Paul Bartholomew
Paul Bartholomew
This is the most recent version I have access to ``` AMD clang version 13.0.0 (CLANG: AOCC_3.2.0-Build#128 2021_11_12) (based on LLVM Mirror.Version.13.0.0) ``` I'm testing building it now.
I also get the same issue using the above AOCC version ``` F90-S-0038-Symbol, decomp_profiler_none, has not been explicitly declared (/work/e01/e01/ptbe01/src/2decomp-fft/src/profiler_caliper.f90) 0 inform, 0 warnings, 1 severes, 0 fatal for decomp_profiler_fin_noarg...
Ah I'd overlooked #6. I think a combination of the two, currently ``decomp_2d.f90`` is >2000 lines long (discarding #include'd files) making it very difficult to work with/change. So as an...
I think this would also tie in with @rfj82982 desire to have decomp2d without MPI eventually
Also, to avoid a "mega commit" I think it will be better to periodically merge this branch into main - the first case might help clarify the plan
You are right, I can reproduce this - I've always used the `ccmake` interface to configure this so never noticed. If you look at the github actions, they follow this...
however I should emphasise this only affects how the tests are run using `ctest` - you can manually run them using `mpirun -np ...` and codes using the library will...
Thanks for the report - I think we'd need to check there's not a reason why 3.21 was set as the minimum (possibly a build requirement for one of the...
> "I see that the BP4 code bent out of its shape to accommodate this situation" - I meant in the adios2 source code, not your code. Not to worry,...
This PR is replaced by #409