Hi HPC Developers,
We are having a Sandy Bridge cluster of 8 nodes having the following:
Hardware:
1U rackmount enclosure
Intel Xeon E5-2680v3 processor x 2ea
128GB Memory
5.5TB Pure Storage
56Gbps Infiband switch
Software:
CentOS 6.6 x64
Intel® Parallel Studio XE 2015
Intel MPI Library 5.0.3
NFS share
In the following cases, operate normally.
1. If you run mpirun from the Terminal Console
2. If you perform a bash mpirun to child processes in a single-threaded program
3. If you perform a bash mpirun in the child process via a Thread in multi-threaded program
However, the operation is delayed in the following cases:
If you perform a bash mpirun in the child process at TCP service hosting program
Attach a reference document to help you understand the above.
Ask for advice on troubleshooting.