Skip to content

McGavin's C backtracker: the throughput story, built here

Peter McGavin's own C backtracker, the community's fastest: a 2007 optimization recipe compounded for two decades through generated code, lookup tables and counter tricks, then built on my M1 and pointed at the real 256-piece puzzle, where single core it drives past 200 of 256 pieces at ~109M placements/s.

by Peter McGavin · write-up by Raphaël Anjou

experimentmeasurementmeasuredGo fasterBacktrackingUpdated 2026-07-21
Reproduceheavy compute — script and results are committedreruns the searchBudget: 30 s per run
Hardware & run
Standardized benchCPU only
0.0083core-hours
Cores
1
RAM
16 GiB
GPUs
0
CPU
Apple M1
Machine
MacBook (Apple M1, 8 cores), single-threaded binary
Budget
30 s per run
Start
time-seeded; deterministic on Joe's puzzle, non-terminating on real E2
Reproduce this result

heavy compute — script and results are committed

fetch genbody71.zip from groups.io msg 11749; gcc -Ofast -DG then without -DG

Keep exploring

Referenced by

Page sourceView as Markdown