[gmx-users] Potential bug in IMD implementation (Interactive Molecular Dynamics)

Christian Blau blau at kth.se
Tue Nov 19 11:30:15 CET 2019


Hi Xavier,


Bugs are usually best reported via "redmine.gromacs.org". You can self-report when registering there. In this case I 
just went ahead and filed the bug report, if you want to add information to the bug report and check how things are 
developing you can check here:

https://redmine.gromacs.org/issues/3209

There, you can also attach files to the bug report. The most helpful thing for fixing a GROMACS bug is a complete set of 
input files (at least the .tpr though) that reliably reproduces the bug. (Ideally the tiniest system you can conceive, 
but any system is much much better than none)


The video that you linked is also extremely nice to have and helpful

Thanks for the report!

Best,

Christian


On 11/19/19 11:11 AM, Xavier Martinez wrote:
> Dear gmx users,
>
> I came across a bug that is reproducible when using the IMD implementation in several versions of Gromacs including 
> 2019.3 and different operating systems.
>
> This bug occurs with the following steps:
>
> - Run an IMD simulation (gmx mdrun -imdwait -imdpull -imdterm -s mysim.tpr)
>
> - Connect to the simulation using VMD (or other visualization tool that implements the IMD protocol)
>
> - Pause the simulation from VMD ("detach sim" button)
>
> - Reconnect to the simulation
>
> The expected behavior is to have the same performance before and after pausing the simulation.
>
> The actual behavior is a slower and stuttering simulation. Note that this bug does not occur when using NAMD.
>
> See an example of the bug here: https://www.youtube.com/watch?v=_gZ6wA7c4b8 
> <https://www.youtube.com/watch?v=_gZ6wA7c4b8&feature=youtu.be>
>
> An error line appears in the gromacs window: "IMD: Failed to shutdown socket. Did the client already disconnect?" and 
> "IMD: Error in file gromacs-2019.3/src/gromacs/imd/imdsocket.cpp on line 293."
>
> How should I proceed to help this bug getting fixed ?
>
> Best regards,
>
> Xavier Martinez, Ph.D
>
> CNRS-IBPC LBT, Paris
>
>


More information about the gromacs.org_gmx-users mailing list