View Issue Details [ Jump to Notes ]

[ Print ]

Project

Category

View Status

Date Submitted

Last Update

0004948

ParaView

(No Category)

public

2007-04-27 15:15

2007-05-17 15:12

Reporter

David Karelitz

Assigned To

Berk Geveci

Priority

high

Severity

major

Reproducibility

always

Status

closed

Resolution

fixed

Platform

OS Version

Product Version

Target Version

Fixed in Version

Summary

0004948: Cell to point on exodus block ids

Description

If I open a 16 file exodus dataset on 4 nodes, run D3, then run CellToPoint. The resulting cellToPoint dataset has something wrong with it. The field that I'm converting from cell data to point data resides on a single processor after running D3, and the cell field has a single value. After running CellToPoint, the point field has a long streak in it at a value not present in the initial cell field. The Cell Field has a data range of [2,2]. The resulting Point field has a data range of of [1,2].

The bug is present regardless of whether D3 is run or not, but in the use case where this comes up, D3 is used to eliminate process and file boundary issues.

Tags

No tags attached.

Project

Topic Name

Type

Attached Files

Relationships

Notes
(0007526) Ken Moreland (manager) 2007-04-30 13:46	I do not think this problem has anything to do with D3, with multiple files, or with running in parallel. I am assuming that Dave was looking at the BlockId of the g1s1 data set, however, I can replicate the same problem with disk_out_ref on one process. To replicate, do the following: 1. Load disk_out_ref. Notice that the range of BlockId is [1,1]. 2. Run through Cell To Point. BlockId range is now [0,1]. I'm guessing there is a problem with cell to point when you have integer data. I base this mostly on the fact that I cannot replicate the problem with data generated by the calculator filter, which outputs double data.

(0007616) Berk Geveci (administrator) 2007-05-10 13:28	I believe this is fixed (both HEAD and 3.0 branch). Can you please verify? > cvs commit -m"BUG: Fixed interpolation for integer types. Absolute truncation was causing problems because weights to not always exactly add up to 1. Now using rounding" vtkDataArray.cxx Checking for path: /cvsroot/ParaView3/ParaView3/VTK/Common Unrestricted user: berk /cvsroot/ParaView3/ParaView3/VTK/Common/vtkDataArray.cxx,v <-- vtkDataArray.cxx new revision: 1.75; previous revision: 1.74

Issue History
Date Modified	Username	Field	Change
2011-06-16 13:10	Zack Galbreath	Category	=> (No Category)