Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

Just for clarity: does the BIOS on the LGA2011 system have HT enabled? > On Apr 8, 2015, at 10:55 AM, Lane, William wrote: > > Ralph, > > I added one of the newer LGA2011 nodes to my hostfile and > re-ran the benchmark successfully and saw some strange results WRT the > binding directives. Why

Re: [OMPI users] parsability of ompi_info --parsable output

Received from Ralph Castain on Wed, Apr 08, 2015 at 12:23:28PM EDT: > Sounds reasonable - I don't have time to work thru it right now, but we can > look at it once Jeff returns as he wrote all that stuff and might see where to > make the changes more readily than me. Made a note of the suggestion

Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

Ralph, I added one of the newer LGA2011 nodes to my hostfile and re-ran the benchmark successfully and saw some strange results WRT the binding directives. Why are hyperthreading cores being used on the LGA2011 system but not any of other systems which are mostly hyperthreaded Westmeres)? Isn't th

Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

> On Apr 8, 2015, at 9:29 AM, Lane, William wrote: > > Ralph, > > Thanks for YOUR help, I never > would've managed to get the LAPACK > benchmark running on more than one > node in our cluster without your help. > > Ralph, is hyperthreading more of a curse > than an advantage for HPC applicati

Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

> On Apr 8, 2015, at 10:20 AM, Lane, William wrote: > > Ralph, > > I just wanted to add that roughly a year ago I was fighting w/these > same issues, but was re-tasked to more pressing issues and had to > abandon looking into these OpenMPI 1.8.2 issues on our CentOS 6.3 > cluster. > > In any c

Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

Ralph, I just wanted to add that roughly a year ago I was fighting w/these same issues, but was re-tasked to more pressing issues and had to abandon looking into these OpenMPI 1.8.2 issues on our CentOS 6.3 cluster. In any case, in digging around I found you had the following recommendation back

Re: [OMPI users] OpenMPI 1.8.2 problems on CentOS 6.3

Ralph, Thanks for YOUR help, I never would've managed to get the LAPACK benchmark running on more than one node in our cluster without your help. Ralph, is hyperthreading more of a curse than an advantage for HPC applications? I'm going to go through all the OpenMPI articles on hyperthreading a

Re: [OMPI users] parsability of ompi_info --parsable output

Sounds reasonable - I don’t have time to work thru it right now, but we can look at it once Jeff returns as he wrote all that stuff and might see where to make the changes more readily than me. > On Apr 8, 2015, at 8:43 AM, Lev Givon wrote: > > Received from Ralph Castain on Wed, Apr 08, 2015

Re: [OMPI users] parsability of ompi_info --parsable output

Received from Ralph Castain on Wed, Apr 08, 2015 at 10:46:58AM EDT: > > > On Apr 8, 2015, at 7:23 AM, Lev Givon wrote: > > > > The output of ompi_info --parsable is somewhat difficult to parse > > programmatically because it doesn't escape or quote fields that contain > > colons, > > e.g., > >

Re: [OMPI users] 1.8.4 behaves completely different from 1.6.5

Hmmm…could you try 1.8.5rc1? We’ve done some thread-related stuff on it, but we may not have solved this level of use just yet. We are working on the new1.9 series that we hope to make more thread friendly http://www.open-mpi.org/software/ompi/v1.8/

Re: [OMPI users] parsability of ompi_info --parsable output

I think the assumption was that people would parse this as follows: * entry before the first colon is the category * entry between first and second colons is the subcategory * everything past the second colon is the value You are right, however, that the current format precludes the use of an

Re: [OMPI users] 1.8.3 executable with 1.8.4 mpirun/orted?

Meantime, I’ve created a patch that should address this problem: https://github.com/open-mpi/ompi-release/pull/227 If you can and would like, please see if it resolves this for you. > On Apr 7, 2015, at 9:29 PM, Ralph Castain wrote: > > Hmm

Re: [OMPI users] OpenMPI 1.8.4 - Java Library - allToAllv()

I think the following paragraph might be useful. Its in MPI-3, page 142 lines 16-20: "The type-matching conditions for the collective operations are more strict than the corresponding conditions between sender and receiver in point-to-point. Namely, for collective operations, the amount of dat

Re: [OMPI users] OpenMPI 1.8.4 - Java Library - allToAllv()

In the interim, perhaps another way of addressing this would be to ask: what happens when you run your reproducer with MPICH? Does that work? This would at least tell us how another implementation interpreted that function. > On Apr 7, 2015, at 10:18 AM, Ralph Castain wrote: > > I’m afraid w

[OMPI users] parsability of ompi_info --parsable output

The output of ompi_info --parsable is somewhat difficult to parse programmatically because it doesn't escape or quote fields that contain colons, e.g., build:timestamp:Tue Dec 23 15:47:28 EST 2014 option:threads:posix (MPI_THREAD_MULTIPLE: no, OPAL support: yes, OMPI progress: no, ORTE progress:

Re: [OMPI users] 1.8.3 executable with 1.8.4 mpirun/orted?

Hmmm…yeah, we’ve been discussing this point. It’s a bit of a mixed bag. We hit problems where people don’t get their paths set correctly on remote machines, and then we hang because of bad connections between incompatible versions. Same time, we hit these situations. We’re getting ready to rele