In complement to Jeff, I would add that using asynchronous messages REQUIRES that you wait (mpi_wait) for all messages at some point. Even though this might not seem obvious it is due to memory allocation "behind the scenes" which are only de-allocated upon completion through a wait statement.
2014-09-18 12:36 GMT+02:00 Jeff Squyres (jsquyres) <jsquy...@cisco.com>: > On Sep 18, 2014, at 2:43 AM, XingFENG <xingf...@cse.unsw.edu.au> wrote: > > > a. How to get more information about errors? I got errors like below. > This says that program exited abnormally in function MPI_Test(). But is > there a way to know more about the error? > > > > *** An error occurred in MPI_Test > > *** on communicator MPI_COMM_WORLD > > *** MPI_ERR_TRUNCATE: message truncated > > *** MPI_ERRORS_ARE_FATAL: your MPI job will now abort > > For the purpose of this discussion, let's take a simplification that you > are sending and receiving the same datatypes (e.g., you're sending MPI_INT > and you're receiving MPI_INT). > > This error means that you tried to receive message with too small a buffer. > > Specifically, MPI says that if you send a message that is X element long > (e.g., 20 MPI_INTs), then the matching receive must be Y elements, where > Y>=X (e.g., *at least* 20 MPI_INTs). If the receiver provides a Y where > Y<X, this is a truncation error. > > Unfortunately, Open MPI doesn't report a whole lot more information about > these kinds of errors than what you're seeing, sorry. > > > b. Are there anything to note about asynchronous communication? I use > MPI_Isend, MPI_Irecv, MPI_Test to implement asynchronous communication. My > program works well on small data sets(10K nodes graphs), but it exits > abnormally on large data set (1M nodes graphs). > > Is it failing due to truncation errors, or something else? > > -- > Jeff Squyres > jsquy...@cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > _______________________________________________ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2014/09/25344.php > -- Kind regards Nick
