I'm officially satisfied with the mods I made to the local coordinate
system to fix the NV* energy drift and I will try to submit by that date
(NTP would only reduce noise since the the nature drift potentially changes
every time the unit cell size changes). At my end, it's a ~2% perf hit
overall probably due to SFU congestion adding 3 additional type conversions
to every coordinate load. On the bright side, it opens the door to further
optimizations in the NL builder and it simplifies orthorhombic cell
simulations.
On Fri, Feb 25, 2022 at 4:59 AM David A Case <david.case.rutgers.edu> wrote:
> Please remember: the code deadline for the upcoming Amber and AmberTools
> releases is March 11, 2022.
>
> Right now, all merge requests at gitlab.ambermd.org are blocked at the CI
> stage, due to a recent change with how mpi4py in the miniconda distribution
> is performing. But please submit your merge requests anyway.
>
> [Jason: should we temporarily skip parallel mm_pbsa.py tests so that we can
> unclog the pipeline?]
>
> Please send me (or the list) an email message if you have significant
> changes that will be coming over the next two weeks.
>
> ....thanks....dac
>
> p.s. I'm still looking for volunteers to be testing release candidates on
> OSX machines, especially on OSX 12 (Monterey) with either Intel or M1
> chips.
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Feb 25 2022 - 11:30:03 PST
