[ITK-users] Variable Matrix Access Violiotation

Mon Sep 11 14:08:11 EDT 2017

Yes, unsigned int’s  max is ~4GB, so it “could” work. But working with unsigned integers in the 2GB-3GB range is problematic and trick when subtraction and offsets need to be computed. It is a big red flag to me.

Another possibility, is that the program was just compiled in 32-bit mode. I don’t think a win32 application can allocate that much memory.

Brad

From: Francois Budin <francois.budin at kitware.com>
Date: Monday, September 11, 2017 at 1:57 PM
To: "Lowekamp, Bradley (NIH/NLM/LHC) [C]" <blowekamp at mail.nih.gov>
Cc: Fabian Torres <fabian.trobles at gmail.com>, "insight-users at itk.org" <insight-users at itk.org>
Subject: Re: [ITK-users] Variable Matrix Access Violiotation

@Brad: I looked in the source code and the types of the column and row size variables are "unsigned int" which should allow up to 2^32-1 so 4GB, isn't that correct?

On Mon, Sep 11, 2017 at 1:47 PM, Lowekamp, Bradley (NIH/NLM/LHC) [C] <blowekamp at mail.nih.gov<mailto:blowekamp at mail.nih.gov>> wrote:
Oh! I think you are experiencing integer overflow with that size of a matrix. The value Francois (correctly) computed 3,200,000,000 is ~3GB which is greater than numeric_traits<int>::max(), which for 32-bit singed integers is 2,147,483,647. The internal implementation of this class is vnl_matrix, so the integer limitation lies within that library. So you matrix is too big for this library.

When I looked at this e-mail earlier I miss computed the size, so I didn’t realized this problem.

HTH,
Brad

From: Francois Budin <francois.budin at kitware.com<mailto:francois.budin at kitware.com>>
Date: Monday, September 11, 2017 at 1:38 PM
To: Fabian Torres <fabian.trobles at gmail.com<mailto:fabian.trobles at gmail.com>>
Cc: "insight-users at itk.org<mailto:insight-users at itk.org>" <insight-users at itk.org<mailto:insight-users at itk.org>>
Subject: Re: [ITK-users] Variable Matrix Access Violiotation

Hello Fabian,
You are most likely allocating more memory than what your computer can allocates. If you compute the memory required for your matrix, it will be at least:
8 (long doubles on Visual Studio C++ are the same size as doubles) * 2*100*100*2*100*100 = 3 200 000 000 Bytes
This means that you need to have at least that much memory available.
If your computer does not have this amount of memory available, you will run into errors.
Hope this helps,
Francois

On Wed, Aug 30, 2017 at 12:24 PM, Fabian Torres <fabian.trobles at gmail.com<mailto:fabian.trobles at gmail.com>> wrote:
Hi evryone.

I´m using itk VariableMatrix to do some image processing. But I have some problems initializing my matrices.
When I use small sizes like this I do not have any problem

//matrix A
regionSize[0] = 15;
regionSze[1] = 15;
 VariableSizeMatrix<long double> matA;
matA.SetSize(2*regionSize[0]*regionSize[1],2*regionSize[0]*regionSize[1]);
matA.Fill(0.0);

But when I use bigger sizes like:

regionSize[0] = 100;
regionSze[1] = 100;

I get an Acces Violation exception
Access violation reading location 0xcdcdcdc1.

I'm using Visualstudio 2008 and ITK 4.4

Does this has something to do with memory allocation?
Is there a way a could verify if I have enough memory to allocate my matrix?

Thanks.

--
Fabián Torres Robles
Maestría en Ciencias en Ingeniería Electrónica
Ingeniería en Sistemas Electrónicos
Tel. 58081280, 0445534661338
E-mail fabian.trobles at gmail.com<mailto:fabian.trobles at gmail.com>, dae.wong at gmail.com<mailto:dae.wong at gmail.com>

_____________________________________
Powered by www.kitware.com<http://www.kitware.com>

Visit other Kitware open-source projects at
http://www.kitware.com/opensource/opensource.html

Kitware offers ITK Training Courses, for more information visit:
http://www.kitware.com/products/protraining.php

Please keep messages on-topic and check the ITK FAQ at:
http://www.itk.org/Wiki/ITK_FAQ

Follow this link to subscribe/unsubscribe:
http://public.kitware.com/mailman/listinfo/insight-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/insight-users/attachments/20170911/c8da9ed6/attachment-0001.html>