4964a5e98b
http://www.open-mpi.org/community/lists/devel/2013/10/13072.php Add support for pinning GPU Direct RDMA in openib BTL for better small message latency of GPU buffers. Note that none of this is compiled in unless CUDA-aware support is requested. This commit was SVN r29680.
44 строки
1.7 KiB
Plaintext
44 строки
1.7 KiB
Plaintext
# -*- text -*-
|
|
#
|
|
# Copyright (c) 2009 Sun Microsystems, Inc. All rights reserved.
|
|
# $COPYRIGHT$
|
|
#
|
|
# Additional copyrights may follow
|
|
#
|
|
# $HEADER$
|
|
#
|
|
[eager_limit_too_small]
|
|
The "eager limit" MCA parameter in the %s BTL was set to a value which
|
|
is too low for Open MPI to function properly. Please re-run your job
|
|
with a higher eager limit value for this BTL; the exact MCA parameter
|
|
name and its corresponding minimum value is shown below.
|
|
Local host: %s
|
|
BTL name: %s
|
|
BTL eager limit value: %d (set via btl_%s_eager_limit)
|
|
BTL eager limit minimum: %d
|
|
MCA parameter name: btl_%s_eager_limit
|
|
|
|
[cuda_eager_limit_too_small]
|
|
The "CUDA eager limit" MCA parameter in the %s BTL was set to a value which
|
|
is too low for Open MPI to function properly. Please re-run your job
|
|
with a higher CUDA eager limit value for this BTL; the exact MCA parameter
|
|
name and its corresponding minimum value is shown below.
|
|
|
|
Local host: %s
|
|
BTL name: %s
|
|
BTL CUDA eager limit value: %d (set via btl_%s_cuda_eager_limit)
|
|
BTL CUDA eager limit minimum: %d
|
|
MCA parameter name: btl_%s_cuda_eager_limit
|
|
|
|
[cuda_rdma_limit_too_small]
|
|
The "CUDA rdma limit" MCA parameter in the %s BTL was set to a value which
|
|
is too low for Open MPI to function properly. Please re-run your job
|
|
with a higher CUDA rdma limit value for this BTL; the exact MCA parameter
|
|
name and its corresponding minimum value is shown below.
|
|
|
|
Local host: %s
|
|
BTL name: %s
|
|
BTL CUDA rndv limit value: %d (set via btl_%s_cuda_rdma_limit)
|
|
BTL CUDA rndv limit minimum: %d
|
|
MCA parameter name: btl_%s_cuda_rdma_limit
|