[Users] meeting minutes for 2023-10-19

[Gabriele Bozzola]

Hi Sam,

I can confirm that if I set OMP_NUM_TRHEADS=1 the test passes.

Best,
Gabriele

On Thu, Nov 2, 2023 at 2:35 PM Cupp, Samuel D. <scupp1 at my.apsu.edu> wrote:

> Hi Gabriele,
>    I've been experimenting with this, and I've reproduced this issue on a
> different machine. In my case, it runs with 1 OMP thread but has the
> segfault with 2. Can you run that test parfile
> (repos/GRHayLET/GRHayLHDX/test/Balsara0.par) without OMP? If this is the
> source, it will help narrow down the problem. I can run the test with
> multiple threads on my local machine though, so I'm not sure why OMP only
> causes a segfault sometimes.
>
>    Samuel Cupp
>    Postdoctoral Researcher
>    Department of Physics
>    University of Idaho
> ------------------------------
> *From:* Gabriele Bozzola <bozzola.gabriele at gmail.com>
> *Sent:* Sunday, October 29, 2023 6:16 PM
> *To:* Cupp, Samuel D. <scupp1 at my.apsu.edu>
> *Cc:* Steven R. Brandt <sbrandt at cct.lsu.edu>; users at einsteintoolkit.org <
> users at einsteintoolkit.org>
> *Subject:* Re: [Users] meeting minutes for 2023-10-19
>
> Hi Sam,
>
> I verified that the testsuite passes if I remove the GRHayL thorns.
> Other than that, ET seems to work fine.
>
> Attached is the cfg file I used to compile ET on Anvil.
>
> Best,
> Gabriele
>
> On Thu, Oct 26, 2023 at 12:45 PM Cupp, Samuel D. <scupp1 at my.apsu.edu>
> wrote:
>
> Hi Gabriele,
>    We discussed this in the call this morning, and there's a few things we
> can try. First, it would be helpful if you created a ticket so we can track
> progress on the issue. This is especially true since the testsuite
> shouldn't hang if a test fails. The expected behavior would be for it to
> continue with testing, but for some reason it didn't. Also, it might help
> to remove GRHayLHDX from the thornlist, recompile, and see if that is the
> only test failing. Knowing if this is specific to the thorn or a broader
> issue would help diagnose the source of the problem.
>
>    Samuel Cupp
>    Postdoctoral Researcher
>    Department of Physics
>    University of Idaho
> ------------------------------
> *From:* Gabriele Bozzola <bozzola.gabriele at gmail.com>
> *Sent:* Tuesday, October 24, 2023 1:27 PM
> *To:* Cupp, Samuel D. <scupp1 at my.apsu.edu>
> *Cc:* Steven R. Brandt <sbrandt at cct.lsu.edu>; users at einsteintoolkit.org <
> users at einsteintoolkit.org>
> *Subject:* Re: [Users] meeting minutes for 2023-10-19
>
> Hi Sam,
>
> This is just CPU. The .out for the entire testsuite was attached to my
> first email.
>
> Best,
> Gabriele
>
>
>
> On Tue, Oct 24, 2023 at 12:14 PM Cupp, Samuel D. <scupp1 at my.apsu.edu>
> wrote:
>
> Do you know if any other CarpetX tests fail? Also, is this using gpus or
> cpus?
>
>    Samuel Cupp
>    Postdoctoral Researcher
>    Department of Physics
>    University of Idaho
> ------------------------------
> *From:* Gabriele Bozzola <bozzola.gabriele at gmail.com>
> *Sent:* Tuesday, October 24, 2023 10:07 AM
> *To:* Cupp, Samuel D. <scupp1 at my.apsu.edu>
> *Cc:* Steven R. Brandt <sbrandt at cct.lsu.edu>; users at einsteintoolkit.org <
> users at einsteintoolkit.org>
> *Subject:* Re: [Users] meeting minutes for 2023-10-19
>
> Hi Sam,
>
> I pulled the latest version. Tests are not failing anymore, but looking at
> the .out, it seems that the Balsara0 test fails and stalls execution of the
> entire testsuite.
>
> I am attaching the .out for the test.
>
> Best,
> Gabriele
>
> On Fri, Oct 20, 2023 at 2:32 PM Cupp, Samuel D. <scupp1 at my.apsu.edu>
> wrote:
>
> Hi Gabriele,
>    It looks like some of the failures are in GRHayLHD. I've been having
> trouble getting consistent output with nproc>1. The data doesn't change, it
> just gets rearranged in the datafile. I made changes to the tests this
> week, so could you rerun the GRHayLHD tests and tell me if they fail? If
> they do, I'll change the tests so that they only run for nproc=1.
>
> I don't know why it would take that long to run, however. Nothing in the
> ET CI suggests it should take that long, best I can tell. Do you have an
> idea of which test is taking so long?
>
>    Samuel Cupp
>    Postdoctoral Researcher
>    Department of Physics
>    University of Idaho
> ------------------------------
> *From:* Users <users-bounces at einsteintoolkit.org> on behalf of Gabriele
> Bozzola <bozzola.gabriele at gmail.com>
> *Sent:* Thursday, October 19, 2023 5:46 PM
> *To:* Steven R. Brandt <sbrandt at cct.lsu.edu>
> *Cc:* users at einsteintoolkit.org <users at einsteintoolkit.org>
> *Subject:* Re: [Users] meeting minutes for 2023-10-19
>
> Hello,
>
> Is there a tested configuration for anvil?
>
> I compiled it one last weekend and ran the tests with the master branch. I
> found a few failures in the
> MPI runs, and the test suite does not complete even with a walltime of 6
> hours.
>
> I am attaching the output.
>
> Best,
> Gabriele
>
> On Thu, Oct 19, 2023 at 7:36 AM Steven R. Brandt <sbrandt at cct.lsu.edu>
> wrote:
>
> Present: Steve, Peter, Sam, Zach, Leo
>
> Release
>      Gallery Examples
>      - TOV is Peter
>      - BBH is Steve
>      - Roland might have students for the other three?
>      Tests
>      - Need to be run
>      - Clusters:
>          Stampede2 isn't listed, and Delta and Anvil are added.
>      - Release name: Lise Meitner
> https://en.wikipedia.org/wiki/Lise_Meitner
>      - Autogenerated codes need to be regenerated, Leo will do it
>
> Mailing list moderation
>      - Remind Roland to send Peter the password
>
> Unanswered question on the mailing list
>      - Need answer for Enzo. Roland was going to answer?
>      - Ticket 2749: Leo says he fixed it.
>      - Ticket 2647: One of Zach's students is looking at that in Grail (sp)
>      - Ticket 2609: Steve thinks he approved PR for Roland
>
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.einsteintoolkit.org/pipermail/users/attachments/20231105/a2355df2/attachment.htm>