[mpich-discuss] Size of a type created with MPI_Type_create_darray

Dorier, Matthieu mdorier at anl.gov
Fri Nov 11 09:26:12 CST 2016


Yes but I have no control over what was used since it's in the trace of an already-executed program.

Thanks,

Matthieu
________________________________
From: Jeff Hammond [jeff.science at gmail.com]
Sent: Friday, November 11, 2016 9:20 AM
To: MPICH
Subject: Re: [mpich-discuss] Size of a type created with MPI_Type_create_darray

I've always found darray types to be profoundly confusing.  You might try to use MPI_Type_create_subarray instead.

Jeff

On Fri, Nov 11, 2016 at 5:36 AM, Dorier, Matthieu <mdorier at anl.gov<mailto:mdorier at anl.gov>> wrote:
Hi,

I have a trace of an MPI program in which parameters of every MPI calls have been recorded. Some communication functions are using a user-defined type created with MPI_Type_create_darray.

I would like to know what is the size of this type as a function of the parameters passed to MPI_Type_create_darray (again, I'm working on a trace, so I can't just call MPI_Type_size). We can assume, for simplicity, that the oldtype parameter is MPI_BYTE. I'm having troubles understanding, from the standard and from online documentation, how this function works and how I can compute this size. Could you help?

Thanks,

Matthieu

_______________________________________________
discuss mailing list     discuss at mpich.org<mailto:discuss at mpich.org>
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss



--
Jeff Hammond
jeff.science at gmail.com<mailto:jeff.science at gmail.com>
http://jeffhammond.github.io/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20161111/e3552d42/attachment.html>
-------------- next part --------------
_______________________________________________
discuss mailing list     discuss at mpich.org
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss


More information about the discuss mailing list