[Lab] GPU Status Checking

less than 1 minute read

Currently, we are maintaining four clusters: mark0, mark1, mark2, mark3.
For GPU status checking, mark0 works as a server where each client (clusters including mark0) runs a shell scripts informing GPU usages and

Server Setting

In mark0, run

$ sh run_server.sh

Client Setting

For every clusters, run

$ nohup ./gpu_daemon.sh mark0 > /dev/null &`

Leave a Comment