[Users] Segmentation Fault with CarpetReduce
Garrison, David
Garrison at uhcl.edu
Fri Apr 29 15:10:15 CDT 2022
I managed to track down where the error is coming from and it does seem to be either cluster or compiler dependent. It’s not something in the Cactus Code in general.
-DG
--
David Garrison, Ph.D.
Associate Dean for the College of Science and Engineering,
Professor and Founding Chair of Physics
University of Houston-Clear Lake
Bayou 3611/STEM 2252
Houston, TX 77058
BAYOU: 281-283-3717
STEM : 281-283-3796
https://www.uhcl.edu/science-engineering/faculty/garrison-david
http://www.uhcl.edu/physics
"If we knew what it was we were doing, it would not be called research, would it?" — Albert Einstein.
On Apr 29, 2022, at 3:37 AM, Bruno Giacomazzo <bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>> wrote:
DG,
it could also be that there is a bug in the code, but other compilers are not affected by it.
My strategy is usually to make sure that I can reproduce the problem on the cluster when running just with one core. In this way it is easier to run within a debugger and spot the problem.
Best regards,
Bruno
Il giorno ven 29 apr 2022 alle ore 08:12 Garrison, David <Garrison at uhcl.edu<mailto:Garrison at uhcl.edu>> ha scritto:
It seems to be an error on a cluster I use but not on my Mac. It may be an issue with how the cluster is configured.
-DG
On Apr 28, 2022, at 9:48 AM, Bruno Giacomazzo <bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>> wrote:
DG,
thanks for the parameter file.
During today's call we also saw that you mentioned that you encountered a segmentation fault when using one ghost point. Do you have a backtrace? Did you try running the code in a debugger to check which variable is causing it?
Cheers,
Bruno
Il giorno gio 21 apr 2022 alle ore 17:32 Garrison, David <Garrison at uhcl.edu<mailto:Garrison at uhcl.edu>> ha scritto:
I don’t use standard Einstein Toolkit thorns. I also don’t use refinement levels in Carpet. Please see attached for an example. The error occurs when I change the ghost zone size to 1.
-DG
--
David Garrison, Ph.D.
Associate Dean for the College of Science and Engineering,
Professor and Founding Chair of Physics
University of Houston-Clear Lake
Bayou 3611/STEM 2252
Houston, TX 77058
BAYOU: 281-283-3717
STEM : 281-283-3796
https://www.uhcl.edu/science-engineering/faculty/garrison-david
http://www.uhcl.edu/physics
"If we knew what it was we were doing, it would not be called research, would it?" — Albert Einstein.
On Apr 21, 2022, at 9:24 AM, Bruno Giacomazzo <bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>> wrote:
DG,
if you could share the parameter file that would be useful in order to try reproducing the error with standard Einstein Toolkit thorns. I do not see any problem with the lines of code you posted in the email.
Do you get the same problem also when using Carpet without refinement levels?
Best regards,
Bruno
Il giorno gio 31 mar 2022 alle ore 19:17 Garrison, David <Garrison at uhcl.edu<mailto:Garrison at uhcl.edu>> ha scritto:
Hello,
It is a custom code so I am attaching the part I used to do the reductions instead.
handle = CCTK_ReductionHandle("average");
ierr = CCTK_Reduce(cctkGH, -1, handle, 1, CCTK_VARIABLE_REAL, &scale_factor, 1, CCTK_VarIndex("MHD_Analysis::aa_out_avg"));
ierr = CCTK_Reduce(cctkGH, -1, handle, 1, CCTK_VARIABLE_REAL, &utime, 1, CCTK_VarIndex("MHD_Analysis::univ_age"));
-DG
On Mar 31, 2022, at 9:21 AM, Bruno Giacomazzo <bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>> wrote:
DG,
could you share the parameter file you are using? So that we may have a better look at the problem.
Thanks,
Bruno
Il giorno mer 30 mar 2022 alle ore 20:52 Garrison, David <Garrison at uhcl.edu<mailto:Garrison at uhcl.edu>> ha scritto:
Hello,
Whenever I set the number of ghost zones to 1, I get segmentation faults from my reduction operations. This wasn’t an issue with Pugh or when I use 2 or more ghost zones with Carpet. Has anyone else had this problem?
-DG
--
David Garrison, Ph.D.
