/pbbsbench

New version of pbbs benchmarks

Primary LanguageC++MIT LicenseMIT

pbbsbench

The Problem Based Benchmark Suite (PBBS) is a collection of over 20 benchmarks defined in terms of their IO characteristics. They are designed to make it possible to compare different algorithms, or implementations in different programming languages. The emphasis is on parallel algorithms. Our default implementations are for shared-memory multicore machines, although we would be delighted if others implemented versions on GPUs or distributed machines and made comparisons.

A list of the benchmarks can be found on here.

Information on the organization and on how to run PBBS can be found on here.

And here is a paper outlining the benchmarks:

The problem-based benchmark suite (PBBS), V2
Daniel Anderson, Guy E. Blelloch, Laxman Dhulipala, Magdalen Dobson, and Yihan Sun
ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming (PPoPP), 2022