Thanks, George
This would be an invaluable reference to me.
Best regards
Durga
On Sun, Nov 15, 2009 at 6:53 PM, George Bosilca wrote:
> Durga,
>
> You can find the answer to your questions in
> http://www.netlib.org/netlib/utk/people/JackDongarra/PAPERS/scop3.pdf.
>
> george.
>
>
> On Nov 15,
Durga,
You can find the answer to your questions in http://www.netlib.org/netlib/utk/people/JackDongarra/PAPERS/scop3.pdf
.
george.
On Nov 15, 2009, at 14:39 , Durga Choudhury wrote:
I apologize for dragging in this conversation in a different
direction, but I'd be very interested to kn
I apologize for dragging in this conversation in a different
direction, but I'd be very interested to know why the behavior with
the Playstation is different from other architectures. The PS3 box has
a single gigabit ethernet and no exapansion ports, so I'd assume it's
behavior would be no differen
By default only one socket per peer per physical network is opened.
However, Open MPI has the possibility to open multiple socket per peer
per network, based on some experiments with the Playstation (where
having multiple socket allow for more bandwidth). The MCA parameter
that allows such
No. The MPI standard only guarantees that the local buffer is
available for re-use when MPI_SEND returns. It does not guarantee
anything about the receiver or the transmission of the message. If
you need a guarantee about receiver behavior, try MPI_SSEND -- it
won't return until the rece
I'm confused about the required behavior of MPI_Send() using TCP sockets.
Does a call to MPI_Send() block until the receiving process actually
receives the message, or does MPI_Send() only block until the send operation
completes locally? In other words, does the sender actually have to wait
for a
Olá Gus
Álvaro would be a better name for me... Oh G Eterno!
;)
On Sat, 14 Nov 2009, Gus Correa wrote:
I think MPI doesn't ensure that the output will come ordered
according to process rank, as in your expected output list.
Even MPI_Barrier doesn't sync the output, I suppose.
It sy
On Nov 15, 2009, at 6:07 AM, Charles Salvia wrote:
With using TCP with OpenMPI 1.3.3, how many sockets does each
process open per peer-process? Does each process open a single
socket to connect to each peer-process, or does it use TWO sockets,
one for sending, one for receiving?
There a
On Nov 13, 2009, at 6:14 PM, Charles Salvia wrote:
I am using OMPI 1.3.3.
I didn't realize that earlier versions may behave differently. Is
there perhaps an advantage/disadvantage to using 1 socket to connect
to each process, versus 2 sockets (one for sending and one for
receiving)?
No
With using TCP with OpenMPI 1.3.3, how many sockets does each process open
per peer-process? Does each process open a single socket to connect to each
peer-process, or does it use TWO sockets, one for sending, one for
receiving?
Thanks,
10 matches
Mail list logo
