charm AT lists.siebelschool.illinois.edu
Subject: Charm++ parallel programming system
List archive
- From: Isaac Dooley <idooley AT gmail.com>
- To: Blesson Varghese <b.varghese AT student.reading.ac.uk>
- Cc: charm AT cs.uiuc.edu
- Subject: Re: [charm] Chunk migration
- Date: Tue, 17 Nov 2009 13:25:14 -0800
- List-archive: <http://lists.cs.uiuc.edu/pipermail/charm>
- List-id: CHARM parallel programming system <charm.cs.uiuc.edu>
In AMPI, each MPI rank can be migrated either by a instrumentation
based load balancer, or explicitly by the programmer by calling
MPI_Migrateto(int newpe).
"Chunk" and "Virtual Processor" are synonymous. Virtualization refers
to having many virtual processors that can be migrated among the
available processors.
Isaac
On Tue, Nov 17, 2009 at 12:59 PM, Blesson Varghese
<b.varghese AT student.reading.ac.uk>
wrote:
> Hello,
>
>
>
> I am just wondering if the following are possible in AMPI:
>
> a. If the programmer would like to have control over chunk migration
> at some particular point in the program would it be possible?
>
> b. Would the programmer be able to specify the processor onto which the
> chunk should migrate?
>
> c. Is the term processor virtualization being used with reference to
> chunk migration?
>
>
>
> Looking forward to assistance.
>
>
>
> Many thanks,
>
> Blesson.
>
> _______________________________________________
> charm mailing list
> charm AT cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/charm
>
>
- [charm] Chunk migration, Blesson Varghese, 11/17/2009
- Re: [charm] Chunk migration, Isaac Dooley, 11/17/2009
Archive powered by MHonArc 2.6.16.