[visit-users] Query over time with a parallel engine

Yann Kempf yann.kempf at fmi.fi
Wed Jul 8 09:16:22 EDT 2015


Hello,

I am using VisIt 2.8.2 in client/server mode with a parallel engine 
(linux, self-compiled at both ends). My dataset is typically something 
like 2000x2000 (three-dimensional) cells arranged in a 2D slab. There's 
one file per time step and it has been written out by a few thousand MPI 
processes, I then open the file series as one database. We have a custom 
data format and our own plugin to read in the data.

I noticed already long ago (i.e. in earlier versions, it's not really a 
new thing) that queries over time for Min and Max (at least) do not 
return correct values for the whole dataset. They only return the min 
(max) value of one process/domain but not of the whole dataset. If I use 
a serial engine I obviously get the correct values, but this is not 
really an option given the time it takes with my files. If I do a 
regular query on a single time step I get the correct value, but not 
with queries over time.

Is there a trick I missed to get the correct min/max out of the query 
over time?

Thanks in advance for your help,
Best regards,

Yann


More information about the visit-users mailing list