running NWCHEM on more than 80 nodes


Clicked A Few Times
Hello all,

I was trying to run some large MP2 calculations therefore I requested many nodes. However, when I tried 90 or 100 nodes, the job died almost immediately with error messages like this:

"armci_TcpNoDelay: getprotobyname on TCP failed!: -1"

Any suggestions?

Thank you!

Yue

Forum Vet
How was the code compiled?
I'm a little surprised to see TCP errors, are you running over TCP/IP?

What is the architecture you are running on, and what is the interconnect?

How did you compile NWChem,i.e. which environment variables did you use to compile the code?

Bert


Forum >> NWChem's corner >> Running NWChem