[Ncep.list.nems.announce] impending NEMS commit
Samuel Trahan - NOAA Affiliate
samuel.trahan at noaa.gov
Mon May 14 15:34:24 UTC 2018
The changes are committed. Please let me know as soon as possible if you
run into problems, unless they appear to be Rocoto-related. Any errors
coming from Rocoto itself should be reported to the Rocoto helpdesk
rdhpcs.rocoto.help at noaa.gov. Anything else should be sent to me.
On Mon, May 14, 2018 at 10:26 AM, Samuel Trahan - NOAA Affiliate <
samuel.trahan at noaa.gov> wrote:
> Hi all,
> As is expected on a day like today, someone was wandering aimlessly on the
> DC Metro red line tracks between stations, and the trains couldn't move for
> about a half hour while the authorities coaxed him out. I'm starting the
> commit now, and I expect to finish around 11-11:30 Eastern.
> Note to self: in the future, schedule NEMS commits in the afternoon.
> Sam Trahan
> On Mon, May 14, 2018 at 7:58 AM, Samuel Trahan - NOAA Affiliate <
> samuel.trahan at noaa.gov> wrote:
>> Hi all,
>> The previously mentioned NEMS commit will go in as scheduled, though it
>> may be closer to 10:30 Eastern if the DC Metro is in a bad mood. There are
>> two further changes to NEMSfv3gfs after the last batch of FV3 commits:
>> 1. On sJet and vJet, the C768 no longer fits in 16 cores per node. I
>> have reduced it to 14 cores per node.
>> 2. On WCOSS Phase 2, some of the tests now take slightly over a half hour
>> (sometimes). I reverted the queue to dev2 and wallclock limit to 45
>> minutes. If you feel like baby-sitting the workflow, you can switch back
>> to debug2 and 30 minutes, but you will have to keep running "rtrewind" on
>> the jobs that hit their wallclock limit. Ultimately, you might finish the
>> test faster that way, but it will be quite tedious.
>> Also, please note that on GAEA and Jet, you must specify which partition
>> or set of partitions to use through the "--platform" option. If you don't
>> do this, NEMSCompsetRun will tell you the available platforms, and how to
>> request one. I'm using GAEA C3 on GAEA, and I test in s/vJet and xJet on
>> Jet. Presently, t/uJet do not work with NEMSfv3gfs because that app needs
>> to be recompiled with new libraries.
>> Sam Trahan
>> On Thu, May 10, 2018 at 5:12 PM, Samuel Trahan - NOAA Affiliate <
>> samuel.trahan at noaa.gov> wrote:
>>> Hi all,
>>> I will be committing some long-awaited NEMS changes Monday at 10 AM if I
>>> don't hear back from any naysayers. The overview of the changes is below.
>>> All tests have passed except a final retest of NEMSfv3gfs on GAEA, which
>>> should finish soon. (It passed twice, so I don't expect the third to
>>> fail.) Please let me know if you plan on committing answer-changing
>>> results to a Supported app before then.
>>> 1. GAEA support is added to NEMSCompsetRun and NEMSAppBuilder. The
>>> NEMSfv3gfs compsets all work on GAEA C3 and C4.
>>> 2. Jet IMPI support in NEMSfv3gfs, plus support for C768 on Jet. Both
>>> xJet and the s+v Jet pair are supported and have separate baselines.
>>> 3. Pass $CFLAGS down to NEMS C code so we can add uJet and tJet
>>> support. It will be a while before the underlying libraries are
>>> recompiled; until then we cannot support anything on t or u Jet.
>>> 4. Add FV3-MOM6-CICE5 to the list of Supported apps.
>>> Relevant locations:
>>> NEMS branch: gaea-support
>>> App branches: gaea-jet-support
>>> URL with test results:
>>> I may not be able to fill in the "GAEA results" box in that website
>>> because GAEA is unable to contact dmzgw right now. However, the logs will
>>> be in the NEMSfv3gfs gaea-jet-support branch once the tests complete.
>>> Sam Trahan
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Ncep.list.nems.announce