Skip to content
This repository was archived by the owner on Aug 28, 2024. It is now read-only.

Commit

Permalink
fixed typo
Browse files Browse the repository at this point in the history
  • Loading branch information
cwishard committed Mar 4, 2023
1 parent d850a41 commit 429ea3a
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion paper/paper.md
Original file line number Diff line number Diff line change
Expand Up @@ -49,7 +49,7 @@ The combination of modern programming practices, flexible data processing tools,

# Performance

Modeling the behavior of a thousands of fully interacting bodies over long timescales is computationally expensive, with typical runs taking weeks or months to complete. The addition of collisional fragmentation can quickly generate hundreds or thousands of new bodies in a short time period, creating further computational challenges for traditional \textit{n}-body integrators. As a result, enhancing computational performance was a key aspect of the development of 'Swiftest'. Here we show a comparison between the performance of 'Swift', 'Swifter-OMP' (a parallel version of 'Swifter'), and 'Swiftest' on simulations with 1k, 2k, 8k, and 16k fully interacting bodies. The number of CPUs dedicated to each run is varied from 1 to 24 to test the parallel performance of each program.
Modeling the behavior of thousands of fully interacting bodies over long timescales is computationally expensive, with typical runs taking weeks or months to complete. The addition of collisional fragmentation can quickly generate hundreds or thousands of new bodies in a short time period, creating further computational challenges for traditional \textit{n}-body integrators. As a result, enhancing computational performance was a key aspect of the development of 'Swiftest'. Here we show a comparison between the performance of 'Swift', 'Swifter-OMP' (a parallel version of 'Swifter'), and 'Swiftest' on simulations with 1k, 2k, 8k, and 16k fully interacting bodies. The number of CPUs dedicated to each run is varied from 1 to 24 to test the parallel performance of each program.

\autoref{fig:performance} shows the results of this performance test. We can see that 'Swiftest' outperforms 'Swifter-OMP' and 'Swift' in each simulation set, even when run in serial. When run in parallel, 'Swiftest' shows a significant performance boost when the number of bodies is increased. The improved performance of 'Swiftest' compared to 'Swifter-OMP' and 'Swift' is a critical step forward in \textit{n}-body modeling, providing a powerful tool for modeling the dynamical evolution of planetary systems.

Expand Down

0 comments on commit 429ea3a

Please sign in to comment.