ci: disable power9, add a100 and roci jobs #348
Conversation
rbberger
force-pushed
the
rberger_ci_upgrade
branch
5 times, most recently
from
e9d0569 to
54b615a
Compare
rbberger
force-pushed
the
rberger_ci_upgrade
branch
2 times, most recently
from
db64436 to
b38205c
Compare
rbberger
changed the title
rbberger
changed the title
rbberger
force-pushed
the
rberger_ci_upgrade
branch
from
cf971c8 to
8070c42
Compare
rbberger
force-pushed
the
rberger_ci_upgrade
branch
2 times, most recently
from
e84508a to
e2999aa
Compare
rbberger
force-pushed
the
rberger_ci_upgrade
branch
from
1abb487 to
c068623
Compare
|
@Yurlungur this is now ready. The CI on one of the Rocinante jobs still fails due to some HDF5 issue, but I think this is beyond the scope of this MR which already packs a lot of changes. |
There was a problem hiding this comment.
I'm approving now. However, I'd like to give @ktsai7 @dholladay00 and @mauneyc-LANL to review if they want to. Please review by early next week or forever hold your peace, then I will merge.
This allows offline jobs to continue.
rbberger
force-pushed
the
rberger_ci_upgrade
branch
from
c068623 to
f75632b
Compare
|
@ktsai7 in particular should look at the different jobs that are still rebuilding various spack packages. |
|
@rbberger I will take a look at the packages. @Yurlungur If this is working, I would say to get it in. I am making rounds adding some CI stuff, but that won't conflict with this. |
| @@ -59,6 +59,8 @@ register_headers( | |||
|
|
|||
| if (SINGULARITY_BUILD_CLOSURE) | |||
| register_headers(closure/mixed_cell_models.hpp) | |||
| register_headers(eos/singularity_eos.hpp) | |||
| register_srcs(eos/singularity_eos.cpp) | |||
There was a problem hiding this comment.
what necessitated this change?
There was a problem hiding this comment.
I didn't change this, @rbberger did, but I think he just moved these registration calls from the fortran if statement to the closure if statement. Since this is C++ code technically, not fortran code.
There was a problem hiding this comment.
this will blow up compile times in RIOT. While it is c++ source, they are really only needed in w/in the context of defining fortran interfaces and for EAP.
There was a problem hiding this comment.
oh this is a good catch. Thank you! Yes let's move it back into the fortran. I can submit a PR. @rbberger will this be ok?
PR Summary
Closes #347
PR Checklist
make formatcommand after configuring withcmake.If preparing for a new release, in addition please check the following: