search

E3SM-Project / e3sm-unified

A metapackage for a unified anaconda environment for analyzing results from the Energy Exascale Earth System Model (E3SM).
BSD 3-Clause "New" or "Revised" License
8 stars 8 forks source link

Update to 1.3.1.2 #78

Closed xylar closed 3 years ago

xylar commented 4 years ago

Change mpi4py to be included explicitly in nompi build (it was there anyway because of ilamb) and make sure mpi is mpich except for the explicit openmpi build, since this is what works well with e3sm_diags.

xylar commented 4 years ago

@chengzhuzhang, @forsyth2 and @golaz, I think this is ready start testing as soon as e3sm_diags gets built on conda-forge: https://github.com/conda-forge/staged-recipes/pull/12570

I will put test environments on Cori, Compy, Anvil and Cooley. You will be able to activate them with a script, load_test_e3sm_unified.sh, in the same place as load_latest_e3sm_unified.sh.

@chengzhuzhang and @forsyth2, based on your experience on Cori compute nodes, I am making the MPICH environment the default. It will be important for us to do some testing on both login and compute nodes on all the systems I listed above. If it turns out that MPICH should be the default on some systems and no-MPI on others, that's an option, too.

@golaz, the new release of MPAS-Analysis should have fixed https://github.com/MPAS-Dev/MPAS-Analysis/issues/729 but I re-opened that issue so we can close it later when you verify that you're not seeing the problem on Compy (or elsewhere) anymore.

chengzhuzhang commented 4 years ago

Thank you @xylar ! @forsyth2 once the test enviroment is ready, would you please help testing based on our testing plan? I'm taking time off this week, hopefully i will still have time to do more tests, once I return to work on Monday next week.

xylar commented 4 years ago

@chengzhuzhang, enjoy your time off! I have pushed back the release date to next Wednesday so that should give you a couple days of testing. But I can push it to later in the week if need be.