Skip to content

jeffhammond/KahanMPI

Repository files navigation

KahanMPI

The goal of this project is to address the challenges of reproducibility and accuracy faced by MPI applications when using floating-point reductions.

In particular, we aspire to implement numerically if not bit-wise reproducible MPI reductions using a variety of well-known algorithms for such purposes, including Kahan (compensated) summation.

Known Issues

The CMake build system is broken. Do not use it. And by, I mean CMake. Ever.

Disclaimer

Neither Professor William Kahan nor anyone else named Kahan has endorsed this software. The name reflects the fact that Kahan summation is used extensively.

About

Implementation of Kahan summation for MPI reductions

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published