[mpiwg-coll] deprecating MPI_GRAPH_CREATE?

[Jeff Hammond]

Does the non-scalable nature of v-collectives prevent people from using them?

:-)

Jeff

On Tue, Dec 2, 2014 at 11:38 AM, Torsten Hoefler <htor at inf.ethz.ch> wrote:
> I consider this risk as low because it says it in the standard that it's
> non-scalable as well as in any recent documentation. Yes, some people will
> write bad codes.
>
> All the Best,
>         Torsten
>
>
> On 12/02/2014 08:28 PM, Jeff Hammond wrote:
>>
>> I created ticket 462 as a placeholder for whenever someone has time to
>> work on this.
>>
>> MPI_GRAPH_CREATE does harm to users because they might use it.
>> Somebody might naively think that distributed graphs are more
>> complicated to use than replicated ones...
>>
>> Jeff
>>
>> On Tue, Dec 2, 2014 at 1:57 AM, Torsten Hoefler <htor at inf.ethz.ch> wrote:
>>>
>>> I am unfortunately too busy these days to drive it (a clean removal will
>>> touch several places). It is ugly but it doesn't do any direct harm to
>>> the
>>> standard or its users.
>>>
>>> Best,
>>>          Torsten
>>>
>>>
>>> On 01.12.2014 17:02, Jeff Hammond wrote:
>>>>
>>>>
>>>> But you're not going to write that proposal? You seem more qualified
>>>> than
>>>> most...
>>>>
>>>> Jeff
>>>>
>>>> Sent from my iPhone
>>>>
>>>>> On Dec 1, 2014, at 7:51 AM, Torsten Hoefler <htor at inf.ethz.ch> wrote:
>>>>>
>>>>> Jeff,
>>>>>
>>>>> There is no such proposal yet (afaik), I'd support a carefully crafted
>>>>> one and I think I wouldn't be alone.
>>>>>
>>>>> Thanks & Best,
>>>>>      Torsten
>>>>>
>>>>>> On 01.12.2014 06:41, Jeff Hammond wrote:
>>>>>> Regarding http://lists.mpi-forum.org/mpiwg-coll/2014/03/0289.php, is
>>>>>> there a proposal to deprecate MPI_GRAPH_CREATE?  It seems that there
>>>>>> are many issues with it and they are deliberately not getting fixed
>>>>>> because a subset of the issues with it make it too horrible to fix.
>>>>>>
>>>>>> I find this situation distasteful as it leads to a poor user
>>>>>> experience.  Let's do the right thing and put MPI_GRAPH_CREATE out of
>>>>>> our misery sooner rather than later.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Jeff
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> ### qreharg rug ebs fv crryF --- http://htor.inf.ethz.ch/ ---
>>>>> Torsten Hoefler           | Assistant Professor
>>>>> Dept. of Computer Science | ETH Zürich
>>>>> Universitätsstrasse 6     | Zurich-8092, Switzerland
>>>>> CAB E 64.1                | Phone: +41 76 309 79 29
>>>>> _______________________________________________
>>>>> mpiwg-coll mailing list
>>>>> mpiwg-coll at lists.mpi-forum.org
>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-coll
>>>>
>>>>
>>>> _______________________________________________
>>>> mpiwg-coll mailing list
>>>> mpiwg-coll at lists.mpi-forum.org
>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-coll
>>>>
>>>
>>> --
>>> ### qreharg rug ebs fv crryF --- http://htor.inf.ethz.ch/ ---
>>> Torsten Hoefler           | Assistant Professor
>>> Dept. of Computer Science | ETH Zürich
>>> Universitätsstrasse 6     | Zurich-8092, Switzerland
>>> CAB E 64.1                | Phone: +41 76 309 79 29
>>> _______________________________________________
>>> mpiwg-coll mailing list
>>> mpiwg-coll at lists.mpi-forum.org
>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-coll
>>
>>
>>
>>
>
> --
> ### qreharg rug ebs fv crryF ------------- http://www.unixer.de/ -----
>
> Torsten Hoefler           | Assistant Professor
> Dept. of Computer Science | ETH Zürich
> Universitätsstrasse 6     | Zurich-8092, Switzerland
> CAB E 64.1                | Phone: +41 76 309 79 29
> _______________________________________________
> mpiwg-coll mailing list
> mpiwg-coll at lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpiwg-coll



-- 
Jeff Hammond
jeff.science at gmail.com
http://jeffhammond.github.io/