[mvapich-discuss] Binding level socket error

[Enrico Calore]

Hi all,
using mvapich2.1-gdr or 2.2-gdr on a node of our cluster we are
experiencing problems trying to use:

MV2_CPU_BINDING_LEVEL=SOCKET

On the other hand; setting:

MV2_CPU_BINDING_LEVEL=CORE

works smoothly.

Do you have any hints about what could be causing this problem and/or
how we could debug it?

In particular, here are all the MV2_* variables we have set:

MV2_PROCESS_TO_RAIL_MAPPING=mlx4_0:mlx4_1
MV2_ENABLE_AFFINITY=1
MV2_CPU_BINDING_POLICY=SCATTER
MV2_CPU_BINDING_LEVEL=SOCKET
MV2_RAIL_SHARING_POLICY=FIXED_MAPPING
MV2_USE_CUDA=1
MV2_SHOW_CPU_BINDING=1
MV2_CUDA_IPC=0
MV2_USE_GPUDIRECT=1
MV2_USE_GPUDIRECT_GDRCOPY=0

And here the error reported when trying to run a simple program as:

mpirun -np 2 ./test

Warning! : Core id 255 does not exist on this architecture!
CPU Affinity is undefined
Error parsing CPU mapping string
INTERNAL ERROR: invalid error code ffffffff (Ring Index out of range) in
MPIDI_CH3I_set_affinity:119
[cli_0]: aborting job:
Fatal error in MPI_Init:
Other MPI error, error stack:
MPIR_Init_thread(514):
MPID_Init(367).......: channel initialization failed
MPIDI_CH3_Init(532)..:



Thanks and Best Regards,

Enrico

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20160421/d84268e2/attachment.sig>