yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx>]

> About 2000 spheres. Each clump is about 4 or 6 spheres. See attached
> files. The .py files aren't in fact the "just run it" scripts :/
> Becasue the Triaxial test from Filegenerator works for me best. When
> run from triax-basic.py - something is different. And I didn't bother
> to find out what. So I just used bodies generated from
> triax-basic-with-clups.py and pasted them into .xml file generated
> from FileGenerator. So the .xml files are "just run it". But remember
> to update to HEAD.
>
>   
Thanks for explanation!

> Be careful to run with single job. With multiple cores results are
> different on each run. In fact the picture that I've sentbefore  was
> with jobs=3 and I had different result on another try. And now I did
> the calculation again, and it is the same everytime, with jobs=1. See
> the new attached picture. Two plots are overlapping each other.
>
>   wget http://geo.hmg.inpg.fr/~janek/clumps.xml
>   wget http://geo.hmg.inpg.fr/~janek/spheres.xml
>
>   
Yes, I experimented this multithread variability in different situations.
Still not sure if it is a rounding error propagation (that's the 
official explanation - different roundings for ordering of operations), 
or if there are really errors.
The differences look preety big for something supposed to be just 
rounding errors, but I was initiating simulations with loose clouds. 
Maybe that is the reason.

Bruno

--
_______________
Bruno Chareyre
Associate Professor
Grenoble INP
Lab. 3SR
BP 53 - 38041, Grenoble cedex 9 - France
Tél : 33 4 56 52 86 21
Fax : 33 4 76 82 70 43
________________