TPL Update 2025-12-15 #1509
TPL Update 2025-12-15 #1509
Conversation
chapman39
force-pushed
the
feature/chapman39/tpl-2025-12-15
branch
from
a44e1d3 to
09bb31f
Compare
chapman39
force-pushed
the
feature/chapman39/tpl-2025-12-15
branch
from
09bb31f to
18cf423
Compare
… docker containers
|
Heads up for you @chapman39 , Axom changed their default type for index: You will get a lot of warnings as errors. |
…ith into feature/chapman39/tpl-2025-12-15
…ith into feature/chapman39/tpl-2025-12-15
| version: [7.2.6] | ||
| externals: | ||
| - spec: py-sphinx@7.2.6 | ||
| - spec: py-sphinx@7.2.6 ^python |
| prefix: /usr | ||
| openblas: | ||
| buildable: false | ||
| # Spack is insistent on building openblas in the CUDA container, so allow building from source |
There was a problem hiding this comment.
Should this be removed then? Why have an external when we are building it?
| prefix: /usr/tce | ||
| - spec: cmake@3.30.5 | ||
| modules: | ||
| - cmake/3.30.5 |
There was a problem hiding this comment.
Is this better than putting a hard-coded path like everything else?
| """Axom provides a robust, flexible software infrastructure for the development | ||
| of multi-physics applications and computational tools.""" | ||
|
|
||
| maintainers("white238") | ||
|
|
||
| homepage = "https://github.com/LLNL/axom" | ||
| git = "https://github.com/LLNL/axom.git" |
There was a problem hiding this comment.
Nitpick: The description, homepage, and git can be removed.
|
|
||
| class Fmt(BuiltinFmt): | ||
| # Fix NVCC error | ||
| patch("system_error_cuda.patch", when="@12.1.0:") |
There was a problem hiding this comment.
Is this the same issue Alan and Axom was running into? Btw Axom just updated their fmt to 12.1.0 as well.
| with when("+sundials"): | ||
| # Going to sundials@7: causes 80%+ test failures | ||
| depends_on("sundials@:6.999") | ||
| depends_on("sundials") |
There was a problem hiding this comment.
| depends_on("sundials") |
w/o the version this line isn't needed anymore.
| depends_on("caliper+mpi~papi", when="+caliper") | ||
| depends_on("caliper+adiak", when="+caliper+adiak") |
There was a problem hiding this comment.
| depends_on("caliper+mpi~papi", when="+caliper") | |
| depends_on("caliper+adiak", when="+caliper+adiak") | |
| with when("+caliper") | |
| depends_on("caliper+mpi~papi") | |
| depends_on("caliper+adiak", when="+adiak") |
I think this is cleaner...
|
|
||
| set_source_files_properties(functional_tet_quality PROPERTIES LANGUAGE CXX) | ||
|
|
||
| # Then add the examples/tests |
There was a problem hiding this comment.
This is great you verified that overiding these languages were no lnoger needed.. thanks!
| { | ||
| const auto currentConductivity = | ||
| conductivity_offset_ + get<0>(parameter) + d_conductivity_d_temperature_ * temperature; | ||
| #if defined(SMITH_USE_CUDA) || defined(SMITH_USE_HIP) |
There was a problem hiding this comment.
We might want to create a SMITH_USE_GPU that encompasses both of these. We should review RAJA's patterns
white238
left a comment
white238
left a comment
There was a problem hiding this comment.
All my comments were nitpicking and I don't want to hold up this being merged after someone else's review. Thanks for the hard slog @chapman39 !
this pr
componentsvariant to Axom's spack package axom#1740about CUDA
related PRs
next