Propane IBI tutorial taking a long time

[Satyen Dhamankar]

Hello, 

I have been running the propane IBI tutorial for propane. The IBI for propane takes quite some time to run. Step_001 has taken more than an hour to complete. 

This is the slurm script I have used to run this job: 

#!/bin/bash
#
#SBATCH --job-name=R1.5-CG # create a short name for your job
#SBATCH --qos=vshort # _quality of service_
#SBATCH --nodes=1 # node count
#SBATCH --ntasks-per-node=12 # number of tasks per node
#SBATCH --cpus-per-task=1 # cpu-cores per task (>1 if multi-threaded tasks)
#SBATCH --mem=10GB # total memory requested
##SBATCH --mem-per-cpu=4G # memory per cpu-core (4G per cpu-core is default)
#SBATCH --gres=gpu:1 # number of gpurs per node
#SBATCH --time=2:30:00 # total run time limit (HH:MM:SS)
#SBATCH --mail-type=all # send email on job start, end, and fail
module purge
module load intel/19.0/64/19.0.5.281 # for running gromacs
module load rh/devtoolset/7
module load intel-mkl/2019.5/5/64
module load cmake/3.x
# run the job
set -e
# keep track of the last executed command
trap 'last_command=$current_command; current_command=$BASH_COMMAND' DEBUG
# echo an error message before exiting
#trap 'echo "\"${last_command\" command filed with exit code $?." ' EXIT
bash run.sh

Is taking such a long time for a sample IBI run normal? 

[Christoph Junghans]

Where does it get stuck? In the gromacs run or after that?

Christoph 

--
Join us on Slack: https://join.slack.com/t/votca/signup
---
You received this message because you are subscribed to the Google Groups "votca" group.
To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca/dadf2e09-20dd-4e63-8eaf-f748a3d390a3n%40googlegroups.com.

--

Christoph Junghans
Web: http://www.compphys.de

[Satyen Dhamankar]

The gromacs  run. It is strange because the spce ibi did not take as long. 

I have attached my md.log file. After 2.5 hours, only 11999 steps were completed for a 3000 molecule simulation... 

[Christoph Junghans]

On Fri, Apr 2, 2021 at 7:16 AM Satyen Dhamankar <saty...@gmail.com> wrote:
>
> The gromacs run. It is strange because the spce ibi did not take as long.
> I have attached my md.log file. After 2.5 hours, only 11999 steps were completed for a 3000 molecule simulation...

Hmm, that might be more a question for the gromacs user mailing list.

From your md.log, I see that Gromacs detected 28 cores
"Running on 1 node with total 28 cores, 28 logical cores"
and also runs on them
"Initializing Domain Decomposition on 28 ranks"

However you only requested 12 cores in your batch script:
#SBATCH --ntasks-per-node=12

If the other cores are used by another user that could lead to
performance problems as gromacs always runs as slow as the slowest
core.
gromacs has an option to limit the number cores, I believe "-nt 12",
which you can hand over to gromacs from the xml file using
cg.inverse.gromacs.mdrun.opts.

Christoph

> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/0847a33c-636b-4a09-82c0-4fdd5a63c1e5n%40googlegroups.com.

[Satyen Dhamankar]

Christoph, 

Thank you for the advice. 

I have added -nt 12 in the following manner:

<inverse>
<!-- 200*0.00831451 gromacs units -->
  <kBT>1.66290</kBT>
  <!-- use gromacs as simulation program -->
  <program>gromacs</program>
  <!-- gromacs specific options -->
  <gromacs>
    <!-- trash so many frames at the beginning -->
    <equi_time>20</equi_time>
    <!-- grid for table*.xvg !-->
    <table_bins>0.002</table_bins>
    <!-- cut the potential at this value (gromacs bug) -->
    <pot_max>1000000</pot_max>
    <!-- extend the tables to this value -->
    <table_end>2.5</table_end>
    <mdrun.opts>-nt 12</mdrun.opts>
</gromacs>

however, i am getting the same message:

"Running on 1 node with total 28 cores, 28 logical cores"
and also runs on them
"Initializing Domain Decomposition on 28 ranks"

I have attached my settings.xml file to this document (the only change i have made is the -nt 12 part). 

Speaking of, I made the change and wrote -nt 8 in the methanol tutorial, and that one worked fine... 

[Satyen Dhamankar]

Christoph,

I also ran some ibi tests on the urea-water tutorial example. 

I am running into the same issues. The IBI is running at an incredibly slow pace. From what I have gathered, it seems to be a problem when you have more than one bead type in the coarse-grained system that IBI gets incredibly slow. I have heard that tabulated potentials are incredibly slow in GMX... Could that be a contributing factor? 

[Christoph Junghans]

On Fri, Apr 2, 2021, 13:10 Satyen Dhamankar <saty...@gmail.com> wrote:

Christoph, 

Thank you for the advice. 

I have added -nt 12 in the following manner:

<inverse>
<!-- 200*0.00831451 gromacs units -->
  <kBT>1.66290</kBT>
  <!-- use gromacs as simulation program -->
  <program>gromacs</program>
  <!-- gromacs specific options -->
  <gromacs>
    <!-- trash so many frames at the beginning -->
    <equi_time>20</equi_time>
    <!-- grid for table*.xvg !-->
    <table_bins>0.002</table_bins>
    <!-- cut the potential at this value (gromacs bug) -->
    <pot_max>1000000</pot_max>
    <!-- extend the tables to this value -->
    <table_end>2.5</table_end>
    <mdrun.opts>-nt 12</mdrun.opts>

Try

<mdrun>

  <opts>-nt 12</opts>

</mdrun>

Here

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/aeaccc1d-3236-409e-8dcf-2377c079f2dbn%40googlegroups.com.

[Christoph Junghans]

On Fri, Apr 2, 2021, 20:58 Satyen Dhamankar <saty...@gmail.com> wrote:

Christoph,

I also ran some ibi tests on the urea-water tutorial example. 

I am running into the same issues. The IBI is running at an incredibly slow pace. From what I have gathered, it seems to be a problem when you have more than one bead type in the coarse-grained system that IBI gets incredibly slow. I have heard that tabulated potentials are incredibly slow in GMX... Could that be a contributing factor? 

Tabulated interactions are slow but not more than a factor 2.

More bead types means more energy groups and that could be the issue, but for multiple tables you need multiple energy groups.

You could try an older gromacs version...

But not many other users have reported that issue before, so I would investigate more why gromacs is slow first.

Christoph

To view this discussion on the web visit https://groups.google.com/d/msgid/votca/75b980dd-748f-4cfd-b0fc-b9a5ccb0741cn%40googlegroups.com.

[Satyen Dhamankar]

Christoph, 

Actually, once i got the 

<mdrun> 
 <opts>-nt 8</opts>

</mdrun> 

into the settings.xml file, it all start moving along faster. 

Thank you so much for your help!!