Re: [OMPI users] openmpi 1.4 and pgi 10

Yep - been fixed in the upcoming 1.4.1 (but there's a problem with the rsh launcher in the 14 nightly tarballs right now). A fix has also been submitted upstream to libtool. Thanks for noticing! -jms Sent from my PDA. No type good. - Original Message - From: users-boun...@open-mpi.o

Re: [OMPI users] Question about senting short MPI messages using RDMA

It's complicated, but the short answer is that a "short" message is defined as one where the cost of a memory copy doesn't matter. You could always use mpi_alloc_mem to get registered memory. But I don't recall offhand if we check to see if the memory is already registered for short messages (

Re: [OMPI users] openib btl slows down application

I filed a trac ticket: https://svn.open-mpi.org/trac/ompi/ticket/2153 Götz Waschk wrote: Hi everyone, I'm seeing a very strange effect with the openib btl. It seems to slow down my application even if not used at all. For my test, I am running a simple application with 8 processes on a

Re: [OMPI users] Dual quad core Opteron hangs on Bcast.

also, you can use -mca btl ^sm which, at least for me, actually gives better performance than does increasing fifos.. Matt On Jan 3, 2010, at 10:04 PM, Louis Rossi wrote: > I am having a problem with BCast hanging on a dual quad core Opteron (2382, > 2.6GHz, Quad Core, 4 x 512KB L2, 6MB L3 Cac

Re: [OMPI users] openib btl slows down application

First, a small comment:  use the MPI standard timer MPI_Wtime() instead of gettimeofday. I think the problem is that MPI_Sendrecv_replace needs a temporary buffer.  Unless the message is very small, the function uses MPI_Alloc_mem and MPI_Free_mem to allocate and free this temporary buffer.  W

[OMPI users] Question about senting short MPI messages using RDMA

Hi, guys. As I understand, to send short MPI messages, OpenMPI copies the messages to preallocated buffer and then uses RDMA. I was wondering if we can avoid the overhead of memory copy. If the user buffers for short messages are reused a lot, we can just register the user buffer instead of using

[OMPI users] openmpi 1.4 and pgi 10

Hi! config/libtool.m4 has a bug when pgi 10 is used. The lines: pgCC* | pgcpp*) # Portland Group C++ compiler case `$CC -V` in *pgCC\ [[1-5]]* | *pgcpp\ [[1-5]]*) matches pgi 10.0 but 10.0 doesn't have the --instantiation_dir flag. -- Ake Sandgren,

Re: [OMPI users] Dual quad core Opteron hangs on Bcast.

On 01/04/2010 01:23 AM, Eugene Loh wrote: 1) What about "-mca coll_sync_barrier_before 100"?  (The default may be 1000.  So, you can try various values less than 1000.  I'm suggesting 100.)  Note that broadcast has somewhat one-way traffic flow, which can have some undesirable flow control issu

Re: [OMPI users] Dual quad core Opteron hangs on Bcast.

Lenny Verkhovsky wrote: have you tried IMB benchmark with Bcast, I think the problem is in the app. Presumably not since increasing btl_sm_num_fifos cures the problem.  This appears to be trac 2043 (again)!  Note that all processes *do* enter the broadcasts.  The first broadcast

Re: [OMPI users] Dual quad core Opteron hangs on Bcast.

have you tried IMB benchmark with Bcast, I think the problem is in the app. All ranks in the communicator should enter Bcast, since you have if (rank==0) else state, not all of them enters the same flow. if (iRank == 0) { iLength = sizeof (acMessage); MPI_Bcast (&iLength, 1, MPI_INT, 0, MPI_

Re: [OMPI users] VampirTrace: time not increasing

Dear Roman, dear all, this error says, that something is very wrong with the trace collection. Do you know if this error happens in the middle of the run or at the end? To fin out, please export the following before the run "export VT_UNIFY=no" and afterwards run the unification manual

[OMPI users] openib btl slows down application

Hi everyone, I'm seeing a very strange effect with the openib btl. It seems to slow down my application even if not used at all. For my test, I am running a simple application with 8 processes on a single node, so openib should not be used at all. Still, the result with the btl enabled is much wor

Re: [OMPI users] Dual quad core Opteron hangs on Bcast.

If you're willing to try some stuff: 1) What about "-mca coll_sync_barrier_before 100"?  (The default may be 1000.  So, you can try various values less than 1000.  I'm suggesting 100.)  Note that broadcast has somewhat one-way traffic flow, which can have some undesirable flow control issues.

[OMPI users] Dual quad core Opteron hangs on Bcast.

I am having a problem with BCast hanging on a dual quad core Opteron (2382, 2.6GHz, Quad Core, 4 x 512KB L2, 6MB L3 Cache) system running FC11 with openmpi-1.4. The LD_LIBRARY_PATH and PATH variables are correctly set. I have used the FC11 rpm distribution of openmpi and built openmpi-1.4 loc

[OMPI users] Cannot get OpenMPI 1.3.3 to work with Torque 2.4.2

Hello, I followed the instructions on the FAQ page to configure and compile openmpi so that it should work with Torque. ./configure --with-tm=/usr/local --prefix=/usr/local The option --disable-server was used to configure torque on the compute nodes. I got openmpi compiled without any error messa