Ok thank you for the github links. I missed those. But the question remains if
the old functionality in ./orte/orted/orted_main.c is still accessible, by some
configuration parameters.
I will also push the github questions.
Hello Mateo,
What version of openmpi are you running?
Also, the OFED-4.17-1 release notes do not claim support for CentOS 7.7. It
supports CentsOS 7.6.
Apologies if you have already tried CentOS 7.6.
We have been able to run openmpi (earlier this month):
OS: CentOS 7.6
mpir
I rolled everything back to stock centos 7.7 installing OFED via:
yum groupinstall @infiniband
yum install rdma-core-devel infiniband-diags-devel
which does not install the ofed_info command, or at least I could
not find it (do you know where it is?).
openmpi is version 3.1.4
the fw v
Have you tried using the UCX PML?
The UCX PML is Mellanox's preferred Open MPI mechanism (instead of using the
openib BTL).
> On Nov 13, 2019, at 9:35 AM, Matteo Guglielmi via users
> wrote:
>
> I rolled everything back to stock centos 7.7 installing OFED via:
>
>
>
>
> yum groupinstall
I'm not using Mellanox OFED because the card
is a Marvell OCP type 25Gb/s 2-port LAN Card.
Kernel drivers used are:
qede + qedr
Beside that,
I did a quick test on two nodes installing
CentSO 7.6 and:
ofed_info -s
OFED-4.17-1:
and now the error message is different:
---
I cannot find a firmware for my card:
https://www.gigabyte.com/za/Accessory/CLNOQ42-rev-10#ov
Do you have the same model?
I found this zip file of the web:
Linux_FWupg_41xxx_2.10.78.zip
which contains a firmware upgrade tool and a firmware
version 8.50.83, but when I run it I get this err
Dear all,
I have a little piece of code shown below that initializes a
multidimensional Fortran array and performs:
- a non-blocking MPI_Iallreduce immediately followed by an MPI_Wait
- a blocking MPI_Allreduce
After both calls, it displays a few elements of the input and output
buffers.
In
Camille,
your program is only valid with a MPI library that features
|MPI_SUBARRAYS_SUPPORTED|
and this is not (yet) the case in Open MPI.
A possible fix is to use an intermediate contiguous buffer
integer, allocatable, dimension(:,:,:,:) :: tmp
allocate( tmp(N,N,N,N) )
and then repla
Dear Gilles,
Thank you very much for your clear answer.
Camille
On 11/13/19 5:40 PM, Gilles Gouaillardet via users wrote:
Camille,
your program is only valid with a MPI library that features
|MPI_SUBARRAYS_SUPPORTED|
and this is not (yet) the case in Open MPI.
A possible fix is to use a
Difficult to know what to say here. I have no idea what your program does after
validating the license. Does it execute some kind of MPI collective operation?
Does only one proc validate the license and all others just use it?
All I can tell from your output is that the procs all launched okay.
Agree with Ralph. Your next step is to try what is suggested in the FAQ: run
hello_c and ring_c.
They are in the examples/ directory in the source tarball. Once Open MPI is
installed (and things like "mpicc" can be found in your $PATH), you can just cd
in there and run "make" to build them.
11 matches
Mail list logo
