Hi folks
I believe I have previously mentioned the PMIx project - an effort to extend
the PMI interface to support exascale operations. The project has been on the
OMPI web site for a little while now, but I finally got around to populating it
with a description of the planned features. Artem
Or use --all.
On May 26, 2014, at 10:21 AM, Ralph Castain wrote:
> Try adding "--level 9" to the cmd line. It's a new "feature" to try and
> reduce the volume of data being thrown at the user as the majority of params
> are frequently of use only to us developers
>
> On
Try adding "--level 9" to the cmd line. It's a new "feature" to try and reduce
the volume of data being thrown at the user as the majority of params are
frequently of use only to us developers
On May 26, 2014, at 7:14 AM, Kevin Brown wrote:
> Greetings,
>
> I've
Greetings,
I've just noticed that ompi_info (Open MPI v. 1.8.1) is not giving a
complete listing when ran with the following options:
rc000:~ > ~/opt/openmpi-1.8.1-base/bin/ompi_info --param all all
MCA btl: parameter "btl_tcp_if_include" (current value: "",
Building optimized on an IB-based machine:
osc_rdma_data_move.c: In function 'ompi_osc_rdma_callback':
osc_rdma_data_move.c:1633: warning: unused variable 'incoming_length'
osc_rdma_data_move.c: In function 'ompi_osc_rdma_control_send':
osc_rdma_data_move.c:221: warning: 'ptr' may be used
Rolf,
the assert fails because the endpoint reference count is greater than one.
the root cause is the endpoint has been added to the list of
eager_rdma_buffers of the openib btl device (and hence OBJ_RETAIN'ed at
ompi/mca/btl/openib/btl_openib_endpoint.c:1009)
a simple workaround is not to use