[Insight-developers] itkLargeImageWriteReadTest and large file (Snow Leopard Update)

[Bradley Lowekamp]

Hello,

The recent release of Mac OS X 10.6 has brought to light the fact that  
despite 64-bit processes, apple's kernel was still only 32-bits. I  
thought that this may help with the issue bellow and mantis bug #9320.

Even with Leopard, the kernel does not default to 64-bits on  
workstations ( it does on Xserves ), but it can be configured to be 64- 
bit on some newer Intel workstation. I have done this, and I can  
confirm that large IO now works! The IO methods can perform > 2GB  
reads and writes, with a 64-bit kernel.

Unfortunately, the LargeImageWriteReadTests are still failing, but  
this time in a different fashion:
http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=420727

Brad


On Aug 15, 2009, at 11:49 AM, Stephen Aylward wrote:

> Hi,
>
> Sorry - just catching up on this thread.
>
> Has this been fixed in MetaImage?
>
> Thanks,
> Stephen
>
> On Fri, Jul 24, 2009 at 9:16 AM, Bradley Lowekamp<blowekamp at mail.nih.gov 
> > wrote:
>> Hello,
>> The itkLargeImageWriteReadTests were added to test to see if  
>> itk::Image can
>> deal with image greater then 2GB or 4GB. However, they fail on Mac  
>> OSX
>> despite it being 64-bits with a 8-byte long. The apple OS is  
>> limited to only
>> being able to read or write a maximum of 2GB at a time (regardless  
>> of 32 or
>> 64-bit build). I don't know if this is the only OS with this  
>> limitation. The
>> MetaIO library writes the files in one large block and therefore  
>> runs into
>> this limitation. However the NRRD file format
>> appears to read and write in smaller blocks, so by modifying the  
>> test to
>>  use NRRD it passes on my system. An alternative is to use streamed  
>> reading
>> and writing (actually I am not sure if that would work to get  
>> around this
>> for writing, as I think the ImageFileWriter writes the whole image  
>> if it is
>> available even if it was requested to stream).
>> The VisibleHumamStreamReadWriteTest is a very interesting test with  
>> regards
>> to large images. Currently this is running only on my mini's, some  
>> are
>> 64-bit, others are 32-bit. I did not realize it at the time, but  
>> this means
>> that were are processing a 16GB image on a 32-bit machine. I am not  
>> saying
>> that there are not issues related to this out of core processing,  
>> just that
>> it is very cool we are already doing this :)
>> Currently I am running an experimental on "victoria", which should  
>> show up
>> on the dashboard shortly.
>> Brad
>>
>> ========================================================
>>
>> Bradley Lowekamp
>>
>> Lockheed Martin Contractor for
>>
>> Office of High Performance Computing and Communications
>>
>> National Library of Medicine
>>
>> blowekamp at mail.nih.gov
>>
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> 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://www.itk.org/mailman/listinfo/insight-developers
>>
>>
>
>
>
> -- 
> Stephen R. Aylward, Ph.D.
> Director of Medical Imaging
> Kitware, Inc. - North Carolina Office
> http://www.kitware.com
> stephen.aylward (Skype)
> (919) 969-6990 x300

========================================================
Bradley Lowekamp
Lockheed Martin Contractor for
Office of High Performance Computing and Communications
National Library of Medicine
blowekamp at mail.nih.gov


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20090908/1b008885/attachment.htm>