Tune JumperGraphSolver heuristics and iteration budget to reach 60% benchmark success #3

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.

Sign up for GitHub

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

Jump to bottom

Draft

seveibar wants to merge 1 commit into main from codex/improve-algorithm-performance-for-benchmark-krv7e4

Contributor

seveibar commented Jan 6, 2026

Motivation

Improve success rate of the Single 1206x4 Jumper Grid Solver benchmark to at least 60% by adjusting solver heuristics and budgets.
Reduce unnecessary ripping penalties and better balance exploration vs heuristic guidance for high-crossing cases.
Provide a reproducible benchmark output capture for verification.

Description

Adjust solver heuristic and costs by setting greedyMultiplier to 0.95, ripCost to 40, and reducing port rip penalty to 1.
Lower per-crossing penalty to computeDifferentNetCrossings(...) * 6 and increase MAX_ITERATIONS to 20000 + input.inputConnections.length * 1500.
Add OUTPUT.txt containing the benchmark run output and commit the updated lib/JumperGraphSolver/JumperGraphSolver.ts.

Testing

Ran bun run scripts/run-benchmark-single-1206x4.ts and observed an average success rate of 60.0% (output saved to OUTPUT.txt).
Ran bunx tsc --noEmit for type checking which completed successfully.
Attempted bun run format which failed because the format script is not present.


          Improve jumper solver benchmark success

88e94aa

seveibar added the codex label

— with

ChatGPT Codex Connector

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

Labels