Mike Wilkins

Base Allreduce Algorithm Selection/Performance Issue

3

## Background information I was investigating an Allreduce performance issue for 64k processes @ 32kB message size. ### What version of Open MPI are you using? (e.g., v4.1.6, v5.0.1, git...

coll: Added recursive multiplying algorithm

2

## Pull Request Description Adding the recursive multiplying allreduce algorithm. This algorithm achieves better performance than existing algorithms. They were designed with DOE's new exascale machines (Frontier, Aurora, El Capitan)...

Compiling on OpenSUSE 15.4, I got the following compile error: ``` In file included from ../src/mpid/ch4/shm/src/../posix/../ipc/src/ipc_p2p.h:14:0, from ../src/mpid/ch4/shm/src/../posix/posix_coll_gpu_ipc.h:61, from ../src/mpid/ch4/shm/src/../posix/posix_coll.h:12, from ../src/mpid/ch4/shm/src/../posix/shm_inline.h:16, from ../src/mpid/ch4/shm/src/shm_coll.h:10, from ../src/mpid/ch4/shm/src/shm_impl.h:18, from ../src/mpid/ch4/include/mpidch4.h:450, from ../src/mpid/ch4/include/mpidpost.h:10,...

Usleep Progress Throttle w/ No Progress Counter

5

## Pull Request Description This PR is an enhanced version of the progress changes from https://github.com/pmodels/mpich/pull/7368. https://github.com/pmodels/mpich/pull/7368 introduces `MPIR_CVAR_CH4_PROGRESS_THROTTLE`, which, when enabled, adds `usleep(1)` to the progress loop, preventing cache...

This algorithm achieves better performance than existing bcast algorithms for both small and large message sizes. The algorithm is based on the circulant graph abstraction and Jesper Larsson Traff's recent...

Replaced default algorithm tuning files with aurora-specific files

1

## Pull Request Description ## Author Checklist * [ ] **Provide Description** Particularly focus on _why_, not _what_. Reference background, issues, test failures, xfail entries, etc. * [ ] **Commits...