[gmx-users] Problems during installation

Rajat Desikan rajatdesikan at gmail.com
Fri Jul 20 08:56:37 CEST 2018 

Hi Szilard,
Your advice was perfect: I've installed gromacs 2018_2 on my computer and
I'm currently testing some sample runs which have speeded up 20 times!

However, I consistently find that mdrun only selects one GPU when two are
available. Why does this happen?

*nvidia-smi*

Fri Jul 20 12:22:26 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 396.24.10              Driver Version: 396.24.10
   |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr.
ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute
M. |
|===============================+======================+======================|
|   0  GeForce GTX 108...  Off  | 00000000:01:00.0 Off |
N/A |
| 64%   60C    P0    79W / 280W |     18MiB / 11176MiB |      0%
Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 108...  Off  | 00000000:02:00.0 Off |
N/A |
|  0%   33C    P8     9W / 280W |      2MiB / 11178MiB |      0%
Default |
+-------------------------------+----------------------+----------------------+

*gmx_2018_2 mdrun -v -deffnm MD_test -nb gpu -gpu_id 01*

Using 1 MPI process
Using 12 OpenMP threads

1 GPU auto-selected for this run.
Mapping of GPU IDs to the 2 GPU tasks in the 1 rank on this node:
  PP:0,PME:0

Regards,
Rajat


On Thu, Jul 19, 2018 at 8:40 PM Szilárd Páll <pall.szilard at gmail.com> wrote:

> On Mon, Jul 16, 2018 at 7:43 PM Rajat Desikan <rajatdesikan at gmail.com>
> wrote:
>
> > Hi Mark,
> >
> > Thank you for the quick answer. My group is experimenting with a
> GPU-heavy
> > processor-light configuration similar to the Amber machines available
> from
> > Exxact (https://www.exxactcorp.com/AMBER-Certified-MD-Systems). In our
> > understanding, for this configuration to be successful performance-wise,
> > gromacs would've to do all the calculating on GPUs, and not just
> > electrostatics. That's why we're looking at openmm enabled versions.
> >
>
> For that you'd need the ~8 year old GROMACS 4.5 with OpenMM 2.x (IIRC)
> support which is ancient and unsupported so really not a good idea.
>
> Instead, use the latest release. Hardware-wise, ddeally you'd want to have
> 2-3 faster (wokstation) or 3-5 slower (server) cores per GPU for good
> CPU-GPU balance.
> --
> Szilárd
>
>
> > If you and others have alternate suggestions, we would be really eager to
> > hear them. Thank you for your time.
> >
> > Regards,
> > Rajat
> >
> > On Mon, 16 Jul 2018, 10:25 pm Mark Abraham, <mark.j.abraham at gmail.com>
> > wrote:
> >
> > > Hi,
> > >
> > > This code was not ever functional in the 4.6 branch, so if you want it
> to
> > > work, try 4.5.7. But really there is no reason to want to build it any
> > more
> > > - use the CUDA and OpenCL ports available in more recent GROMACS.
> > >
> > > Mark
> > >
> > > On Mon, Jul 16, 2018 at 6:47 PM Rajat Desikan <rajatdesikan at gmail.com>
> > > wrote:
> > >
> > > > Dear all,
> > > > I'm trying to install openmm enabled version of gromacs 4.6.7 on a
> > 6-core
> > > > i7 machine with two GTX 1080 Ti GPUs. I am repeatedly facing a rather
> > > > strange error while running make:
> > > >
> > > > /home/sujit/gromacs-4.6.7/src/contrib/mdrun_openmm.c: In function
> > > ‘cmain’:
> > > > /home/sujit/gromacs-4.6.7/src/contrib/mdrun_openmm.c:221:23: warning:
> > > > missing braces around initializer [-Wmissing-braces]
> > > >    gmx_hw_opt_t hw_opt={0,0,0,0,TRUE,FALSE,0,NULL};
> > > >                        ^
> > > > */home/sujit/gromacs-4.6.7/src/contrib/mdrun_openmm.c:221:23: note:
> > (near
> > > > initialization for ‘hw_opt’)*
> > > > */home/sujit/gromacs-4.6.7/src/contrib/mdrun_openmm.c:247:41: error:
> > > > ‘gmx_hw_opt_t {aka struct <anonymous>}’ has no member named ‘gpu_id’;
> > did
> > > > you mean ‘gpu_opt’?*
> > > > *     { "-gpu_id",  FALSE, etSTR, {&hw_opt.gpu_id},*
> > > >                                          ^
> > > > src/kernel/CMakeFiles/mdrun.dir/build.make:302: recipe for target
> > > > 'src/kernel/CMakeFiles/mdrun.dir/__/contrib/mdrun_openmm.c.o' failed
> > > > make[2]: ***
> > > [src/kernel/CMakeFiles/mdrun.dir/__/contrib/mdrun_openmm.c.o]
> > > > Error 1
> > > > CMakeFiles/Makefile2:1902: recipe for target
> > > > 'src/kernel/CMakeFiles/mdrun.dir/all' failed
> > > > make[1]: *** [src/kernel/CMakeFiles/mdrun.dir/all] Error 2
> > > > Makefile:162: recipe for target 'all' failed
> > > > make: *** [all] Error 2
> > > >
> > > > I have installed cuda 9.1, openmm, fftw, open-mpi and all other
> > necessary
> > > > software, so I'm at a loss. Please help me out. Google did not have
> > much
> > > > leads.
> > > >
> > > > Thank you.
> > > >
> > > > Regards,
> > > > Rajat
> > > >
> > > > --
> > > > Dr. Rajat Desikan (Post Doctoral Fellow)
> > > > Prof. Narendra M Dixit's Lab (no 1),
> > > > Dept. of Chemical Engineering,
> > > > Indian Institute of Science, Bangalore
> > > > --
> > > > Gromacs Users mailing list
> > > >
> > > > * Please search the archive at
> > > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > > > posting!
> > > >
> > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > >
> > > > * For (un)subscribe requests visit
> > > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> or
> > > > send a mail to gmx-users-request at gromacs.org.
> > > --
> > > Gromacs Users mailing list
> > >
> > > * Please search the archive at
> > > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > > posting!
> > >
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > > * For (un)subscribe requests visit
> > > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > > send a mail to gmx-users-request at gromacs.org.
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.



-- 
Dr. Rajat Desikan (Post Doctoral Fellow)
Prof. Narendra M Dixit's Lab (no 1),
Dept. of Chemical Engineering,
Indian Institute of Science, Bangalore

More information about the gromacs.org_gmx-users mailing list