Hello Reza,
I don't believe you can redefine mpiexec's environment on the fly
(though there might be some magic you could find in MPI_Finalize() and
MPI_Init() to redefine comm_world?)
If I was trying to do what you describe, I'd ask for all tasks on all
nodes for my mpiexec, pin the tasks to
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 18/09/13 10:40, Bill Wichser wrote:
> If we get through a whole month then I would say that after all the
> firmware and iDrac and CMC updates that a chassis power cycle is
> the answer.
One possibility there is that there were some firmware upda
Hello guys,
let me explain my question with an example:
Assume I want to run 24 thread on my Beowulf cluster which use MPICH2 as
the MPI library. I run these 24 threads on six servers giving each 4
threads with "mpiexec" command. Can I change this configuration during the *run
time* for example do
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 18/09/13 10:49, Douglas O'Flaherty wrote:
> "Run in C1. C0 over commits unpredictably, then throttles."
I've seen a recommendation in a public Mellanox document of using C1
not C0 when using hyperthreading/SMT, could be related to this..
- --
C
We have tested using c1 instead of c0 but no difference. We don't use
logical processors at all. When the problems happens, it doesn't matter
what you set the cores for C1/C0, they never get up to speed again
without a power cycle/reseat. We believe this to be something related
to power. Ma
