Hi Ralph,
I saw another hangup with openmpi-1.8 when I used more than 4 nodes
(having 8 cores each) under managed state by Torque. Although I'm not
sure you can reproduce it with SLURM, at leaset with Torque it can be
reproduced in this way:
[mishima@manage ~]$ qsub -I -l nodes=4:ppn=8
qsub: wai
Would the respective parties please clean these up for v1.8.1?
common_verbs_find_ports.c:164: warning: 'transport_name_to_str' defined but not
used
btl_openib_component.c: In function 'btl_openib_component_init':
btl_openib_component.c:2696: warning: unused variable 'qp_index'
In file included
memheap_buddy.c:93:5: warning: "__SIZEOF_LONG__" is not defined
I tracked it down - not Torque specific, but impacts all managed environments.
Will fix
On Apr 1, 2014, at 2:23 AM, tmish...@jcity.maeda.co.jp wrote:
>
> Hi Ralph,
>
> I saw another hangup with openmpi-1.8 when I used more than 4 nodes
> (having 8 cores each) under managed state by Torque. Al
Soliciting input from the community:
WHAT: Modify PML cm component to remove unnecessary initializations,
optimizing blocking operations
WHY:Remove overhead in fast-path by allowing a "direct mode" increases
single packet latency
HOW:In PML cm, even if the request starts and ends wit
Hello everyone,
I am trying to run OpenMPI-1.6.5 on a Linux on a system based on ARM Cortex
A9. The linux system and the hardware is provided by Xilinx Inc., and for
those who may have related experiences the system is called Zynq, which is
an embedded SoC system with ARM cores and FPGA fabrics. X
This change just looks wrong - you can't split the variables on a "space"
as there is no way to know how many variables there might be, and thus how
to parse the rest of the cmd line. At best, you need a comma-delimited list.
Please fix this!
Ralph
On Tue, Apr 1, 2014 at 2:14 PM, wrote:
> Aut
not sure what you mean, could you please provide example?
comma is used often as part of the value, here is a example:
-mca base_env_list "HCOLL_BCOL=basesmuma,mlnx_p2p HCOLL_SBGP=basesmuma,p2p
HCOLL_ML_USE_KNOMIAL_ALLREDUCE=1"
On Wed, Apr 2, 2014 at 2:12 AM, Ralph Castain wrote:
> This chan
Thanks Ralph.
Tetsuya
> I tracked it down - not Torque specific, but impacts all managed
environments. Will fix
>
>
> On Apr 1, 2014, at 2:23 AM, tmish...@jcity.maeda.co.jp wrote:
>
> >
> > Hi Ralph,
> >
> > I saw another hangup with openmpi-1.8 when I used more than 4 nodes
> > (having 8 cores
If you are expecting the user to put quotes around the string, then you
better tell them that in the help message. Otherwise, they would do what I
did - simply list the envars with a space in-between, and everything fails.
Also, you need to update mpirun.1in to reflect this new option or else
nobo
Actually, the more I think about this, the more puzzled I get. We already
have a mechanism for forwarding and/or setting envars, though it applies to
the daemons, who then drop it down to the apps. Just use the "-x" option,
which will either forward the current value of the envar, or you can set
th
yes, it is expected that the "string" value should be quoted. will clarify
it in the help message and man page. Thanks for spotting.
The underlying libraries used from OMPI
(mxm,psm,hcoll,glibc,pmi2,slurm,...) all have shell environment variables
to control their behave.
It is unreasonable to exp
The "-x var=val" cannot be placed into mca params file.
On Wed, Apr 2, 2014 at 2:34 AM, Mike Dubman wrote:
> yes, it is expected that the "string" value should be quoted. will clarify
> it in the help message and man page. Thanks for spotting.
>
> The underlying libraries used from OMPI
> (mxm,p
Understood - my point, however, was that we can easily add that capability
to the mca params file.. This would seem far preferable to creating a
completely new, parallel way of setting envars.
Why not do it that way?
On Tue, Apr 1, 2014 at 4:34 PM, Mike Dubman wrote:
> The "-x var=val" cannot
mca param file treats any key=val as mca parameter only.
In order to add parser support for something that is not mca param, will
require change file syntax and it will look bad, i.e.:
mca btl = sm,self,openib
env DISPLAY = console:0
I think the current implementation is less intrusive and re-use
HmmmI'll have to ponder this one. Regardless, it doesn't belong in the
1.8 series as it isn't a bug fix.
Meanwhile, let's table it until Jeff returns next week and discuss it at
the weekly telecon. It just seems bothersome that we now have two
orthogonal ways of specifying the same thing - thi
16 matches
Mail list logo
