[gmx-developers] new 64-bit int support and MPI

Jeff Hammond jeff.science at gmail.com
Tue Dec 24 19:08:53 CET 2013


You can always communicate these in bytes/chars, so it isn't a big problem.  And I think long long suffices for reductions. 

Jeff

Sent from my iPhone

> On Dec 24, 2013, at 10:15 AM, Erik Lindahl <erik.lindahl at scilifelab.se> wrote:
> 
> I’d rather not - our CMake configuration is already _way_ to complex, and I’d prefer not to turn the Gromacs distribution into a kitchen sink that builds every single tool needed for molecular simulation :-)
> 
> However, for the time being the only int64_t's I imagine we might need communicating are related to sizes (e.g. number of particles, particle index, etc.). In absolutely worst case, we can always turn that into MPI_OPAQUE objects, or manually separate the high/low 32-bit components.
> 
> Cheers,
> 
> Erik
> 
> 
>> On 24 Dec 2013, at 16:30, Jeff Hammond <jeff.science at gmail.com> wrote:
>> 
>> Can't you use cmake to build MPI if it is necessary? I can provide enough configure invocations to cover the cases where this would be used. 
>> 
>> Jeff
>> 
>> Sent from my iPhone
>> 
>>> On Dec 24, 2013, at 6:41 AM, Erik Lindahl <erik.lindahl at scilifelab.se> wrote:
>>> 
>>> Hi,
>>> 
>>> Hm - good point. However, can’t we simply ask the MPI library to transport 8 chars instead in that case? (assuming that the endian is the same on all nodes in an MPI job)
>>> 
>>> Cheers,
>>> 
>>> Erik
>>> 
>>>> On 24 Dec 2013, at 13:34, Mark Abraham <mark.j.abraham at gmail.com> wrote:
>>>> 
>>>> Hi,
>>>> 
>>>> I'm happy with the recent change to requiring 64-bit integers, but I note that this means we have required that the MPI library support the MPI_INT64_T datatype from MPI 2.2. AFAIK that's probably OK in practice, but I thought it worth mentioning / discussing.
>>>> 
>>>> We should update the install guide and add a CMake test for its support, however, since there will still be plenty of people with earlier non-compliant MPI libraries. I imagine CentOS will take about 5 years to update that far...
>>>> 
>>>> Cheers,
>>>> 
>>>> Mark
>>>> -- 
>>>> Gromacs Developers mailing list
>>>> 
>>>> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_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-developers or send a mail to gmx-developers-request at gromacs.org.
>>> 
>>> -- 
>>> Gromacs Developers mailing list
>>> 
>>> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_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-developers or send a mail to gmx-developers-request at gromacs.org.
>> -- 
>> Gromacs Developers mailing list
>> 
>> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_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-developers or send a mail to gmx-developers-request at gromacs.org.
> 
> -- 
> Gromacs Developers mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_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-developers or send a mail to gmx-developers-request at gromacs.org.


More information about the gromacs.org_gmx-developers mailing list