Re: [OMPI devel] Neighbor collectives with periodic Cartesian topologies of size one

Nathan, here is a revision of the previously attached patch, and that supports graph and dist graph. Cheers, Gilles On 2014/09/30 0:05, Nathan Hjelm wrote: > An equivalent change would need to be made for graph and dist graph as > well. That will take a little more work. Also, I was avoiding

[OMPI devel] MPI_Comm_spawn crashes with the openib btl

Folks, the dynamic/spawn test from the ibm test suite crashes if the openib btl is detected (the test can be ran on one node with an IB port) here is what happens : in mca_btl_openib_proc_create, the macro OPAL_MODEX_RECV(rc, _btl_openib_component.super.btl_version,

[OMPI devel] recomended software stack for development?

Hi all, I kind of broke something with mail mail configuration so I haven't been able to properly answer to this earlier, sorry. @Jsquyres We are planning to work on fault tolerance and improved scheduling cappabilities for HPC. To do so, we are first focusing on serial tasks, and in a next step

Re: [OMPI devel] recomended software stack for development?

Kewl. FWIW: we already have the ability to migrate processes in the ORTE code. You can tell the system to try and restart the process in its existing location N number of times before requesting relocation. Of course, if a node fails, then we automatically relocate the procs to other nodes.

Re: [OMPI devel] Neighbor collectives with periodic Cartesian topologies of size one

Not quite right. There still is no topology information at collective selection time for either graph or dist graph. -Nathan On Tue, Sep 30, 2014 at 02:03:27PM +0900, Gilles Gouaillardet wrote: >Nathan, > >here is a revision of the previously attached patch, and that supports >

Re: [OMPI devel] [patch] libnbc intercommunicator iallgather bug

Hi Takahiro, Thanks very much for the patch and the test! After the git migration we'll open an issue and patch nbc_iallgather. This will get pushed to 1.8.4. Howard -Original Message- From: devel [mailto:devel-boun...@open-mpi.org] On Behalf Of Kawashima, Takahiro Sent: Monday,

[OMPI devel] Github migration: tomorrow

It's happening tomorrow, October 1, 2014, starting at 8am US Eastern time. There was discussion about Bitbucket vs. Github, and all things being equal (except the cost!), we're going with the original plan of the main OMPI repo at Github. The plan tomorrow is the same as it was last week: -

Re: [OMPI devel] MPI_Comm_spawn crashes with the openib btl

I fixed this in r32818 - the components shouldn't be passing back success if the requested info isn't found. Hope that fixes the problem. On Sep 30, 2014, at 1:54 AM, Gilles Gouaillardet wrote: > Folks, > > the dynamic/spawn test from the ibm test suite

Re: [OMPI devel] Github migration: tomorrow

Hi Jeff, When's the latest today that we can do checkins without causing problems? Howard -Original Message- From: devel [mailto:devel-boun...@open-mpi.org] On Behalf Of Jeff Squyres (jsquyres) Sent: Tuesday, September 30, 2014 9:57 AM To: Open MPI Developers List Subject: [OMPI

Re: [OMPI devel] Github migration: tomorrow

8am US Eastern tomorrow morning is the cutoff. Anything you do tonight will be fine. Note, however, that open CMRs will NOT be moved over to Github -- you'll have to re-file them as pull requests after the migration. On Sep 30, 2014, at 2:09 PM, Pritchard Jr., Howard

Re: [OMPI devel] Broken abort backtrace functionality

Should be fixed in r32821, scheduled for 1.8.4 On Sep 29, 2014, at 2:00 PM, Deva wrote: > I looks like OMPI_MCA_mpi_abort_print_stack=1 is broken. I'm seeing > following warning with it. > > -- > $mpirun -np 2 -x