Dear OpenMPI experts
I am confused about the role and values of priorities set to
the OpenMPI collective components, and why the shared memory (sm)
collective component disqualifies itself (apparently by default).
I describe the problem, then ask a some questions, if you have the
patience to read
Hi,
I'd like mpirun to run tasks with specific ranks on specific hosts,
but I don't want to provide any particular sockets/slots/cores.
The following example uses just one host, but generally I'll use more.
In my hostfile I just have:
root@host01 slots=4
I was playing with my rankfile to achieve
In my previous mail I said that slot=0-3 would be a solution.
Unfortunately it gives me exactly the same segfault as in case with
*:*
2010/6/9 Grzegorz Maj :
> Hi,
> I'd like mpirun to run tasks with specific ranks on specific hosts,
> but I don't want to provide any particular sockets/slots/cores
asmae.elbahlo...@mpsa.com
hello, i'm doing a tutorial on OpenFoam, but when i run in parallel by typing "mpirun -np 30 foamProMesh -parallel | tee 2>&1 log/FPM.log"On the terminal window , after 10 minutes of run, it iterate but i have at the end:..Feature refinement iteration 5--Marked for r
On Jun 9, 2010, at 7:30 AM, wrote:
> mpirun noticed that process rank 0 with PID 18900 on node linux-qv31 exited
> on signal 9 (Killed).
This error message means that some external agent (outside of Open MPI) killed
your OpenFoam process. You might want to check with your system administrator
Dear Sir/Madam,
I'm running OpenMPI 1.4.2 version. The operation system is Ubuntu 9.10
with kernel version 2.6.31-14.
$ mpirun -np 1 -cpus-per-proc 1 -bind-to-core a.out
* This works fine on single core P4 machine.*
$ mpirun -np 1 -bind-to-core a.out
*This also
On Jun 9, 2010, at 12:43 AM, Gus Correa wrote:
> btl_self_priority=0 (default value)
> btl_sm_priority=0 (default value)
These are ok. BTL selection is a combination of priority and reachability.
The self BTL can *only* reach its own process. So process A will use the
"self" BTL to talk to p
On Jun 8, 2010, at 12:33 PM, David Turner wrote:
> Please verify: if using openib BTL, the only threading model is
> MPI_THREAD_SINGLE?
Up to MPI_THREAD_SERIALIZED.
> Is there a timeline for full support of MPI_THREAD_MULTIPLE in Open MPI's
> openib BTL?
IBM has been making some good strides
I would recommend using the sequential mapper instead:
mpirun -mca rmaps seq
You can then just list your hosts in your hostfile, and we will put the ranks
sequentially on those hosts. So you get something like this
host01 <= rank0
host01 <= rank1
host02 <= rank2
host03 <= rank3
host01 <= r
Hi,
I have installed following openMPI packge on worker node from repo
openmpi-libs-1.4-4.el5.x86_64
openmpi-1.4-4.el5.x86_64
mpitests-openmpi-3.0-2.el5.x86_64
mpi-selector-1.0.2-1.el5.noarch
torque-client-2.3.6-2cri.el5.x86_64
torque-2.3.6-2cri.el5.x86_64
torque-mom-2.3.6-2cri.el5.x86_64
Havin
what does your my-script.sh looks like?
On Wed, Jun 9, 2010 at 8:17 AM, Govind wrote:
> Hi,
>
> I have installed following openMPI packge on worker node from repo
> openmpi-libs-1.4-4.el5.x86_64
> openmpi-1.4-4.el5.x86_64
> mpitests-openmpi-3.0-2.el5.x86_64
> mpi-selector-1.0.2-1.el5.noarch
>
>
#!/bin/sh
/usr/lib64/openmpi/1.4-gcc/bin/mpirun hello
On 9 June 2010 16:21, David Zhang wrote:
> what does your my-script.sh looks like?
>
> On Wed, Jun 9, 2010 at 8:17 AM, Govind wrote:
>
>> Hi,
>>
>> I have installed following openMPI packge on worker node from repo
>> openmpi-libs-1.4-4.el5
You need to include the path to "hello" unless it sits in your PATH environment!
On Jun 9, 2010, at 9:37 AM, Govind wrote:
>
> #!/bin/sh
> /usr/lib64/openmpi/1.4-gcc/bin/mpirun hello
>
>
> On 9 June 2010 16:21, David Zhang wrote:
> what does your my-script.sh looks like?
>
> On Wed, Jun 9, 2
Thanks Ralph after giving full path of hello it runs.
But it run only on one rank
Hello World! from process 0 out of 1 on node56.beowulf.cluster
there also a error
>cat my-script.sh.e43
stty: standard input: Invalid argument
On 9 June 2010 16:46, Ralph Castain wrote:
> You need to include the
Thanks a lot, it works fine for me.
But going back to my problems - is it some bug in open-mpi or I should
use "slot=*" option in some other way?
