Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Hi, On Thu, 31 Mar 2016 19:00 Michael Shirts wrote: > > One more question: why does the free energy code have to use its own > kernel? There are some aspects that are only used in (some kinds of) free-energy calculations e.g. soft-core options. So maybe coulomb-only

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Dear Nicolas and Mark: Indeed, I have spent the last 2 days testing the plumed Hrex code for gromacs 5.1.2. This is likely the way that I will go. I was looking for an alternate route because I am cautious of bugs and I thought that if I could do it easily in tested gromacs + a bit of code

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

See: https://groups.google.com/forum/#!topic/plumed-users/eJ0xpnHPb_s and https://github.com/GiovanniBussi/plumed2/tree/v2.2-hrex Nicolas 2016-04-02 12:09 GMT+02:00 Mark Abraham : > Hi, > > On Fri, Apr 1, 2016 at 11:39 PM Christopher Neale < >

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Hi, On Fri, Apr 1, 2016 at 11:39 PM Christopher Neale < chris.ne...@alum.utoronto.ca> wrote: > Dear developers: > > Is it possible with only a small amount of work to modify calc_delta() in > src/programs/mdrun/repl_ex.cpp to add another case to re->type such that > the program sends the

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Dear developers: Is it possible with only a small amount of work to modify calc_delta() in src/programs/mdrun/repl_ex.cpp to add another case to re->type such that the program sends the coordinates to the alternate .tpr information and evaluates the energy completely? This would allow for

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

> One more question: why does the free energy code have to use its own kernel? > I realize that I'm going to sound like an idiot here, but why can't one just > tweak parameters outside of the kernel and then have the optimized kernel do > the dynamics? I presume that one has to step outside of

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Dear Szilárd, Mark, and Michael: Thank you for your suggestions. 1. I did use the icc compiler 2. I tested usage of nstlist=10 by setting verlet-buffer-tolerance=-1 (I had already set nstlist=10 but mdrun changes that to 25 in my previous setup) -- this did not improve the performance 3. I

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Chris, I can suggest two possible tweaks, these won't do miracles, but may give you a bit better performance. * icc is better than gcc at optimizing the naiive free energy kernel code. I observed in the past up to 1.5x faster free energy kernel performance with icc 15 vs gcc 4.9 or so. * The

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Hi, Yeah, unfortunately Michael's pretty much right - the free-energy kernel is currently that cousin nobody talks about (or to). It's essentially unchanged since GROMACS 4.0 days, except that the Verlet scheme has some kludge so that it can call the same kernel that the group scheme used to

Re: [gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Hi, Chris: I'm pretty sure that it's because the nonbonded free energies are much slower than the standard free energies. You state: > I took a look at gmxlib/nonbonded/nb_free_energy.c in v.5.1.2, but I was > unable to find a function called "gmx_waste_time_here()" and beyond that I > was out

[gmx-users] Looking for the source of the H-REMD slowdown when decoupling a lot of atoms

Dear Users: I am trying to do some Hamiltonian replica exchange (H-REMD) in gromacs 5.1.2 and am running up against really large slowdowns when decoupling a large number of atoms. I am decoupling 5360 atoms out of the 15520 atoms in my system. The goal is not to get a PMF, but to enhance