Batch rendering with VisIt on Stampede1 @ TACC
Often there is a need to perform visualization on large number of files. We could create a VisIt session file on our personal computer then copy the included session file to a cluster then point to the larger data on the cluster and render visualizations in batch. This process is not very intuitive and requires a bit of VisIt foo, so here is how to accomplish batch rendering on Stampede cluster at Texas Advanced Computing Center (TACC)
1) Download and unzip this bundle anywhere on Stampede. This zip bundle is a standalone setup to perform batch rendering on Stampede. The VisIt session file was created on desktop computer using the curv3d.silo example data available here. You may optionally examine this session file on your computer as follows: Open VisIt > File > Restore Session With Sources > Browse and choose curv3d_example.session file > Locate the curv3d.silo data on your computer.
2) batch_render.sh - Add your account number on line #17, adjust nodes, time, name, etc as needed for other uses, for this example only account change is needed.
3) render.py - Examine this file, we restore the session with data on Stampede at /opt/apps/intel13/mvapich2_1_9/visit/data/curv3d.silo
4) Submit batch job on Stampede as: % sbatch batch_render.sh
5) Once the batch job completes you should have an image named curv_img_0000.png which looks like one below.
Amit Chourasia, SDSC, UC San Diego
XSEDE - ECSS: This work used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation grant number ACI-1053575.