[mvapich-discuss] Error in MPI_Scatter: 'Buffers must not be aliased'

Mehmet Belgin mehmet.belgin at oit.gatech.edu
Sat Jun 21 20:25:39 EDT 2014


Hi Akshay,

This helps a lot, thanks for your clarification. This rule can be a bit 
counter intuitive for programmers, but an explicit in-place makes sense. 
I am surprised that OpenMPI silently ignores something that is clear in 
the MPI standard.

Thank you!
-Mehmet

On 06/21/2014 06:43 PM, Akshay Venkatesh wrote:
> Mehmet,
>
> This error occurred because there was an attempt to perform a memcpy 
> from source to destination with source = destination at the root. This 
> occurs when you pass overlapping memory regions to sendbuf and recvbuf 
> arguments. The MPI standard says the following about the use of 
> MPI_IN_PLACE with MPI_Scatter on an intracommunicator.
>
> """
> The "in place" option for intracommunicators is specified by passing 
> MPI_IN_PLACE as
> the value of recvbuf at the root. In such a case, recvcount and 
> recvtype are ignored, and
> root "sends" no data to itself. The scattered vector is still assumed 
> to contain n segments,
> where n is the group size; the root-th segment, which root should 
> "send to itself," is not
> moved.
> """
>
> I believe this is your intention.
>
> Also, the standard mentions:
> """
> The send buffer is ignored for all non-root processes.
> """
>
> I've attached a copy of the changes required to your program with this 
> mail. I saw the following output with the fix
>
> bash-4.1$ ../install/bin/mpirun_rsh -np 2 node001 node002 ./scatter_fixed
> [rank=0][ 0 0 ]
> [rank=1][ 1 1 ]
>
> Hope this helps.
>
> -- 
> - Akshay
>
> http://www.cse.ohio-state.edu/~akshay 
> <http://www.cse.ohio-state.edu/%7Eakshay/>
>
>
> _______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss

-- 
=========================================
Mehmet Belgin, Ph.D. (mehmet.belgin at oit.gatech.edu)
Scientific Computing Consultant | OIT - Academic and Research Technologies
Georgia Institute of Technology
258 Fourth Street, Rich Building, Room 326
Atlanta, GA  30332-0700
Office: (404) 385-0665

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20140621/903d06dd/attachment.html>


More information about the mvapich-discuss mailing list