[mvapich-discuss] [EXTERNAL] Running LQCD chroma with MVAPICH2 encountered errors

Ran Du bella.ran.du at gmail.com
Wed Sep 11 05:24:23 EDT 2019


Dear Balint,

        Thanks a lot for your reply.

        My colleague, Yujiang, did the Chroma test, and I am cc'ing him to
answer your questions about the test details.

        And I agree with your advice about issue reproduction, I will help
Yujiang, my colleague to reproduce the issue and post more detailed logs
here.

Kind regards,
Ran


On Wed, Sep 11, 2019 at 12:06 AM Balint Joo <bjoo at jlab.org> wrote:

> Hi Ran,
>  On the offchance that this is a chroma/QUDA related issue, we can take
> this discussion offline
> (I am one of the Chroma delveopers… ), and am cc’ing Kate Clark who is the
> lead on QUDA.
>
> I saw that on your log, the relative solver residuum was cut off. Did the
> BiCGStab solver actually reach the desired target?
>
> Also what size of lattice are you trying to run ( could it be that one or
> two of the components
> is running out of memory). I see you use a prebuilt Chroma which you
> loaded as a ‘module’.
>
> It would maybe be useful if we could reproduce your issue?
>
> Best,
> B
>
> > On Sep 10, 2019, at 5:33 AM, Ran Du <bella.ran.du at gmail.com> wrote:
> >
> > <mvapich2_test_log.pdf>
>
>
> -----------------------------------------------------------------------------
> Dr Balint Joo   High Performance Computational Scientist
> Jefferson Lab
> 12000 Jefferson Ave Suite 3, MS 12B2, Room F217
> Newport News VA 23606, USA
> Tel: +1-757-269-5339         email: bjoo AT jlab.org
>
> -----------------------------------------------------------------------------
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20190911/5b7e5f69/attachment.html>


More information about the mvapich-discuss mailing list