Thanks Orion!!
> On Oct 29, 2015, at 6:07 PM, Orion Poplawski wrote:
>
> On 10/28/2015 01:19 PM, Ralph Castain wrote:
>> Last call - please give this version a whirl.
>>
>> http://www.open-mpi.org/software/ompi/v1.10/
>>
>
> Fedora rpm builds fine.
>
>
> --
> Orion Poplawski
> Technical Ma
Great -- glad you figured it out!
> On Oct 29, 2015, at 3:26 PM, Nick Papior wrote:
>
> Problem resolved, the later intel versions source files automatically adds
> their own MPI library to the LD_LIBRARY_PATH, hence the linked mpi library
> was the intel MPICH library and _not_ the openmpi li
On 10/28/2015 01:19 PM, Ralph Castain wrote:
Last call - please give this version a whirl.
http://www.open-mpi.org/software/ompi/v1.10/
Fedora rpm builds fine.
--
Orion Poplawski
Technical Manager 303-415-9701 x222
NWRA/CoRA DivisionFAX: 303-415-9702
Thank you Paul!
> On Oct 29, 2015, at 5:24 PM, Paul Hargrove wrote:
>
> I have completed testing to the best of my ability.
> That includes waiting for my slow QEMU-emulated ARM and MIPS systems.
> My testing also includes real h/w testing of ARMv7, ARMv8 (aka AARCH64),
> PPC64el and IA64.
> Th
Thank you, Paul!!
> On Oct 29, 2015, at 5:24 PM, Paul Hargrove wrote:
>
> I have completed testing to the best of my ability.
> That includes waiting for my slow QEMU-emulated ARM and MIPS systems.
> My testing also includes real h/w testing of ARMv7, ARMv8 (aka AARCH64),
> PPC64el and IA64.
>
I have completed testing to the best of my ability.
That includes waiting for my slow QEMU-emulated ARM and MIPS systems.
My testing also includes real h/w testing of ARMv7, ARMv8 (aka AARCH64),
PPC64el and IA64.
Though a few systems have been retired since my last round of testing, I
count 79 conf
Problem resolved, the later intel versions source files automatically adds
their own MPI library to the LD_LIBRARY_PATH, hence the linked mpi library
was the intel MPICH library and _not_ the openmpi library.
Bottomline, be aware that LD_LIBRARY_PATH has to be corrected (or prepended
the correct p
Hi all,
I am not really sure whether this is an intel or ompi bug?
I have now tried with all 1.10.* versions (including 1.10.1rc[23]) to test
whether this could be circumvented.
I use the intel 2016.1.0.423501 compiler.
I can successfully compile OpenMPI and use the C-compiler. But as soon as I
