Hi
I am using open-mpi 1.3.4 with BLCR. Sometimes I am running into a
strange problem with ompi-checkpoint command. Even though I see that all
MPI processes (equal to np argument) are running, ompi-checkpoint
command fails at times. I have seen this failure always when the MPI
processes spawned
On May 11, 2010, at 9:18 , Gijsbert Wiesenekker wrote:
> An OpenMPI program of mine that uses MPI_Isend and MPI_Irecv crashes after
> some non-reproducible time my Fedora Linux kernel (invalid opcode), which
> makes it hard to debug (there is no trace, even with the debug kernel, and if
> I ru
FWIW, I have PGI 10.0 (but nothing more recent) and it compiles on RHEL5.4 for
me. I know there's subtle interactions between the PGI compiler suite and the
back-end compiler; there could be some set of combinations thereof between 10.1
and 10.4 that got mucked up in there somewhere...?
Anyhoo
Sorry, the kernel is 2.6.32.12, not 2.6.32.2. And I forgot to mention
the system is CentOS 5.4.
And further ... 25MB/s is after tweaking btl_sm_num_fifos=8 and
btl_sm_eager_limit=65536. Without those the rate is 9MB/s for 1MB
packets and 1.5MB/s for 10kB packets :-(
On 05/11/2010 08:19 PM, Oskar
I have a cluster with two Intel Xeon Nehalem E5520 CPU per server
(quad-core, 2.27GHz). The interconnect is 4xQDR Infiniband (Mellanox
ConnectX).
I have compiled and installed OpenMPI 1.4.2. The kernel is 2.6.32.2 and
I have compiled the kernel myself. I use gridengine 6.2u5. Openmpi was
compiled
Dick is, of course, correct. This topic has come up several times on this
list: Open MPI currently does not do this kind of check. It is therefore
possible for a sender to exhaust memory on a receiver if, for example, it
continually sends short/eager messages that the receiver consumes off the
On May 11, 2010, at 6:49 AM, Terry Dontje wrote:
> Correct the completion of an MPI_Isend request only say's the message buffer
> is no longer needed. You could use synchronous mode sends MPI_Issend which
> requests will complete when the message is being processed at the destination
> (that i
Dave Love wrote:
> I wrote:
>
>> I'll see if we can get a compiler update and report back.
>
> Installing PGI 10.5 has fixed the configure problem for me. So, for the
> archives (or FAQ?) it was a PGI bug, fixed sometime between 10.1 and
> 10.4, and apparently not present in 9.0-3. It's also pr
I wrote:
> I'll see if we can get a compiler update and report back.
Installing PGI 10.5 has fixed the configure problem for me. So, for the
archives (or FAQ?) it was a PGI bug, fixed sometime between 10.1 and
10.4, and apparently not present in 9.0-3. It's also present in 8.0-3.
Thanks to Pre
The MPI standard requires that when there is a free running task posting
isends to a task that is not keeping up on receives, the sending task will
switch to synchronous isend BEFORE the receive side runs out of memory and
fails.
There should be no need for the sender to us MPI_Issend becaus
Yes,
it's correct, but you can use number of MPI_Test fails on MPI_ISend routines
to define a limit. Remember that if you use large buffer (more than
eager_limit) send many times in a few time, maybe it's better to use
MPI_Send routine than MPI_ISend, to avoid too much buffer memory copy.
2010/5/
Gijsbert Wiesenekker wrote:
On May 11, 2010, at 9:29 , Gabriele Fatigati wrote:
Dear Gijsbert,
>Ideally I would like to check how many MPI_Isend messages have not
been processed yet, so that I can stop >sending messages if there are
'too many' waiting. Is there a way to do this?
you can
This problem is solved now. After deinstalling all other MPI distributions on
that machine openMPI 1.4.1 now works perfectly well.
Thanks very very much for your advises!!
Von: users-boun...@open-mpi.org [users-boun...@open-mpi.org] im Auftrag von
Eugen
Prentice Bisbal writes:
> Since I was successful compiled 1.4.1 with PGI 9 and 1.4.2 with PGI
> 10.4,
Thanks. The difference appears to be the compiler versions.
> I suspect the problem is local to you. Can you go through your
> environment and make sure you don't have any settings that are in
On May 11, 2010, at 9:29 , Gabriele Fatigati wrote:
> Dear Gijsbert,
>
>
> >Ideally I would like to check how many MPI_Isend messages have not been
> >processed yet, so that I can stop >sending messages if there are 'too many'
> >waiting. Is there a way to do this?
>
>
> you can check numbe
Dear Gijsbert,
>Ideally I would like to check how many MPI_Isend messages have not been
processed yet, so that I can stop >sending messages if there are 'too many'
waiting. Is there a way to do this?
you can check number of message pending simply using MPI_Test function. It
return false if the
An OpenMPI program of mine that uses MPI_Isend and MPI_Irecv crashes after some
non-reproducible time my Fedora Linux kernel (invalid opcode), which makes it
hard to debug (there is no trace, even with the debug kernel, and if I run it
under valgrind it does not crash).
My guess is that the kern
17 matches
Mail list logo
