Skip to content

Benchmark LinkedCell performance

Sam Reeve edited this page Oct 16, 2023 · 2 revisions

The plot below shows performance of the LinkedCellList on the ORNL Frontier supercomputer. Both CPU and GPU performance are compared as a function of total particle count (with a single MPI rank).

Create refers to creation of the permutation vector for spatially binning the particles (construction of the LinkedCellList). Sort refers to spatially binning the particles. Each line represents a binning a single property or all particle properties.

Frontier

Implementation

Default parameters with the commandline "large" setting were used for these results.

Clone this wiki locally