether the
> other limitation is graphic card or processor.

the question is what the limitation in your system is. You say, "Fiji is not able to process it" - why?
Do you have a 64bit processing system on your computer? this is "advantageous", or, I would say, a prerequisite. (in some way, a 32bit-operating Windows system prevents the PC from knowing what "4GB" really means; it ends somewhere at 3.3 GB or so). Some time ago, we stopped working with 32bit-operating systems - too many limitations. Second, did you "tell" Fiji that your computer in fact has 8GB RAM, i.e. how much of this memory is Fiji able to use? there is a feature in ImageJ where you have to explicitly allow the program to use a suitable amount of memory: -Edit -Options -Memory&Threads --> here, the "maximum memory" needs to be set appropriately. - Finally, but this is not essential: usually, the integrated graphic cores on mother boards are ok for the daily work with office-like programs, but not as good as dedicated graphic cards from third parties.
Here, my knowledge ends, and you might have to consult the real ImageJ and Fiji experts ...
kind regards,
Reinhard

--
Prof. Dr. Reinhard Rachel
University of Regensburg
Centre for EM / Anatomy
Faculty of Biology & Preclin. Med.
Universitaetsstrasse 31
D-93053 Regensburg - Germany
tel +49 941 943 2837, 1720
fax +49 941 943 2868
mail [hidden email]
office: VKL 3.1.29
 
next microscopy conferences:
http://www.mc2013.de/
MC2013 in Regensburg, Germany
http://www.imc2014.com/
18th IMC 2014 in Prague, Czech Rep.

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