Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

TODO carefully investigate QUDA mixed solver parameters #545

Open
kostrzewa opened this issue Jun 22, 2022 · 1 comment
Open

TODO carefully investigate QUDA mixed solver parameters #545

kostrzewa opened this issue Jun 22, 2022 · 1 comment

Comments

@kostrzewa
Copy link
Member

Investigate the interaction of

  • rel_delta / use_alternative_reliable
  • max_res_increase
  • max_res_increase_total

On the one hand, this is relevant for multi-shift CG with mixed-precision refinement for the 1+1 sector, on the other hand, it is also relevant for intermediate and large tm_rho shifts in the determinant and determinant ratios.

@kostrzewa
Copy link
Member Author

Initial investigation

compare_mixed_cg.pdf

suggests that rel_delta = 1e-1 should be set for sloppy == half and for safety, the defaults for max_res_increase and max_res_increase_total should be relaxed.

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

No branches or pull requests

1 participant