Intel® oneAPI HPC Toolkit
Get help with building, analyzing, optimizing, and scaling high-performance computing (HPC) applications.
1890 Discussions

More than 18 nodes of the MPI test run slowly

RuiXi
Beginner
86 Views

Hello,

 

Here is my environment information

Intel one API 2021.2

CentOS 7.6

MLNX_OFED_LINUX-5.3-1.0.0.1

UCX 1.10

I have two HCAs per node, but only one Active.

 

When I run a test of 18 nodes, I get results in 0.9 seconds.

 

[root@n00001 mpi-test]# time mpirun -np 18 -ppn 1 -f hostfile2  hostname
n07061
n07068
n07069
n07063
n07064
n07074
n07073
n07072
n07062
n07067
n07066
n07070
n07065
n07071
n07059
n07060
n07057
n07058

real 0m0.939s
user 0m0.841s
sys 0m0.871s

 

 

When I run tests with more than 18 nodes, I need 41 seconds to get results.

 

[root@n00001 mpi-test]# time mpirun -np 20 -ppn 1 -f hostfile2  hostname
n07066
n07070
n07073
n07067
n07075
n07068
n07065
n07069
n07074
n07071
n07072
n07076
n07059
n07057
n07058
n07060
n07064
n07063
n07061
n07062

real 0m41.026s
user 0m0.790s
sys 0m0.875s

 

 

Do you have any suggestions on this question? Thank you

 

0 Kudos
0 Replies
Reply