2010/6/9 Ralph Castain :
> I would recommend using the sequential mapper instead:
>
> mpirun -mca rmaps seq
>
> You can then just list your hosts in you
I would have to look at the code, but I suspect it doesn't handle "*". Could be
upgraded to do so, but that would depend on the relevant developer to do so :-)
On Jun 9, 2010, at 10:16 AM, Grzegorz Maj wrote:
> Thanks a lot, it works fine for me.
> But going back to my problems - is it some bug
I'll take a stab at this since I don't remember seeing any other
replies.
At least in the original code you sent out, you used Isend/sleep/Wait
to send messages. So, I'm guessing that part of the message is sent,
Iprobe detects that a matching message is in-coming, and then the
receiver goes
On Jun 9, 2010, at 10:00 AM, Govind Songara wrote:
> Thanks Ralph after giving full path of hello it runs.
> But it run only on one rank
> Hello World! from process 0 out of 1 on node56.beowulf.cluster
Just to check things out, I would do:
mpirun --display-allocation --display-map -np 4
T
On Jun 9, 2010, at 12:31 PM, Ralph Castain wrote:
>> >cat my-script.sh.e43
>> stty: standard input: Invalid argument
>
> Not really sure here - must be an error in the script itself.
...or an error in your shell startup files (e.g., $HOME/.bashrc).
--
Jeff Squyres
jsquy...@cisco.com
For corpor
Hi Govind
Besides what Ralph said, make sure your OpenMPI was
built with Torque ("tm") support.
Suggestion:
Do:
ompi_info --all | grep tm
It should show lines like these:
MCA ras: tm (MCA v2.0, API v2.0, Component v1.4.2)
MCA plm: tm (MCA v2.0, API v2.0, Component v1.4.2)
...
***
If your Op
On Jun 4, 2010, at 5:02 PM, Peter Thompson wrote:
> It was suggested by our CTO that if these files were compiled as to
> produce STABS debug info, rather than DWARF, then the debug info would
> be copied into the executables and shared libraries, and we would then
> be able to debug with Open MPI
On Jun 5, 2010, at 7:52 AM, Scott Atchley wrote:
> I do not think this is a supported scenario. George or Jeff can correct me,
> but when you use the MX MTL you are using the pml cm and not the pml ob1. The
> BTLs are part of ob1. When using the MX MTL, it cannot use the TCP BTL.
>
> You only s
Yes, Open MPI does not implement cancels for sends.
Cancels *could* be implemented in Open MPI, but no one has done so. There are
three reasons why:
1. It can be really, really hard to implement cancels (lots of race conditions
and corner cases involved).
2. Very, very few people ask for it
Hi Gus,
OpenMPI was not built with tm support.
The submission/execution hosts does not have any of the
PBS environment variable set
PBS_O_WORKDIR, $PBS_NODEFILE.
How i can make set it
regards
Govind
On 9 June 2010 18:45, Gus Correa wrote:
> Hi Govind
>
> Besides what Ralph said, make sure you
Hi Ralph,
The allocation looks fine, but why it show number of slots as 1. The
executions host has 4 processor, in nodes file also defined np=4.
== ALLOCATED NODES ==
Data for node: Name: node56.beowulf.clusterNum slots: 1Max slots: 0
==
Hi Govind
Govind Songara wrote:
> Hi Gus,
>
> OpenMPI was not built with tm support.
>
I suspected that.
Reading your postings, it seems to be an OpenMPI rpm from
a Linux distribution, which I would guess are generic,
and have no specific support for any resource manager like Torque.
> The subm
Hi Govind
It may work with the suggestion I sent you,
even with the OpenMPI with no Torque support that you have.
However, since you have Torque installed on your cluster,
it may be better to install OpenMPI from the source code tarball,
so as to have full Torque support built in, which is much
Dear all,
I setup a client application in a Linux machine and server application in a
Windows machine. In the server side, the function* MPI_Open_port* will
generate the following port_name:
tag=0 port=2001 description=awwase-Laptop ifname=192.168.1.4
Now I have to to pass this information into
Hi Govind
Govind Songara wrote:
> Hi Gus,
> OpenMPI was not built with tm support.
> The submission/execution hosts does not have any of the
> PBS environment variable set
> PBS_O_WORKDIR, $PBS_NODEFILE.
> How i can make set it
> regards
> Govind
>
I missed the final part of your message,
about
Thanks to both Scott and Jeff !
Next time I have a problem, I will check the README file first (Doh !).
Also we might mitigate the problem by connecting the workstation to the Myrinet
switch.
Martin
-Original Message-
From: users-boun...@open-mpi.org [mailto:users-boun...@open-mpi.org]
Thanks for the reply. I've somewhat figured it out the reason. It seems
that when a non-blocking send is posted, mpi doesn't spawn another process
that takes care the sending. The sending occurs only when the processor is
idle. Since the data I sent was immense (2) elements, there wasn't
su
32 matches
Mail list logo
