[SimGrid-user] Question regarding the network topology "FAT_TREE" and its topo_parameters

Christian Heinrich franz-christian.heinrich at inria.fr
Mon Oct 30 10:12:40 CET 2017


Hi Ahmed,

the examples are correct. The parsing is in
src/kernel/routing/FatTreeZone.cpp starting from line 378 and I just
verified that.

Thank you for bringing this to our attention, I will fix the
documentation.

Cheers
Christian

On Mon, 2017-10-30 at 08:50 +0000, Ahmed Hamdy Mohamed Eleliemy wrote:
> Dear users of SimGrid,
> 
> I have one concern regarding the use of the network topology
> “FAT_TREE” in the platform file.
> According to the definition of the topo_parameters in the
> documentation  http://simgrid.gforge.inria.fr/simgrid/3.16/doc/classs
> imgrid_1_1kernel_1_1routing_1_1FatTreeZone.html , they follow this
> ordering
> h ; m_h, ..., m_1 ; w_h, ..., w_1 ; p_h, ..., p_1
> This ordering is reversed comparing to the working examples that are
> distributed with the SimGrid toolkit (see
> examples/platforms/cluster_fat_tree.xml), and it is also reversed
> comparing to the order mentioned in the following paper  https://webe
> e.eedev.technion.ac.il/wp-content/uploads/2014/08/publication_574.pdf
>   which is h ; m_1, ..., m_h ; w_1, ..., w_h ; p_1, ..., p_h ( see
> section 2.B figure 1).
> 
> My question is which one is correct? Or do I misunderstand something?
> 
> Best regards
> Ahmed
> 
> Ahmed Eleliemy
> High Performance Computing Group
> Mathematics and Computer Science Department
> University of Basel
> Spiegelgasse 1
> CH-4051 Basel Switzerland
> Phone: +41 (0)768232581
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Simgrid-user mailing list
> Simgrid-user at lists.gforge.inria.fr
> https://lists.gforge.inria.fr/mailman/listinfo/simgrid-user


More information about the Simgrid-user mailing list