On Sep 22, 2010, at 3:46 AM, Gabriele Fatigati wrote: > i'm tuning collectives of OpenMPI 1.4.2 with OTPO. I have a little question > about BTL. This layer is involves just in point-to-point communication or > also in collectives routines? > > Because i've noted that changing some blt parameters like btl_sm_eager_limit > and doing one collective routine, performances can have very different > behaviour.
The BTLs are dumb byte transport layers -- they have no idea of any MPI semantics. But that being said, they act as the lowest layers for the MPI point-to-point functions (MPI_SEND and the like). They may *also* act as the lowest layers for all of the collective functions. Some collective algorithms call MPI_SEND to do their individual sends, for example. Others may invoke BTL send/receive functions directly. Others don't use the BTLs at all. -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
