[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [bgl-discuss] MPI failure, simplified



At 11:28 AM 2/2/2006, Steven Pieper wrote:
As I said a week or so ago on this issue, when I had a similar (I think
it was rather similar) problem, I was told to use MPI_SSEND which
completely fixed the problem.  Is there something wrong with that fix
here?
There's nothing wrong with that fix, other than it shouldn't be necessary. The nonblocking routines in MPI are present to avoid the need for large buffering requirements and to avoid unnecessary buffer copying; an implementation should exploit this. I suspect that there is some simple oversight in the BG/L code that can be adjusted for such large receive buffers.

Bill


Steve

Steven C. Pieper, Senior Physicist:  spieper@anl.gov
Argonne National Laboratory, Physics Division, Bldg. 203, Argonne, IL 60439
Phone:  630-252-4232         Fax -6008
Secretary (Donna Nelson or Debra Morrison) -4100

- --------------------------------------------------------------------
To add or remove yourself from this mailing list, use the 'notifyme'
command on any BGL machine. To remove: notifyme -n, to add: notifyme -y.
William Gropp
http://www.mcs.anl.gov/~gropp
- --------------------------------------------------------------------
To add or remove yourself from this mailing list, use the 'notifyme'
command on any BGL machine. To remove: notifyme -n, to add: notifyme -y.