HPC High Performance Computing: 8.1. TensorBoard


Run TensorBoard in your batch script or by command line in an interactive job:

nsurname@ohpc:~$ salloc --gres=gpu:1
salloc: Granted job allocation 793814
nsurname@node021:~$ ml TensorFlow/2.7.1-foss-2020b-CUDA-11.4.3
nsurname@node021:~$ tensorboard --logdir=/tmp/retrain_logs --port=8008 --bind_all
TensorBoard 2.8.0 at http://node021:8008 (Press CTRL+C to quit)

Find in which node is running the job and configure the port that TensorBoard is using, in this case, it is the node021 and the port number 8008.

Open a new terminal session and create a SSH tunneling:

 ssh -p 22 -L 16006:node021:8008 nsurname@hpc.s.upf.edu

Open a browser in your computer and connect to the following URL:


Take into account that we have not generated any graph. In your case, you will see the data.