Ok, I patched it to use -a 0.03 for the time being. That makes the test
suite pass for us.
On 8/25/22 15:45, Josh Berryman wrote:
> Well.... 0.064 kcal/mol is still not much, but it is looking like I
> might have to take the trouble myself to verify that this really is just
> some kind of rounding error.
>
> Please just ignore the test variance for now and I will get around to
> confirming if it is rounding, and then see if I can stabilise things
> numerically in any way, avoid dividing small numbers by small numbers etc.
>
> Josh
>
> On Thu, 25 Aug 2022 at 14:06, Åke Sandgren via AMBER-Developers
> <amber-developers.ambermd.org <mailto:amber-developers.ambermd.org>> wrote:
>
> And after "fixing" those two tests (-a 0.01)
> I get this instead
>
> ===
> b-an02 [~/Projects/easybuild-dev/easybuild-framework]$ less
> /dev/shm/eb-ake/Amber/22.0/foss-2021b-AmberTools-22.3/test/kmmd/kmmd_pmemd/mdout.MPI.dif
>
>
> 918c918
> < EELEC = 90.4140 EHBOND = 0. RESTRAINT =
> 0.4002
> > EELEC = 90.4141 EHBOND = 0. RESTRAINT =
> 0.4259
> 919c919
> < EAMBER (non-restraint) = 124.5939
> > EAMBER (non-restraint) = 124.5681
> ### Maximum absolute error in matching lines = 2.58e-02 at line 919
> field 4
> ### Maximum relative error in matching lines = 6.42e-02 at line 918
> field 9
> ===
>
> How about that one?
>
> On 8/25/22 12:46, Åke Sandgren via AMBER-Developers wrote:
> > And if it should be +-0.01 kcal/mol the test needs to be
> something else
> > than -t 3 since that would allow up to +-0.09 kcal/mol
> >
> > On 8/25/22 12:43, Åke Sandgren via AMBER-Developers wrote:
> >>
> >>
> >> On 8/25/22 11:28, Josh Berryman wrote:
> >>> I'm happy to allow 0.01 of a kcal/mol deviation. Will you make
> the
> >>> commit Ake?
> >>
> >> The test needs to do -t 3 apparently, I misinterpreted the -t flag
> >> initially.
> >>
> >> I don't have commit access to the repo, only read (I hope)
> >>
> >>> Josh
> >>>
> >>> On Thu, 25 Aug 2022 at 10:51, Åke Sandgren via AMBER-Developers
> >>> <amber-developers.ambermd.org
> <mailto:amber-developers.ambermd.org>
> <mailto:amber-developers.ambermd.org
> <mailto:amber-developers.ambermd.org>>>
> >>> wrote:
> >>>
> >>> Hi!
> >>>
> >>> When building here with GCC 11.2.0 I get a small diff in
> kmmd_pmemd
> >>> test
> >>> results. Should the test be changed from dacdif -t 2 to
> dacdif -t
> >>> 1 or
> >>> is that to aggressive?
> >>>
> >>> ===
> >>> b-an02 [/scratch/ake/amber22/amber22_src]$ cat
> >>>
> /dev/shm/eb-ake/Amber/22.0/foss-2021b-AmberTools-22.3/test/kmmd/kmmd_pmemd/mdout.dif
>
> >>>
> >>> 919c919
> >>> < EAMBER (non-restraint) = 124.42
> >>> ---
> >>> > EAMBER (non-restraint) = 124.41
> >>> ===
> >>>
> >>> ===
> >>> b-an02 [/scratch/ake/amber22/amber22_src]$ cat
> >>>
> /dev/shm/eb-ake/Amber/22.0/foss-2021b-AmberTools-22.3/test/kmmd/kmmd_pmemd_gb/mdout.dif
>
> >>>
> >>> 803c803
> >>> < EELEC = 42.26 EGB = 40.14 RESTRAINT =
> >>> 1.20
> >>> ---
> >>> > EELEC = 42.26 EGB = 40.14
> RESTRAINT =
> >>> 1.19
> >>> ===
> >>>
> >>> -- Ake Sandgren, HPC2N, Umea University, S-90187 Umea,
> Sweden
> >>> Internet: ake.hpc2n.umu.se <mailto:ake.hpc2n.umu.se>
> <mailto:ake.hpc2n.umu.se <mailto:ake.hpc2n.umu.se>> Mobile: +46 70
> >>> 7716134 Fax: +46 90-580 14
> >>> WWW: http://www.hpc2n.umu.se <http://www.hpc2n.umu.se>
> <http://www.hpc2n.umu.se <http://www.hpc2n.umu.se>>
> >>>
> >>> _______________________________________________
> >>> AMBER-Developers mailing list
> >>> AMBER-Developers.ambermd.org
> <mailto:AMBER-Developers.ambermd.org>
> <mailto:AMBER-Developers.ambermd.org
> <mailto:AMBER-Developers.ambermd.org>>
> >>> http://lists.ambermd.org/mailman/listinfo/amber-developers
> <http://lists.ambermd.org/mailman/listinfo/amber-developers>
> >>> <http://lists.ambermd.org/mailman/listinfo/amber-developers
> <http://lists.ambermd.org/mailman/listinfo/amber-developers>>
> >>>
> >>
> >
>
> --
> Ake Sandgren, HPC2N, Umea University, S-90187 Umea, Sweden
> Internet: ake.hpc2n.umu.se <mailto:ake.hpc2n.umu.se> Mobile: +46 70
> 7716134 Fax: +46 90-580 14
> WWW: http://www.hpc2n.umu.se <http://www.hpc2n.umu.se>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org <mailto:AMBER-Developers.ambermd.org>
> http://lists.ambermd.org/mailman/listinfo/amber-developers
> <http://lists.ambermd.org/mailman/listinfo/amber-developers>
>
--
Ake Sandgren, HPC2N, Umea University, S-90187 Umea, Sweden
Internet: ake.hpc2n.umu.se Mobile: +46 70 7716134 Fax: +46 90-580 14
WWW: http://www.hpc2n.umu.se
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Aug 25 2022 - 07:30:04 PDT
