Boost logo

Boost Users :

Subject: Re: [Boost-users] mpi: sending/receiving different types
From: Brian Budge (brian.budge_at_[hidden])
Date: 2010-12-05 12:27:18


The way we do this is send a special message saying that a broadcast
will be sent next.

  Brian

On Sun, Dec 5, 2010 at 4:44 AM, Hicham Mouline <hicham_at_[hidden]> wrote:
>> -----Original Message-----
>> From: boost-users-bounces_at_[hidden] [mailto:boost-users-
>> bounces_at_[hidden]] On Behalf Of Brian Budge
>> Sent: 03 December 2010 16:34
>> To: boost-users_at_[hidden]
>> Subject: Re: [Boost-users] mpi: sending/receiving different types
>>
>> On Thu, Dec 2, 2010 at 11:49 PM, Hicham Mouline <hicham_at_[hidden]>
>> wrote:
>> > Hi,
>> >
>> > I will be sending  messages of different kinds.
>> >
>> > The destination will not know what kind of message it is trying to
>> receive.
>> >
>> > Imagine I send many messages of the same kind. I understand the tag
>> in
>> > send/receive will be different in each case? Or can I send 2 messages
>> of the
>> > same king with the same tag? That way the receiver can use the tag to
>> > identify what it is it is receiving. If so, is the receiver supposed
>> to use
>> > the recv() function with different template argument T until the call
>> > succeeds ?
>> >
>> > Or should I always use packed_<oi>archive?
>> >
>> >
>> >
>> > rds,
>> Hi Hicham -
>>
>> You should look into probe().  I the result of a probe will give you a
>> valid MPI_Status, which contains the tag and source.  You can use this
>> tag to determine how you can construct an appropriate recv().
>>
>>   Brian
>>
>
> If from the root process I both mpi::broadcast() and mpi::send(), will a
> receiver process be able to use probe() to see if a message that was either
> sent or broadcast is available?
>
> How can I do that?
>
> rds,
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>


Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net