bench: add optional CP-SAT reference #22

2026-07-30T17:40:45+01:00

mcp commented

2026-07-30 17:40:45 +01:00

Summary

add an optional OR-Tools CP-SAT first-solution reference model
model NoOverlap2D, exact fill, edge exclusions, and equal-copy ordering
validate reports through a dependency-free implementation of the native validation contract
report model size, workers, elapsed time, memory, placements, and solver statistics
document order-8/order-9 comparisons and keep CP-SAT as a reference only
record why DLX is not currently justified

OR-Tools remains outside the native build and default dependency set.

Review

Review strengthened validator parity, subprocess error and model-stability handling, generated-artifact ignores, and the explicit order-9 single-observation limitation. Final review found no outstanding issues.

Testing

Release CTest with system Python and no OR-Tools: 9/9 passed
optional Python 3.13 / OR-Tools 9.15.6755 reference tests: passed
Python compilation checks: passed
independent review order-8 CP-SAT solve: valid
git diff --check

Results

CP-SAT order 8, 8 workers, 3 valid runs: 2.080 s median (2.019–3.046 s), 220–232 MB peak RSS
native order 8, 1 worker: 1.854 s median
CP-SAT order 9, 8 workers, one valid run: 54.623 s, 364 MB peak RSS
native order 9: 1.846 s predecessor-search median plus 0.250 us construction

Closes #5

## Summary - add an optional OR-Tools CP-SAT first-solution reference model - model `NoOverlap2D`, exact fill, edge exclusions, and equal-copy ordering - validate reports through a dependency-free implementation of the native validation contract - report model size, workers, elapsed time, memory, placements, and solver statistics - document order-8/order-9 comparisons and keep CP-SAT as a reference only - record why DLX is not currently justified OR-Tools remains outside the native build and default dependency set. ## Review Review strengthened validator parity, subprocess error and model-stability handling, generated-artifact ignores, and the explicit order-9 single-observation limitation. Final review found no outstanding issues. ## Testing - Release CTest with system Python and no OR-Tools: 9/9 passed - optional Python 3.13 / OR-Tools 9.15.6755 reference tests: passed - Python compilation checks: passed - independent review order-8 CP-SAT solve: valid - `git diff --check` ## Results - CP-SAT order 8, 8 workers, 3 valid runs: 2.080 s median (2.019–3.046 s), 220–232 MB peak RSS - native order 8, 1 worker: 1.854 s median - CP-SAT order 9, 8 workers, one valid run: 54.623 s, 364 MB peak RSS - native order 9: 1.846 s predecessor-search median plus 0.250 us construction Closes #5

mcp added 1 commit 2026-07-30 17:40:46 +01:00

bench: add optional CP-SAT reference 0a7ce1e49e

Provide a separately installed OR-Tools model to compare a generic constraint solver with the native first-solution path without adding a production or default-test dependency.

Use no-overlap, exact-fill, edge and equal-copy symmetry constraints, validate placements independently, and record model size, memory, worker count and timings for orders 8 and 9. Keep the tool only as a reference and defer DLX absent new evidence.

Tests: Release CTest (9 passed)

Tests: Python reference tests and compilation checks

Tests: independently validated CP-SAT orders 8 and 9

Refs: #5

mcp merged commit 0a7ce1e49e into main

2026-07-30 17:41:02 +01:00

mcp deleted branch codex/issue-5-cp-sat-reference

2026-07-30 17:41:02 +01:00

mcp referenced this pull request

2026-07-30 17:41:26 +01:00

roadmap: improve time to the first square solution #16

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: mgrettondann/partridge-cpp#22