h5md-user
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fortran support


From: Pierre de Buyl
Subject: Re: fortran support
Date: Fri, 29 May 2020 10:10:21 +0200
User-agent: NeoMutt/20180716

Dear Alin,

As a followup: ESPResSo uses H5MD. You can find the implementation here:
https://github.com/espressomd/espresso/tree/python/src/core/io/writer

It uses collective I/O and relies on h5xx https://github.com/fhoefling/h5xx , a
template-based C++ wrapper for the HDF5 library.

There is also the lua code from Peter Colberg that uses collective I/O:
https://git.colberg.org/peter/nano-dimer/tree/nanomotor/h5md.lua

To make parallel I/O possible, H5MD allows particles to be stored in an
unordered manner. You can store the ID to retrieve the order if needed. There
can also be gaps in the dataset so that every MPI process can write to its own
chunk of data.

The examples above are not in Fortran indeed, but the HDF5 library in Fortran
maps closely the C library.

What I sometimes do for "observables" time series it to store them in memory and
write them after the simulation loop is done, to avoid parallel sync to write a
few variables. I believe this is what nano-dimer does as well.

Best regards,

Pierre

On Mon, May 25, 2020 at 09:09:59PM +0100, Alin Marin Elena wrote:
> Dear Pierre,
> 
> thank you very much this is very helpful. we are looking in possible
> adoption of h5md for dlpoly.
> 
> Regards,
> Alin
> 
> Without Questions there are no Answers!
> ______________________________________________________________________
> Dr. Alin Marin ELENA
> http://alin.elena.space/
> ______________________________________________________________________
> 
> On Mon, 25 May 2020 at 20:54, Pierre de Buyl <pierre.debuyl@kuleuven.be> 
> wrote:
> >
> > Dear Alin,
> >
> > The Fortran implementation is complete in the sense that it can be used to 
> > write
> > H5MD 1.1 files.
> >
> > I believe that I am the only user though, so there is not much 
> > documentation. I
> > use it in my colloid simulation code, you can see an example here:
> > https://github.com/pdebuyl-lab/RMPCDMD/blob/pre_1.1/programs/single_dimer.f90
> >
> > It is only serial, as I use OpenMP for parallel programming.
> >
> > Best regards,
> >
> > Pierre
> >
> > On Mon, May 25, 2020 at 06:39:01PM +0100, Alin Marin Elena wrote:
> > > Hi,
> > >
> > > What is at the moment the status of the fortran support.
> > > I see there is nothing listed on the website page but I see
> > > a "reference" implementation
> > > https://github.com/pdebuyl/fortran_h5md
> > >
> > > 1. is this complete?
> > > 2. a quick look indicates is serial only. Am I missing something?
> > >
> > > Regards,
> > > Alin
> > >
> > > Without Questions there are no Answers!
> > > ______________________________________________________________________
> > > Dr. Alin Marin ELENA
> > > http://alin.elena.space/
> > > ______________________________________________________________________
> > >
> >
> > --
> > -----------------------------------------------------------
> > Pierre de Buyl
> > KU Leuven - Institute for Theoretical Physics
> > T +32 16 3 27355
> > W http://pdebuyl.be/
> > -----------------------------------------------------------
> 

-- 
-----------------------------------------------------------
Pierre de Buyl
KU Leuven - Institute for Theoretical Physics
T +32 16 3 27355
W http://pdebuyl.be/
-----------------------------------------------------------



reply via email to

[Prev in Thread] Current Thread [Next in Thread]