search

radical-collaboration / extasy-grlsd

Repository to hold the input data and scripts for the ExTASY gromacs-lsdmap work
1 stars 1 forks source link

speed difference rp and standalone #83

Closed euhruska closed 6 years ago

euhruska commented 6 years ago

My issue is that standalone openmm (aprun on a separate script) leads to 250ns/day, but the same commands with radical-pilot barely 23ns/day.

The md.log in /mnt/c/scratch/sciteam/hruska/radical.pilot.sandbox/rp.session.leonardo.rice.edu.eh22.017787.0007/pilot.0000/unit.000005 shows the radical-pilot results. The unit.sh script gives the module load commands. In the standalone case I was running the commands in unit.sh, without -L 7063 in aprun, but including loading the RP variables, but that gave a 10x faster speed

euhruska commented 6 years ago

removed bwpy-environ, but didn't help

andre-merzky commented 6 years ago

I don't have permissions to access the unit script - can you please attach that, and also the the aprun script you used for manual testing? Thanks!

euhruska commented 6 years ago

extasy: unit.000005.txt standalone: unit.000005.pbs.txt

euhruska commented 6 years ago

is there a way to change "-cc 0" to "-cc none" in the aprun command in entk?

andre-merzky commented 6 years ago

Those two scripts result in different performance?? That is really surprising. Could you do me the favor, and pack up the whole pilot sandbox, including all units subdirs? Thanks!

euhruska commented 6 years ago

hmm, I when I ran it interactively I removed traj writing, so that caused the difference.