Poor Performance using PLUMED 2.3.2 Metadynamics with GROMACS 2016.3 on CPU/GPU

[dan.k...@gmail.com]

Hello,

I've having trouble getting good performance with PLUMED 2.3.2 patched to GROMACS 2016.3. Without PLUMED (just GROMACS) my system (10,000 atoms) easily reaches 200 ns/day, on 6CPUS/1GPU with 1MPI/6OpenMP.

With PLUMED I'm having trouble getting even 20 ns/day.

Here is my plumed.dat file in which I'm trying to perform metadynamics over the rmsd, alpharmsd,and betasheet content to analyze the stability of a protein:

MOLINFO STRUCTURE=protein.pdb

alphahelix: ALPHARMSD RESIDUES=all TYPE=DRMSD LESS_THAN={RATIONAL R_0=0.08 NN=8 MM=12}

betasheet: ANTIBETARMSD RESIDUES=all TYPE=DRMSD LESS_THAN={RATIONAL R_0=0.08 NN=8 MM=12}

rmsd: RMSD REFERENCE=protein_ref.pdb

METAD ...

LABEL=metad

ARG=rmsd,alphahelix.lessthan,betasheet.lessthan PACE=500 SIGMA=0.1,3,3 FILE=HILLS BIASFACTOR=20.0 TEMP=300 TAU=10

GRID_MIN=0,0,0

GRID_MAX=5,30,30

... METAD

# monitor the two variables and the metadynamics bias potential

PRINT STRIDE=10 ARG=rmsd,alphahelix.lessthan,betasheet.lessthan,metad.bias FILE=COLVAR

ENDPLUMED

and here is the command I'm executing with (2MPI/7OpenMP/2GPU):

export OMP_NUM_THREADS=7

mpirun -n 2 gmx mdrun -deffnm protein_sim -plumed plumed.dat -tunepme yes -pin on -ntomp 7

I have looked in the PLUMED forum for help but I have been unable to get any kind of significant improvement.

Any advice for optimizing this execution is much appreciated.

Thanks,

Dan

[Carlo Camilloni]

Hi 

Start from here

https://plumed.github.io/doc-v2.3/user-doc/html/_performances.html

C

Sent from my iPhone

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/0e02c97c-a499-43ab-a711-93af95014e86%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Gareth Tribello]

Hello

The big one for you is you should use STRANDS_CUTOFF=1.0 with ANTIBETARMSD.  This will lower the number of RMSD calculations you have to do and should make the calculation much cheaper.

Gareth

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/D15F23BF-8E53-4E90-B142-FBB735F7F5A3%40gmail.com.

[Daniel Kozuch]

Thanks, I got some speed up by including STRANDS_CUTOFF=1.0 and D_MAX=1.0, but I'm still at ~50 ns/day. Is that the best I can hope for? 

I read through https://plumed.github.io/doc-v2.3/user-doc/html/_performances.html but the only other change I saw was moving from TYPE=DRMSD to TYPE=OPTIMAL-FAST but this appears to require WHOLEMOLECULES at every timestep which I assume would be slower.

Dan

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/0e02c97c-a499-43ab-a711-93af95014e86%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/D15F23BF-8E53-4E90-B142-FBB735F7F5A3%40gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "PLUMED users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/plumed-users/_rYW3VjIpMw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to plumed-users+unsubscribe@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/061A7A6B-94F8-469A-8B19-9DD6C2B8569A%40gmail.com.

[Carlo Camilloni]

Hi

Have you tried to see if it is actually slower?

In particular because you are already using RMSD so you already need WHOLEMOLECULES for that.

Have you tried to use OPENMP threads also for PLUMED?

Anyway you will lose something you cannot expect to calculate a lot of conformational parameters for free.

Carlo

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.

To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAOKxWzLbRn7%3Dhgh7YQbLtpnaWsmFm4Dp_2fG9JF03H1gT06uQw%40mail.gmail.com.

[Daniel Kozuch]

I tested WHOLEMOLECULES and it had little impact on performance, but I will keep it since you are correct, I do need it for the RMSD calculation. I have tried different threading settings but increasing the number of threads past 1 or 2 for PLUMED seems to cause drastic performance loss.

Thanks for the help,

Dan

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAOKxWzLbRn7%3Dhgh7YQbLtpnaWsmFm4Dp_2fG9JF03H1gT06uQw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "PLUMED users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/plumed-users/_rYW3VjIpMw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to plumed-users+unsubscribe@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/B3F51BA6-4FA0-4065-9BEA-1FF27E4DC3E8%40gmail.com.