PMEMD on Blueridge

Baby steps.

This is pmemd compiled against native MPI libraries executing on the RENCI Blueridge cluster.

The job submission workflow uses the RENCI-CI script library. The job_run script uses Globus tools to transfer the pmemd application and input files to Blueridge. There, it uses PBS tools directly to submit the job.

Job submission is done directly to the new Blueridge CE.

It executes in about two minutes on a regular basis. It takes about 25 minutes on my 8 core Nehalem machine.

The user

  • Logs into the engage-submit node
  • Sets RCI_HOME to the location of the RCI scripts
  • Sources $RCI_HOME/bin/environment.sh
  • Changes directory to the cpmemd app
  • Runs the following command -
job_run --host=brgw1 \
        --user=scox \
        --app=/home/scox/dev/cpmemd \
        --exe=blueridge.sh 

Here’s example output -

At this point the script currently waits for the PBS job to finish then transfers output back to the submit node.

Elegant it’s not. But it was a necessary part of getting oriented with HTPC and gives our researcher something that works now.

This entry was posted in Amber9, Compute Grids, Continuous Integration (CI), Engage VO, Globus, grid, High Throughput Computing (HTC), High Throughput Parallel Computing (HTPC), multicore, OSG, pmemd, RENCI. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s