Thanks all for your answers, I've added some details about the tests I have
run. See below.
Ralph Castain wrote:
Not precisely correct. It depends on the environment.
If there is a resource manager allocating nodes, or you provide a hostfile
that specifies the number of slots on the nodes,
Thanks a lot, that looks right! Looks like some reading to do..
Do you know if in the OpenMPI implementation the MPI_T-interfaced MCA
settings are thread-local, or rank-local?
Thanks!
Marcin
On 09/15/2015 07:58 PM, Nathan Hjelm wrote:
You can use MPI_T to set any MCA variable before MPI_In
I have run into a freeze / potential bug when using MPI_Comm_accept in a
simple client / server implementation. I have attached two simplest
programs I could produce:
1. mpi-receiver.c opens a port using MPI_Open_port, saves the port
name to a file
2. mpi-receiver enters infinite loop and
Can you check with an MPI_Finalize in the receiver ?
Jalel
Le 16/09/2015 16:06, marcin.krotkiewski a écrit :
I have run into a freeze / potential bug when using MPI_Comm_accept in
a simple client / server implementation. I have attached two simplest
programs I could produce:
1. mpi-receiver.
But where would I put it? If I put it in the while(1), then
MPI_Comm_Accept cannot be called for the second time. If I put it
outside of the loop it will never be called.
On 09/16/2015 04:18 PM, Jalel Chergui wrote:
Can you check with an MPI_Finalize in the receiver ?
Jalel
Le 16/09/2015 16:
Right, anyway Finalize is necessary at the end of the receiver. The
other issue is Barrier which is invoked probably when the sender has
exited hence changing the size of intercom. Can you comment that line in
both files ?
Jalel
Le 16/09/2015 16:22, Marcin Krotkiewski a écrit :
But where woul
I have removed the MPI_Barrier, to no avail. Same thing happens. Adding
verbosity, before the receiver hangs I get the following message
[node2:03928] mca: bml: Using openib btl to [[12620,1],0] on node node3
So It is somewhere in the openib btl module
Marcin
On 09/16/2015 04:34 PM, Jalel
As I said, if you don’t provide an explicit slot count in your hostfile, we
default to allowing oversubscription. We don’t have OAR integration in OMPI,
and so mpirun isn’t recognizing that you are running under a resource manager -
it thinks this is just being controlled by a hostfile.
If you
On Sep 16, 2015, at 8:22 AM, marcin.krotkiewski
wrote:
>
> Thanks a lot, that looks right! Looks like some reading to do..
>
> Do you know if in the OpenMPI implementation the MPI_T-interfaced MCA
> settings are thread-local, or rank-local?
By "rank local", I assume you mean "process local"
With openmpi-1.7.5, the sender segfaults.
Sorry, I cannot see the problem in the codes. Perhaps people out there
may help.
Jalel
Le 16/09/2015 16:40, marcin.krotkiewski a écrit :
I have removed the MPI_Barrier, to no avail. Same thing happens.
Adding verbosity, before the receiver hangs I
Hi
We are currently using openmpi 1.8.5 and gcc 4.4.7, we would like to change
the associated gcc to gcc 4.1.2 for our openmpi 1.8.5 installation.
Is this possible. If so how can it be done.
Thanks
Sudhir Kumar
CLASSIFICATION: UNCLASSIFIED
Have you attempted using 2 cores per process? I have noticed that
MPI_Comm_accept sometimes behaves strangely on single core variations.
I have a program that makes use of Comm_accept/connect and I also call
MPI_Comm_merge. So, you may want to look into that call as
Have you tried just rebuilding OMPI after setting gcc 4.1.2 at the front of
your PATH and LD_LIBRARY_PATH?
> On Sep 16, 2015, at 8:58 AM, Kumar, Sudhir wrote:
>
> Hi
> We are currently using openmpi 1.8.5 and gcc 4.4.7, we would like to change
> the associated gcc to gcc 4.1.2 for our openmpi
Haven't tried that. Will try that approach.
Thanks
Sudhir Kumar
-Original Message-
From: users [mailto:users-boun...@open-mpi.org] On Behalf Of Ralph Castain
Sent: Wednesday, September 16, 2015 11:05 AM
To: Open MPI Users
Subject: Re: [OMPI users] open mpi gcc
Have you tried just rebuil
Thank you all for your replies.
I have now tested the code with various setups and versions. First of
all, the tcp btl seems to work fine (I had patience to check ~10 runs),
the openib is the problem. I have also compiled using the Intel compiler
and the story is the same as when using gcc.
I found the thread below from May. I’m setting up a new cluster and using
openmpi 1.10. I have a gnu build and an Intel. Neither has libmpi.so.1. I
created a symlink and it’s working. My question is if I should try to
rebuild LAPACK, and is it wise to be adding that link? For me it’s just
burn-in a
It's worth a mention that I made several attempts to add a NASA email
address to this list. Nothing ever happened. Nothing bounced. I never got a
validation email. Nothing appeared in our spam filter. I emailed
webmaster@openmpi and got no reply. There seems to be no contact conduit
but these lists
Sorry for the trouble. FWIW, there actually is a real, live sysadmin at
Indiana University who actually receives the webmaster emails; he usually
forwards such emails to me.
Ping me off-list and we can dig into why your NASA email address didn't work
(i.e., I can ask the IU sysadmins to look i
Jeff will undoubtedly start typing before he reads my response, so I'll
spare you from reading all the ugly details twice :-)
There was an unintentional ABI break in the 1.8 series that necessitated a
version numbering change to libmpi. It involves the code that handles the
connection between a pr
Hmm, I'm pretty sure my xhpl binary is/was dynamic linked. Here is my env:
LD_LIBRARY_PATH=/share/apps/openmpi-1.10-intel-x86_64/lib:/share/apps/Intel/composer_xe_2015.2.164/compiler/lib/intel64:/share/apps/Intel/composer_xe_2015.2.164/mkl/lib/intel64:/opt/python/lib
The binary fails without a sy
Looks like you are trying to link it against the 1.10 series? You could
probably get away with the symlink, but unless there is some reason to
avoid it, I'd just recompile to be safe.
On Wed, Sep 16, 2015 at 7:36 PM, Mark Moorcroft wrote:
>
>
> Hmm, I'm pretty sure my xhpl binary is/was dynamic
21 matches
Mail list logo
