[Cbc] multithreading

[John J Forrest]

If Cbc is configured using --enable-cbc-parallel you get a version which
can use multicore architectures.

To use then in standalone version set -threads N

I use N 1 or 2 more than number of Cpus.

A major problem with parallelism is that runs are not repeatable.  We are
considering doing a slightly less efficient deterministic version.

John Forrest


                                                                                                                                             
  From:       JAFULLER at up.com                                                                                                                
                                                                                                                                             
  To:         cbc at list.coin-or.org                                                                                                           
                                                                                                                                             
  Date:       12/05/2007 01:38 PM                                                                                                            
                                                                                                                                             
  Subject:    [Cbc] multithreading?                                                                                                         
                                                                                                                                             







I've followed a bit of the development of frameworks that implement some
sort of parallelism
(DECOMP,BCP,(ALPS),etc) and distributed computing at a grid-level (ex:
Bussiek/Ferris),
but I was curious if anyone's given thought to taking advantage of
multicore architectures
with multiple threads (rather than multiple boxes with mpi)?

I suppose at some level different osi solver objects could be resolving
given different
configurations outside the framework, but it seems like there would be
opportunities to
help performance dividing and conquering.

I suppose the same question applies less intuitively to Clp as well.






.
This message and any attachments contain information from Union Pacific
which may be confidential and/or privileged.
If you are not the intended recipient, be aware that any disclosure,
copying, distribution or use of the contents of this message is strictly
prohibited by law. If you receive this message in error, please contact the
sender immediately and delete the message and any attachments.


_______________________________________________
Cbc mailing list
Cbc at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/cbc