Test recoverML is failing

Create issue
Issue #539 closed
Ian Hinder created an issue

Since 26.08.2011, the recoverML test has failed with the stable version of Carpet. The error is:

   phi.x.asc: substantial differences
      significant differences on 1 (out of 42) lines
      maximum absolute difference in column 13 is 1.922567332939e-08
      maximum relative difference in column 13 is 3.71924165894473e-07
      (insignificant differences on 12 lines)
   phi.y.asc: substantial differences
      significant differences on 2 (out of 66) lines
      maximum absolute difference in column 13 is 1.92256733397983e-08
      maximum relative difference in column 13 is 3.71924166087253e-07
      (insignificant differences on 20 lines)

Possible candidate commits for causing this failure are:


User: eschnett
Date: 2011/08/25 12:40 PM

 vectors-4-SSE.h, vectors-8-SSE2.h

Suggest asm statements to support SSE4a with Intel compilers.
Indent vector architecture definitions.


commit 2888d221261da25e7802b76c46ee3e10ca578418
Author: Barry Wardell <barry.wardell@gmail.com>
Date:   Thu Aug 18 00:05:43 2011 +0200

    Enable Vectorisation.

Keyword: McLachlan

Comments (4)

  1. Barry Wardell
    • removed comment

    It can't have been the change to Vectors as that shouldn't have changed the behavior of the code at all.

    I think it is quite likely that it is the commit after the mentioned McLachlan one:

    commit 16c7cc3c66cab59041cfb2203953eaa21cd7812d Author: Barry Wardell <barry.wardell@gmail.com> Date: Thu Aug 18 00:12:46 2011 +0200

    Regenerate code with Vectorisation enabled.

    Note that this also applies the changes from commit 3ba8a55ae2578cb6dc06f0ec8b81f86b3a2654ac to ML_BSSN_MP, ML_BSSN_MP_Helper, ML_BSSN_MP_O8, ML_BSSN_MP_O8_Helper, ML_BSSN_O2, ML_BSSN_O2_Helper, ML_BSSN_O8, ML_BSSN_O8_Helper, ML_BSSN_UPW and ML_BSSN_UPW_Helper, which were accidentally missed in that commit.

    In particular, this applied the recent schedule updates (which were already applied to other McLachlan thorns) to ML_BSSN_O2. So I think the root cause of the change was:

    commit 3ba8a55ae2578cb6dc06f0ec8b81f86b3a2654ac Author: Erik Schnetter <schnetter@cct.lsu.edu> Date: Tue Aug 2 20:37:19 2011 -0400

    Correct schedule, in particular for checkpoint/recovery

    Do not mark ADMBase variables for non-checkpointing if they have multiple timelevels. (Variables with multiple timelevels must always be checkpointed, because the past timelevels cannot be regenerated after recovery.)

    Finally remove all perl post-processing of the auto-generated code; instead, use proper Kranc mechanisms.

    Schedule the ADM constraints and ADM quantities after MoL_PostStep, since this is where the ADMBase variables are set.

    Schedule enforcing the BSSN constraints in the new schedule group MoL_PostStepModify, since they should not be enforced after recovery. (This would lead to inconsistencies at floating-point round-off level.)

    Regenerate all thorns.

  2. Ian Hinder reporter
    • removed comment

    This test uses the old !CartGrid3D symmetries. Could this be related? These are deprecated now because they don't work with the Mercurial version of Carpet, and will be replaced as per #481.

  3. Log in to comment