Skip to content

Conversation

@widlarizer
Copy link

When clang is used with -flto, nbody and cubic would get absolutely optimized away. This was caused by pure (side-effect free) functions being run only once, even though they're called in a loop N times with the same parameters. For this reason, their parameters must have been changed to create N distinct sets of parameters. This PR adds small changes and reduces the cubic and nbody LTO scores by 2-4 factors of magnitude while bringing minor changes to non-LTO scores.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant