Build error with AMD compiler in FatesLitterMod.F90

[ndkeen]

After updating AMD compiler version #6003 (which was necessary as previous version removed), for several tests, ie SMS_Ld30.f45_f45.IELMFATES.pm-cpu_amdclang.elm-fates_satphen, I see build error with AMD compiler:

F90-S-0155-Derived type has not been declared - integer (/global/cfs/cdirs/e3sm/ndk/repos/nexty-oct26/components/elm/src/external_models/fates/biogeochem/FatesLitterMod.F90: 450)
  0 inform,   0 warnings,   1 severes, 0 fatal for adjust_sf_cwd_frac
Target CMakeFiles/lnd.dir/__/__/elm/src/external_models/fates/biogeochem/FatesLitterMod.F90.o built in 0.107528 seconds
make[2]: *** [cmake/lnd/CMakeFiles/lnd.dir/build.make:3452: cmake/lnd/CMakeFiles/lnd.dir/__/__/elm/src/external_models/fates/biogeochem/FatesLitterMod.F90.o] Error 1

I can work-around by simply using integer type:
In components/elm/src/external_models/fates/biogeochem/FatesLitterMod.F90

     !ndk type(integer), intent(in)          :: ncwd !number of cwd pools                                                                                                                                                                                                                                   
     integer, intent(in)          :: ncwd !number of cwd pools                                                                                                                                                                                                                                              

[rgknox]

Thanks for posting this @ndkeen , we'll work with you to get this fix in

[glemieux]

Per our discussion yesterday, the current plan is to address this within #5555, by rolling it into fates PR NGEET/fates#1109.

[rgknox]

Added a new PR with the changes, they did not make it into FATES: 1109, see 1121 linked above

[glemieux]

@ndkeen I tested NGEET/fates#1121 and I'm not hitting this any more, but I'm getting the following error during build:

F90-F-0004-Unable to open MODULE file m_string.mod (/global/u1/g/glemieux/E3SM-project/e3sm-test/externals/mct/mct/m_MCTWorld.F90: 25) 
F90/x86-64 Linux Flang - 1.5 2017-05-01: compilation aborted
gmake[1]: *** [Makefile:62: m_MCTWorld.o] Error 1
gmake: *** [/global/u1/g/glemieux/E3SM-project/e3sm-test/externals/mct/Makefile:10: subdirs] Error 2
ERROR: F90-F-0004-Unable to open MODULE file m_string.mod (/global/u1/g/glemieux/E3SM-project/e3sm-test/externals/mct/mct/m_MCTWorld.F90: 25)
F90/x86-64 Linux Flang - 1.5 2017-05-01: compilation aborted
gmake[1]: *** [Makefile:62: m_MCTWorld.o] Error 1
gmake: *** [/global/u1/g/glemieux/E3SM-project/e3sm-test/externals/mct/Makefile:10: subdirs] Error 2

Are you hitting this as well?

[ndkeen]

When I checkout branch:

perlmutter-login32% git branch
* lnd/fates-api-readparams-refactor

I still see same build error:

/pscratch/sd/n/ndk/e3sm_scratch/pm-cpu/SMS_Ld30.f45_f45.IELMFATES.pm-cpu_amdclang.elm-fates_satphen.20231117_154451_s4ck84/bld/e3sm.bldlog.231117-154714

Ah maybe that's what you are saying -- need to get in 1121 -- I'm not sure how best to do that?

[glemieux]

@ndkeen I realized I mislead you; I haven't been able to get to the e3sm build step as it's failing in the mct build step. I just attempted this with the current master commit 93ab10c and got a different MCT build failure:

F90-F-0004-Unable to open MODULE file m_sortingtools.mod (/global/u1/g/glemieux/E3SM-project/e3sm/externals/mct/mct/m_AttrVect.F90: 3341)
F90/x86-64 Linux Flang - 1.5 2017-05-01: compilation aborted
gmake[1]: *** [Makefile:62: m_AttrVect.o] Error 1
gmake: *** [/global/u1/g/glemieux/E3SM-project/e3sm/externals/mct/Makefile:10: subdirs] Error 2
ERROR: F90-F-0004-Unable to open MODULE file m_sortingtools.mod (/global/u1/g/glemieux/E3SM-project/e3sm/externals/mct/mct/m_AttrVect.F90: 3341)
F90/x86-64 Linux Flang - 1.5 2017-05-01: compilation aborted
gmake[1]: *** [Makefile:62: m_AttrVect.o] Error 1
gmake: *** [/global/u1/g/glemieux/E3SM-project/e3sm/externals/mct/Makefile:10: subdirs] Error 2

Should I make this a new issue?

[glemieux]

Ah maybe that's what you are saying -- need to get in 1121 -- I'm not sure how best to do that?

The way I do this is by adding the PR remote to the fates submodule (in this case [email protected]:rgknox/fates.git) and then checking out the branch directly. This is a bit complicated though as that PR branch is based on an e3sm-fates API update that's coming in with #6018.