LAM/MPI logo

LAM/MPI General User's Mailing List Archives

  |   Home   |   Download   |   Documentation   |   FAQ   |   all just in this list

From: Robert Latham (robl_at_[hidden])
Date: 2002-09-21 15:23:59


On Fri, Sep 20, 2002 at 04:04:21PM -0400, Paul Howard wrote:
 
> My question is: how will a tree-based sending algorithm (within
> MPI_Scatter()) save any time? Isn't the limiting factor the number of
> bytes transmitted from the master, which will always be the same?

tree-based means the master sends to a few clients, then those clients
each send to a few other clients, and so on

maybe a crude ascii-art graph of a naive tree will help?

   ,4
 ,2
1 `5
 `, ,6
   3,
     7

the point is with a tree-based sending algorithm, it's *not* just the
master sending information.

==rob

-- 
Rob Latham
Mathematics and Computer Science Division    A215 0178 EA2D B059 8CDF
Argonne National Labs, IL USA                B29D F333 664A 4280 315B
_______________________________________________
This list is archived at http://www.lam-mpi.org/MailArchives/lam/