[mvapich-discuss] mvapich2 slow with default mapping

[Hari Subramoni]

Hello All,

We were able to resolve this issue off the list. It was related to the HCA
aware process mapping feature in MVAPICH2. In MVAPICH2 2.2rc1, there is an
environment variable named "MV2_HCA_AWARE_PROCESS_MAPPING". By setting this
to 0, the user was able to prevent over subscription and the poor
performance he was observing. Unfortunately, this parameter is not
available in the MVAPICH2-2.1 series.

We are working on resolving this issue and hope to have a solution
available with the next release.

Regards,
Hari.

On Thu, Apr 28, 2016 at 11:57 AM, Hari Subramoni <subramoni.1 at osu.edu>
wrote:

> Hello Götz,
>
> It looks like some sort of oversubscription is happening here. Could you
> please send us the following information?
>
> 1. Output of program run after setting MV2_SHOW_CPU_BINDING=1
>
> 2. Output of ibv_devinfo executed on the system where you're seeing the
> degradation.
>
> Thanks,
> Hari.
> On Apr 28, 2016 10:11 AM, "Götz Waschk" <goetz.waschk at gmail.com> wrote:
>
>> Dear Mvapich2 experts,
>>
>> I'm currently evaluating OpenHPC packages, including mvapich2 2.1.
>> I've tested the speed using the Intel MPI benchmarks and I have
>> noticed, that the first benchmark PingPong is behaving differently
>> when run with 16 cores vs. 2 cores, although only two cores are in use
>> and the remaining processes simply wait. The full results are in
>> OpenHPC's issue tracker on github:
>> https://github.com/openhpc/ohpc/issues/207#issuecomment-212319647
>>
>> As you can see there, the configuration change to set these variables
>> helped:
>>
>> export MV2_SHOW_CPU_BINDING=1
>> export MV2_CPU_MAPPING=0:1:2:3:4:5:6:7
>>
>> I still wonder why they have such an influence and the default setting
>> isn't sufficient here.
>>
>> Regards,
>> Götz Waschk
>>
>> _______________________________________________
>> mvapich-discuss mailing list
>> mvapich-discuss at cse.ohio-state.edu
>> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20160505/c65ded7e/attachment-0001.html>