Hello, I am a beginner on MPI programming, and right now I am working on an MPI project. I got a few questions related to implementation issues: 1. when I run a simple MPI hello world on multiple nodes, (I already installed mpich3 library on master node, mount the nfs, shared the executable file and mpi library, set slave node to be keyless ssh), my program was stoped there say: bash: /mnt/mpi/mpich-install/bin/hydra_pmi_proxy: /lib/ld-linux.so.2: bad ELF interpreter: No such file or directory. I can not get rid of it for a long times. even though I reset everything (I already add PATH=/mnt/mpi/mpich-install/bin:$PATH in .bash_profile). Do you have any clues on this problems? 2. for multiple servers, each of them has 10G ethernet card. for example, one network card address is eth5: 10.0.5.55. So if I want to launch MPI communication through 10G network card. Should I set the hostfile as: 10.0.5.55:$(PROCESS_NUM)? Or using iface eth5 Thanks a lot! -- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219