Interim Associate Dean for the College of Science and Engineering,
Professor and Founding Chair of Physics
University of Houston-Clear Lake
Bayou 3611/STEM 2252
Houston, TX 77058
BAYOU: 281-283-3717
STEM : 281-283-3796
https://www.uhcl.edu/science-engineering/faculty/garrison-david
http://www.uhcl.edu/physics
"If we knew what it was we were doing, it would not be called research, would it?" — Albert Einstein.
_______________________________________________
Users mailing list
Users at einsteintoolkit.org<mailto:Users at einsteintoolkit.org>
http://lists.einsteintoolkit.org/mailman/listinfo/users<https://urldefense.com/v3/__http://lists.einsteintoolkit.org/mailman/listinfo/users__;!!BCR0FSePrR4x!GUGB7CSrDExPdA7mPVZuCzM9_713wMhA6f8kQaO9C64EN4b4r13kCh02xdXjodRmQMcnuEwDpO-G513ivCR5jfsvfMFCBA$>
--
Prof. Bruno Giacomazzo
Department of Physics
University of Milano-Bicocca
Piazza della Scienza 3
20126 Milano
Italy
email: bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>
phone: (+39) 02 6448 2321
web: http://www.brunogiacomazzo.org<https://urldefense.com/v3/__http://www.brunogiacomazzo.org/__;!!BCR0FSePrR4x!GUGB7CSrDExPdA7mPVZuCzM9_713wMhA6f8kQaO9C64EN4b4r13kCh02xdXjodRmQMcnuEwDpO-G513ivCR5jfvd-LgExw$>
----------------------------------------------------------------------
There are only 10 types of people in the world:
Those who understand binary, and those who don't
----------------------------------------------------------------------
--
David Garrison, Ph.D.
Interim Associate Dean for the College of Science and Engineering, Professor and Former Chair of Physics
University of Houston-Clear Lake
Bayou 3611
Houston, TX 77058
Tel: 281-283-3796
Fax: 281-283-3709
http://sce.uhcl.edu/garrison
http://www.uhcl.edu/physics
"If we knew what it was we were doing, it would not be called research,
would it?" ‹ Albert Einstein.
--
Prof. Bruno Giacomazzo
Department of Physics
University of Milano-Bicocca
Piazza della Scienza 3
20126 Milano
Italy
email: bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>
phone: (+39) 02 6448 2321
web: http://www.brunogiacomazzo.org<https://urldefense.com/v3/__http://www.brunogiacomazzo.org/__;!!BCR0FSePrR4x!DqdugUWohGBKF9_Wl__sHgXYw1vCqt6Rw0l_AuKu5GS6Xh14wzBgWopnoJL3cax3rxn42oIsJW6QCOoYgGoKOrxkYCB7dw$>
----------------------------------------------------------------------
There are only 10 types of people in the world:
Those who understand binary, and those who don't
----------------------------------------------------------------------
--
Prof. Bruno Giacomazzo
Department of Physics
University of Milano-Bicocca
Piazza della Scienza 3
20126 Milano
Italy
email: bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>
phone: (+39) 02 6448 2321
web: http://www.brunogiacomazzo.org<https://urldefense.com/v3/__http://www.brunogiacomazzo.org/__;!!BCR0FSePrR4x!De0kFnjWOdFfWZLrtJ1dyPjh8U0NRsUHBxUJCDlFCu258B4dtytsKJzPrQGeWiO6NI92DOTRQcQWsHCGTm7bD8H2p8VpCw$>
----------------------------------------------------------------------
There are only 10 types of people in the world:
Those who understand binary, and those who don't
----------------------------------------------------------------------
--
Prof. Bruno Giacomazzo
Department of Physics
University of Milano-Bicocca
Piazza della Scienza 3
20126 Milano
Italy
email: bruno.giacomazzo at unimib.it<mailto:bruno.giacomazzo at unimib.it>
phone: (+39) 02 6448 2321
web: http://www.brunogiacomazzo.org<https://urldefense.com/v3/__http://www.brunogiacomazzo.org/__;!!BCR0FSePrR4x!G8QCN-PYIs8St4OyCehFnkWB7djOBoqOoBU9IWkO43zQWrLwzYsqG4QDzRQV-s2fLlvqSh-u75EnfsjTxQv1Qd-w6r_YiQ$>
----------------------------------------------------------------------
There are only 10 types of people in the world:
Those who understand binary, and those who don't
----------------------------------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20220429/a38948a5/attachment-0001.html
More information about the Users
mailing list