As a heads up - openmpi 5.0.0 drops support for 32-bit builds [1]. I'm not sure how far away it is from release - we're on rc10 at the moment.

Affected packages seem to be:

amg4psblas-1.1.0-4.fc38.src.rpm arbor-0.7-4.fc38.src.rpm auryn-0.8.2-15.fc38.src.rpm boost-1.78.0-11.fc38.src.rpm bout++-4.4.2-5.fc37.src.rpm cgnslib-4.3.0-6.fc38.src.rpm cgnslib-4.3.0-7.fc38.src.rpm coin-or-Ipopt-3.14.9-2.fc38.src.rpm combblas-1.6.2-0.15.beta2.fc37.src.rpm cp2k-2023.1-2.fc38.src.rpm dolfin-2019.1.0.post0-36.fc38.src.rpm elpa-2022.05.001-2.fc38.src.rpm fftw-3.3.10-3.fc37.src.rpm freefem++-4.12-2.fc38.src.rpm ga-5.7.2-13.fc38.src.rpm gmsh-4.11.1-3.fc38.src.rpm gpaw-22.8.0-5.fc38.src.rpm gretl-2022c-2.fc38.src.rpm h5py-3.7.0-4.fc38.src.rpm hdf5-1.12.1-11.fc38.src.rpm hpx-1.8.1-1.fc37.src.rpm hypre-2.24.0-3.fc37.src.rpm intel-mpi-benchmarks-2021.3-3.fc38.src.rpm lammps-20220623-3.fc38.src.rpm libcircle-0.3-12.fc38.src.rpm libneurosim-1.2.0-6.20210110.gitafc003f.fc38.src.rpm mathgl-8.0.1-2.fc38.src.rpm mpi4py-3.1.4-2.fc38.src.rpm MUMPS-5.5.1-1.fc38.src.rpm MUSIC-1.1.16-10.20201002git8c6b77a.fc38.src.rpm nest-3.3-1.fc38.src.rpm netcdf-4.9.0-5.fc38.src.rpm netcdf-cxx4-4.3.1-8.fc38.src.rpm netcdf-fortran-4.5.4-3.fc38.src.rpm netgen-mesher-6.2.2202-5.fc38.src.rpm neuron-8.0.2-6.fc38.src.rpm nwchem-7.0.2-12.fc38.src.rpm openmpi-4.1.4-8.fc38.src.rpm paraview-5.11.0-3.fc38.src.rpm petsc-3.17.4-7.fc38.src.rpm psblas3-3.8.0-3.fc37.src.rpm python-dijitso-2019.1.0-12.fc38.src.rpm python-ffc-2019.1.0.post0-11.fc38.src.rpm python-lfpy-2.2.6-6.fc38.src.rpm python-steps-3.6.0-27.fc38.src.rpm quantum-espresso-7.0-5.fc38.src.rpm scalapack-2.2.0-3.fc38.src.rpm scotch-6.1.2-3.fc37.src.rpm sundials2-2.7.0-12.fc38.src.rpm sundials-5.8.0-11.fc38.src.rpm superlu_dist-8.1.1-1.fc38.src.rpm vtk-9.2.5-2.fc38.src.rpm

Personally I've been planning on dropping 32-bit support in a buch of packages I maintain in this stack like vtk, paraview, etc. This seems like as good a driver for it as any.

That said, packages that build serial and parallel versions don't need to drop 32-bit support completely, just for the openmpi builds.

1 - https://github.com/open-mpi/ompi/pull/11282