search

ESMCI / ccs_config_cesm

CESM CIME Case Control System configuration files
3 stars 46 forks source link

Update for mksurfdata_esmf build #37

Closed ekluzek closed 2 years ago

ekluzek commented 2 years ago

Changes needed for the mksurfdata_esmf build

Fixes #36

I've only tested in the context of mksurfdata_esmf so far:

I also did the following tests on izumi and they PASS:

ERS_Ld3.f19_g17.X.izumi_intel
ERS_Ld3.f45_g37_rx1.A.izumi_intel

ekluzek commented 2 years ago

Bringing this in would cause the #38 bug to come back as well as the slowdown that caused #40 to revert the previous change. So this really can't come in as is right now.

mvertens commented 2 years ago

I would confirm a performance slowdown before backing this out. My current understanding is that the biggest culprit is CAM (not sure why).

ekluzek commented 2 years ago

I'd like to look at this again, so that we can update the mksurfdata_esmf to use the latest ESMF version that's in ccs_config_cesm0.0.39.

ekluzek commented 2 years ago

@fischer-ncar OK I backed out the ESMF changes, so this seems really safe to me now and not in need of further testing. But, let me know if you disagree. The only thing in there now is setting the specific version of cmake on cheyenne (to the current default), and adding the setting of the env variable PIO on izumi.

With the ESMF changes I thought it needed more testing, but now I don't think so. So let me know if you consider this good to go as it is.

ekluzek commented 2 years ago

OK I tried the following tests in the context of a cesm2_3_alpha09d tag and they pass.

ERS_Ld3.f19_g17.X.izumi_intel
ERS_Ld3.f45_g37_rx1.A.izumi_intel