View Issue Details Jump to Notes ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0012123ParaViewBugpublic2011-04-26 13:082016-08-12 09:58
ReporterJames Whitby 
Assigned ToUtkarsh Ayachit 
PrioritynormalSeverityminorReproducibilityalways
StatusclosedResolutionmoved 
PlatformPCOSMS Windows XP ProfessionalOS VersionSP3
Product Version3.10 
Target VersionFixed in Version 
Summary0012123: bug importing raw data when multicore support enabled
DescriptionImporting a RAW dataset (binary) fails silently when Multi-Core support is enabled i both 3.10.0 and 3.10.1 (32 bit). i.e. load operation appears to have succeeded (no error messages) but dataset is identically zero.
Steps To ReproduceEnable or disable Multi-Core support under Edit/Settings, exit, relaunch, try and load dataset...
TagsNo tags attached.
Project
Topic Name
Type
Attached Files? file icon Shortened_ nickel_3D.mtx [^] (655,360 bytes) 2011-04-29 11:59
png file icon xy_slice_nickel_PV3.10.1_test.png [^] (17,616 bytes) 2011-04-29 12:00


png file icon PView_test_slice_mcore.png [^] (6,753 bytes) 2011-05-09 05:19

 Relationships

  Notes
(0026265)
Utkarsh Ayachit (administrator)
2011-04-27 14:54

What do you mean by "identically zero"? I could not reproduce this. I see an nice outline when I open the image data with multi-core support enabled.

Are you trying to volume render and then you don't see anything but the wireframe and that's the issue here? If so, yes, that's a limitation. Currently volume rendering is not supported in multi-core mode.

(0026274)
James Whitby (reporter)
2011-04-28 09:51

Thank you for your attempt to reproduce the bug, and for your comments.

By 'identically zero' I mean that the range of values shown in the Data Array section of the Information tab of the Object Inspector shows [0,0] instead of the expected [0.0046875, 0.276562]. Similarly the histogram view etc all indicate that the imported data array elements are all zero, instead of the expected values. Not unsurprisingly, none of the rendering options then give me an image of may data.

In the import filter list there are filters both for 'RAW' and for 'Raw (binary)' files, I am referring to the latter. (To import a 128x128x179 matrix of little endian floats, which works fine when multi-core support is not enabled).

I hadn't realised that volume rendering is not supported by the multi-core mode (even having reread the documentation and Wiki this is not clear, but thank you for letting me know), and I had indeed been trying to use this amongst other viewing options. That volume rendering is not supported does not seem however to explain the symptoms that I am observing.
(0026275)
Utkarsh Ayachit (administrator)
2011-04-28 10:06

Thanks for getting back. James, can you give me the sample dataset? feel free to email me directly if you cannot upload it to the public wiki.
(0026286)
James Whitby (reporter)
2011-04-29 12:10

I shortened my datafile to make it more portable. A bug remains, but now instead of being zero, the array reports a range of (false) values when loaded with the multi-core option enabled (when the full-size array (18 time bigger) was loaded, Paraview reported only zeroes). See the uploaded file shortened_nickel_3D.mtx for test data that should enable you to reproduce the problem (load using raw (binary) reader as little endian floats, with data extents 0..127, 0..127, 0..9). The uploaded file xy_slice_nickel_PV3.10.1_test.png is an image from Paraview of an xy slice of the dataset correctly loaded without multi-core support enabled.

(In case you're interested, this is secondary ion mass spectrometry data, in this case for nickel ions showing of part of a nickel electron microscope finder grid. We are trying to find a good way of handling medium-large 4-D datasets for visualization e.g. by colour coding different masses, and using volume rendering or segmentation followed by rendering or surface shading).
(0026449)
Utkarsh Ayachit (administrator)
2011-05-06 11:39

Alas. I still can't reproduce this. In both cases the "ImageFloat" range goes from [0, 0.171875] for me. Is that not correct?

Can you also post the incorrect image that you see when using multi-core?

Thanks
(0026460)
James Whitby (reporter)
2011-05-09 05:26

Strange. The range of data values you report is correct, but I do not obtain those values when mutlicore support is enabled (with the test dataset mentioned above, I see the range [-1.58456e+29,5.8486e+35] which is clearly false, and the resulting slice image is appended Pview_test_slice_mcore.png). To reiterate, this is observed for the 32 bit version of Paraview 3.10.1 (and 3.10.0) on Windows XP.

I also did some testing with the 64 bit version of Paraview on a Mac, using the Darwin pre-compiled version from Kitware. Version 3.10.1 worked as expected with and without multicore. Version 3.10.0 however reported the range of values to be [0, 0.015625] with multicore support on, but [0, 0.171875] without multicore support. There is definitely something odd happening when multicore support is enabled, but I have no idea what nor why you can not reproduce the bug that I see on two different operating systems. Could it be some kind of data alignment problem? What architecture/OS are you testing on?
Regards,
  James
(0026551)
Utkarsh Ayachit (administrator)
2011-05-19 09:07

I've been testing on 64 bit linux. Let me try on a Mac (Snow Leopard) and see if that changes anything.
(0037964)
Kitware Robot (administrator)
2016-08-12 09:58

Resolving issue as `moved`.

This issue tracker is no longer used. Further discussion of this issue may take place in the current ParaView Issues page linked in the banner at the top of this page.

 Issue History
Date Modified Username Field Change
2011-04-26 13:08 James Whitby New Issue
2011-04-27 14:54 Utkarsh Ayachit Note Added: 0026265
2011-04-27 14:54 Utkarsh Ayachit Assigned To => Utkarsh Ayachit
2011-04-27 14:54 Utkarsh Ayachit Status backlog => @20@
2011-04-28 09:51 James Whitby Note Added: 0026274
2011-04-28 09:51 James Whitby Status @20@ => tabled
2011-04-28 10:06 Utkarsh Ayachit Note Added: 0026275
2011-04-29 11:59 James Whitby File Added: Shortened_ nickel_3D.mtx
2011-04-29 12:00 James Whitby File Added: xy_slice_nickel_PV3.10.1_test.png
2011-04-29 12:10 James Whitby Note Added: 0026286
2011-05-06 11:39 Utkarsh Ayachit Note Added: 0026449
2011-05-06 11:39 Utkarsh Ayachit Status tabled => @20@
2011-05-09 05:19 James Whitby File Added: PView_test_slice_mcore.png
2011-05-09 05:26 James Whitby Note Added: 0026460
2011-05-09 05:26 James Whitby Status @20@ => tabled
2011-05-19 09:07 Utkarsh Ayachit Note Added: 0026551
2016-08-12 09:58 Kitware Robot Note Added: 0037964
2016-08-12 09:58 Kitware Robot Status backlog => closed
2016-08-12 09:58 Kitware Robot Resolution open => moved


Copyright © 2000 - 2018 MantisBT Team