Lisandro Dalcin

> What if your oversubscription ratio? We are improving that part of functionality I have 2 (virtual) CPUs (GitHub actions limit for free runners), and I would like to run...

Sorry, looks like I got confused and made a minor mistake. I've updated the list with a fix. The `MPI_{Reduce|Ireduce}_scatter` with `MPI_IN_PLACE` issue is fixed in both Linux and Windows...

OK, great, may thanks for the tips. I'm running 2021.11.0 on Fedora Linux 38 and looks that things got better. The first issue in the list related to matched probes...

@ddurnov Could you please backport this trivial frix from MPICH https://github.com/pmodels/mpich/pull/6933 ?

@ddurnov The latest Intel MPI release 2021.12 introduced a somewhat serious ABI regression. The (weak) symbols `MPI_Status_(c2f|f2c)` are missing from `libmpi.so`. However, the corresponding `PMPI_Status_(c2f|f2c)` are available. You can easily...

@ddurnov I'm seeing regressions with latest version 2021.12, but only on Windows, all related to user-defined struct datatypes with `long`/`unsigned long`/`long double` members in most cases, and `int32_t`/`uint32_t` in one...

Thanks for the update, @rafael-lago. I'm working on reverting my previous workarounds: https://github.com/mpi4py/mpi4py/commit/ff8139edc7b48ca89791e11397d0de7a9b7657df https://github.com/mpi4py/setup-mpi/commit/520633b286fcd0f3dbc25ef6d84d78f6d6bb1ab6 https://github.com/mpi4py/mpi4py-publish/commit/8228cba6389c3febbc2f1f64b583e8d3920b1dbb

@rafael-lago @ddurnov In PR #518 I'm adding support for the new MPI-4.x routines you guys have been incrementally adding into Intel MPI since release 2021.9.0. All the important stuff (e.g....

@Vizonex You should test your work against mpi4py ``` git clone https://github.com/mpi4py/mpi4py.git cd mpi4py python conf/cythonize.py ``` Your code currently crashes Cython: ``` Error compiling Cython file: ------------------------------------------------------------ ... bint...

> FYI, I'm pretty sure (but haven't tested it) that this works correctly in scikit-build-core. Oh, sorry, I forgot to update this issue. I've switched to `scikit-build-core`, and the issue...